2ND BASIC COURSE ON ECONOMIC BASIS OF HEALTH CARE INTERVENTIONS. 19™ TO 2 1st AUGUST 1999.

Item

Title
2ND BASIC COURSE ON ECONOMIC BASIS OF
HEALTH CARE INTERVENTIONS.
19™ TO 2 1st AUGUST 1999.
extracted text
JU

1

2nd BASIC COURSE ON ECONOMIC BASIS OF
HEALTH CARE INTERVENTIONS.
19™ TO 2 1st AUGUST 1999.

COURSE CO-ORDINATOR: DR. K.R. JOHN, MD
VENUE: CHTC

Overall objective of the course
1) Define health economics and state its uses
2) a) state type of costs direct and indirect costs , fixed and variable
costs, marginal and average costs b) Enumerate steps in costing
3) Distinguish cost minimisation - cost effectiveness, cost benefit and
cost utility studies
4) How to critically review an article and derive conclusions

5) How to ask research questions relating to economic evaluation
start collecting informations.

d'1 •

1 f
11 1^'i

2nd BASIC COURSE ON ECONOMIC BASIS OF HEALTH CARE
INTERVENTIONS ( 19th to 21st of September)

Programme

Day 1
19.8.99
8.00 a.m.

8.45
9.00 a.m.
10.00 a.m.
10.30 a.m.

1.00 p.m.
2.00 p.m.
3.00 p.m.
5.00 p.m.

Introduction / pre evaluation
Expectations
Definition, scope, uses
Types of studies in published literature
Inauguration
Principles of costing, Types of costs,
depreciation, annualization
Coffee break
Exercise on costing Group I - OP visit
Group II - IP bed
Group III - Dressing
Group IV - Haemoglobin test
Developing the spread sheet
Lunch
Components of economic analysis / types of studies
How to review articles relating to economic evaluation one
worked out example of each type
Assignments (take home) for journal review

Day 2
20.8.99
8.00 a.m.
10.00 a.m.
10.30 a.m.

1.00 p.m.
2.00 p.m.

4.00 - 5.00 p.m.

Review of articles - Group I, II, in, IV
Coffee break
Group Costing exercises CAT Scan
OP department
- Pharmacy department
Lab costing
IP department
TB program
Lunch
Developing spread sheet and entering in the computer
using excel and Epi info
How to conduct studies relating to economic evaluation
Review of articles take home

Day 3
21.8.99
8.00 a.m.
10.00 a.m
11. 00 a.m.
12.00

Presentation of review of articles/
How to develop a proposal group work
Study proposals presentations
Post evaluation
Concluding session

Follow up
Submission of assignments

1

Critical review of one article
Prnnaqal writino

1.<>v
f

CLINICAL ECONOMICS

A

MODULE 1

AN INTRODUCTION TO CLINICAL ECONOMICS

At the completion of this module you should understand:
a.

the foundations on which economics is based;

b.

why economics is relevant to health;

c.

the role of clinical economics.

1

INTRODUCTION TO CLINICAL ECONOMICS
1970s,
health expenditure as a proportion of Gross
In the
Product
(an indicator of the total expenditure on goods
Domestic
services
in
a country)
rose rapidly in most developed
and
countries (Table 1).
This raised interesting questions about
TABLE 1

HEALTH EXPENDITURE AS A PERCENTAGE OF GDP
(current prices)

COUNTRY

1970

YEAR
1975

1980

1983

New Zealand
United Kingdom
Australia
Sweden
USA
OECD average

4.5
4.5
5.7
7.2
7.6
5.6

5.2
5.5
7.6
8.0
8.6
6.7

5.7
5.8
7.4
9.5
9.5
7.2

5.7
6.2
7.5
9.6
10.8
7.6

Source:
Harvey,
R.
’’Trends
in Health Service Provision and
Expenditure
in
Australia,
and their relevance to public
hospitals”, unpublished, Australian Institute of Health, 1987.
level of health expenditure - for example, does
the appropriate
allocate
too
few and the US too many resources to
New Zealand
In
trying
to
answer
this, related questions about value
health?
in
the
health
sector
were raised.
Does increased
for money
better
levels
of
health,
are the right
lead to
expenditure
on
preventive
vs curative
expenditure
choices made between
aged vs care for infants, and
the
for
between care
services,
low technology interventions, for
technology vs
between high
These concerns are increasingly being expressed in the
example?
probably even more pertinent to
and are
literature,
medical
the shortage of resources available
countries where
developing
i s more apparent than in developed
to
the health sector
countries.

not surprising that economists have played a role in this
It is
The
discipline
of economics is based on two-'
debate.
about
the
world,
firstly that society (consisting
observations
individuals
and
institutions)
has virtually unlimited demands
of
services,
and
secondly that the resources
goods
and
f or
the
goods
and
services society values are
available to produce
not
scarce
in
the sense of non-renewable
They are
scarce.
there
will never be enough to
because
scarce
resources,
but
would
like.
Choices between
people
that
produce everyth ing

2
uses
of these scarce
resources are,
therefore,
competing
Accordingly,
the choice to use resources in a
inevi table.
particular fashion always involves a cost in that the resources
used in other ways which would have produced
cannot
be
The value of the greatest possible benefit _wh£ch
benefits.
could have been obtained by using the resources elsewhere is the
cost of foregoing the opportunity to use them elsewhere and is
known as the ^jpportunity cqs±.
economics can be defined as
the
Given these observations,
concerned
with
the problem of using or
"social
science
administering scarce resources ... so as to attain the greatest
f ulf iIment of society’s unlimited wants” (Jackson &
or maximum
1987,
pl9).
This general problem is usually
McConnell
subdivided
into a number of more tractable questions, including
what goods
and services
to produce, in what quantities and at
what time,
what resources will be used in their production, and
how to distribute the goods and services that are produced,
The
guiding principle is
the search
for ef f iciency.
Productive
ef f iciency
involves producing the greatest possible output from
a given quantity of resources, or stated another way,producing
a given output aT the lowest possible cost.
However, there is
no point being efficient
in this sense if goods which society
does not value very highly are being produced.
Allocative
which
involves producing
the goods and services
ef f iciency,
which
society values most highly, is important as well;
This
the fact that economics is
concern with efficiency highlights
not simply an exercise in reducing costs.
Economics is as much
about
increasing outputs, the goods and services society values,
as about reducing inputs or costs.

The

observations about resource scarcity which form the basis of
economics apply to the health sector as well,
well.
Peoples’ desires
and
services,
including health,
are virtually
for
goods
and governments must decide how many
unlimited and individuals
of their scarce resources should be allocated to health care or
to other ways of gaining benefits.
Health economics is simp1y
»» discipline
of economics applied
the
to the topic of health”
It is a very broad topic, encompassing all
(Mooney
1986,
p4 ) .
the general economic questions raised above, and efficiency is
Clinical economics examines a narrower
important concern.
an
quest i ons
relating to the efficiency of clinical
set
of
is concerned mainly with clinical
This course
interventions.
the wider concerns of health
economics,
although
one of
economics is introduced in Module 10.

A

few general points should be made in conclusion, 1 Firstly, two
Positive economics
is
types
of economic analyses exist.
with
describing
what
happens
concerned
objectively.
normative economics
involves subjective value
Alternatively,
of
the
outcomes
which are considered to be
judgements
It is sometimes argued that clinical economics is an
desirable.
example of normative economics because the choice of what

3
constitutes
a benefit of a medical
subjective (Drummond et al 1987).

intervention is necessarily

of
is not about
The problems
about. money.
economics
Secondly,
even
under
scarcity
confront
all
societies,
resource al location
barter
rather than monetary
have been based on
those which
Money
is simply a useful tool for valuing goods and
exchange.
services.

solutions can sometimes
be
inequitable,
efficient
Thirdly,
has
Little
to contribute to the debate about whether
Economics
should be made to reduce such inequities, but it does
attempts
show that reductions in inequality can at times be achieved only
at the expense of efficiency.

Finally,
decisions
the appropriate form of treatment are
dec i s i ons about
not made on
the grounds of efficiency consideration alone, and
no economist would suggest that they should be.
However,
efficiency should not be ignored.
Governments may legitimately
to
follow a less efficient policy for the sake of equity
dec i de
social
justice,
but it is important that they be aware of the
or
costs
of
doing
this.
Similarly, physicians cannot
economic
that economics is irrelevant to their clinical decision
pretend
Every one of their decisions to use resources involves
making.
an opportunity cost in that it prevents the resources being used
somewhere else which would have produced benefits.

REFERENCES
Major Reference.
Drummond M.F. et al,
"Health economics: an introduction for
1 .
clinicians", Annals of Internal Medicine , 107(1), 1987,
88-92.
Other References.
Eisenberg J.M. et
e t al,
"Clinical economics training in the
2.
International Clinical Epidemiology Network", Jnl Clinical
Epidemiology,

3.

"Medical education and economics", Medical
Evans, D.B.
Education, 23, 1989, 48-54.

Economics (2nd.
4.^Jackson, J. & C.R. McConnell,
ed.), McGraw-Hill, Sydney, 1987, chapter 2.

Australian

5. 1 Mooney, G.H.
Economics, Medicine and Health Care, Harvester
Press, Brighton, 1986, chapter 2 .

4

DISCUSSION QUESTIONS
i.

• ’


Read’ the
article
by Drummond et al and discuss the following
proposition.
It
is ethical
for a national Cancer Council to
consider costs and
benefits when discussing
the appropriate
national policy for screening for cancer of the cervix, but it
is not ethical
for a doctor to consider costs when treating an
eth ica1
individual patient.

r

2.
Discuss:
opportunity costs are not involved when patients
are
fully covered by health insurance as they do not pay for the
services they receive.
3.
To what
practice?

extent

are

economic issues relevant to clinical

c

a

i/

I

Reprinted from ANNALS OF INTERNAL MEDICINE Vol. 107; No. 1 July 1987
Printed in U.S.A.

<*

Health Economics: An Introduction for Clinicians
MICHAEL DRUMMOND, Ph.D.; GREG STODDART, Ph D.; ROBERTA LABELLE, M.A.; and ROBERT
CUSHMAN, M.D.; Birmingham, United Kingdom; Hamilton and Ottawa, Ontario, Canada
Economic issues have had a growing importance in the
health care field as the sector's share of the gross
national product has risen. Clinicians are under increasing
pressure to adopt more cost-effective treatment practices
as a result of initiatives being taken by the major thirdparty payers, government, and business. However, recent
publications suggest that there are some misconceptions
about economics in health care and the extent to which it
is in conflict with good clinical practice. To provide a
foundation for the understanding of this field by clinicians,
we have outlined several basic notions of health
economics.^-, a,
^MeSH termsf cost benefit analysis; cost control; decision
maKmg; delivery of health care; diagnostic-related groups;
economics, hospital; economics, medical; health
maintenance organizations; health planning; health policy;
health resources; health services needs and demand;
insurance, health reimbursement; preferred provider
organizations; prospective payment system ]

Economic issues have had a growing importance in the
health care field as the sector’s share of the gross national
product has risen. In the United States, in common with
most developed countries, there has been concern about
health care cost containment and the promotion, by the
major third-party payers, of specific measures to bring
about a more efficient use of resources. For examplejjn
the hospital sector there has been a movement toward
prospective _payment systems, the best known being
-Medicare’s approach, based on diagnosis-related groups
(PROS). Such schemes give hospitals an added incen­
tive to control costs, and the extension of prospective
payment to physicians is now being considered (1).
In the primary care setting the largest change has been the
growth of health maintenance organizations (HMOs) and
preferred provider organizations (PPOs). To a great ex­
tent -this.change is being supported by private company
executives who view it as a way of cutting the costs of
health_care-fbr their-workers. These company officials
have also considered other measures, such as self-insur­
ance and contracting out of specific services (2, 3). It is
in clinicians’ best interests to be aware of these changes
and the economic forces behind them, as they are likely
to have a profound impact on the market for medical
services in the future (4).
The other majoi&eason for clinicians to understand
more about economics stems from their key role as gate­
keepers to the use of health care resources. It has been
► From the Health Services Management Centre, University of Birmingham, Bir­
mingham, United Kingdom; and the Departments of Clinical Epidemiology and
Biostatistics, and Economics. McMaster University; the Ottawa Carleton Region­
al Health Department; and the Department of Community Medicine and Epide­
miology, University of Ottawa, Otuwa. Canada.

88

Annals of Internal Medicine. 1987;107:88-92.

pointed out that physicians represent less than one half of
1% of the population yet determine, through the deci­
sions they make about the care of their patients, how
nearly 10% of the nation's gross national product will be
spent (5). In discharging their responsibilities, clinicians
perform a difficult dual role, of securing good care for
their patients yet having a broader social responsibility
for the careful use of health care resources (6). This
broader role is likely to be reinforced by the changes in
organization and payment and by other initiatives to in­
fluence practice behaviour such as education, feedback on
resource use, peer review schemes, and direct financial
incentives (6-9).
Given this pressing need for clinicians to have a good
grasp of economic issues, it is disturbing to find evidence
that all is not well. Although clinicians are likely to be
more often jpxposed to economic issues and arguments
thrp ugh the , increase in the publication of economic papers in medical journals (10), a recent article (11) point­
ed tCL_the-frequent misuse of the term cost-effectivc in
medicine. The commonest errors are a confusion between
cost containment and economic efficiency, and the sug­
gestion that economic analysis is restricted to considera­
tion of cost savings and production gains, and not con­
cerned with general improvements in the quality of life.
Over t hepastfew years we have been communicating
economics. cQncepts_and__ideas_tQ_practising clinicians
within a teaching hospital setting. This paper, based on
our experience, sets out ten basic notions of health eco­
nomics for those clinicians wishing to have a good grasp
of this field (Table 1).
Ten Basic Notions of Health Economics
HUMAN WANTS ARE UNLIMITED BUT RESOURCES ARE
FINITE

The original intention of some socialisec health care
systems, such as the British National Health Service, was
that moreJnvestment in health care treatipents and pro­
grams ,_wpuld remove existing ’’health needs.” We now
know that health needs will never be eliminated, as these
are continually redefinedrihat is,"tfiere~are always more
legitimate ways of using resources (manpower, equip­
ment, supplies) than there are resources available. Even
if, by some miracle, a stage was reached where the com­
munity did not want more investment in health services,
there would be plenty of other suggestions for using re­
sources outside the health care system in other public or
private sector investments.
Therefore, most resources used in the health care sys­
tem have alternative beneficial uses inside and outside the
© 1987 American College of Physicians

health care sector, although in the short run some re­
sources may be difficult to redeploy. In general, though
the restrictions on the resources for health care are more
a reflection of the human condition than they are the
creation of governments. The implication for those work­
ing within the health care system is that choices, in allo­
cation of scarce resources to competing activities, are inescapable.
ECONOMICS IS AS MUCH ABOUT BENEFITS AS IT IS
ABOUT COSTS

Table 1. Ten Basic Notions of Health Economics

Human wants arc unlimited but resources are finite.
Economics is as much about benefits as it is about costs.
The costs of health care programmes and treatments are not
restncted to the hospital, or even to the health sector.
Choices in health care (that is, in health planning, or in treatment mode) inescapably involve value judgments.
Many of the simple rules of market operation do not apply in
the case of health care.
Consideration of costs is not necessarily unethical.
Most choices in health care relate to changes in the level or
extent of a given activity; the relevant evaluation concerns
these marginal changes, not the total activity.
The Provision of health care is but one way of improving the
health of the population.
®
As a community we prefer to postpone costs and to bring forward benefits.

Given the limitations on resources, choices need to be
made between competing beneficial activities at the plan­
ning and clinical levels. Q^BF^Tof engaging in a.
particular activity/jrthe'Benefit lost by nm Sing ihe same'
resources in their best (that is. most highly valued) alter­ Equity in health care i
‘ ‘ ‘ but reducing inequalimay *„
be desirable,
native use. Thus, although one may tend to think of econ­
ties usually comes at a price.
omists as being interested in health care expenditures, it
would actuallytetnierto.think of them as being interestdeinstitutionalization of patients should not be advocated
that is, in maximising the total benefits
on
cost savings alone; the development of adequate com­
from the use of the community’s scarce resources. This is
munity
care programmes for the elderly or mentally ill is
What economists mean by efficiency. Therefore, when an
not without cost. Moreover, community care may or may
economist poses the question, “What is the cost of X?”
not be more beneficial than its institutional counterpart
he or she is talking about the sacrifice made, or benefits
for
various categories of patient.
foregone, by engaging in X. This is the distinction beIt should also be remembered that the health care secween the economist’s notion of opportunity cost and the
‘orisjwtJ^onljLrKource. for health care. Other public
commoner notion of money expenditures.
and private agencies are involved, and patients (and their
Often_£but not always) money prices may be considfamilies) incur costs. Family costs include the time and
ere^_to reflect true opportunity costs. This situation de­
expense
in traveling to, or waiting at,' healtlTcare'facili­
pends on the extent to which Qpe believes that markets—
ties, the provision of informal “home nursing” for sick
the mam mechanism through which resources, goods,
re atives and any extra medications, equipment, or facili­
and services
are functioning
traded—i perfectly. That
-- -are
ties required for treatment but not funded by the health
is, a l
------- uc
unc wnere
are where
many there are :
perfect
market
wouldmere
be one
service or insurance scheme. Furthermore, if patient or
buyers and sellers,
where; no
-------------no barrier
barrier to
to entry
entry to
to the
the mar­
ket exists, and where consumers have perfect informa­
is a gen
tion.
be lost.
output may
Of course, any one decision maker cannot conceive of
Many clinicians are aware of these nonmonetary costs
all possible states of the world and identify the best alter­
falling on patients and their families, and may moderate
native use for a given resource. However, on a pragmatic
therapies accordingly. However, more attention could be
evel it is possible to use the notion of opportunity cost if
paid
to them in scheduling clinic attendances and in deone is a decision maker with a fixed budget, such as
_y.clppmg admission and discharge j^licies. Such costs
someone running an HMO. For example, the opportunimust also be taken into account in the economic evalua­
more jiigh-technoiogy medicine may be that
tion (from society’s point of view) of health care pro­
prevention programmes are not given funds to expand.
grammes.
H
(In individual clinical decisions, the opportunity cost of
the use of resources may not be so apparent, as discussed
CHOICES IN HEALTH CARE (IN HEALTH PLANNING OR
later.)
the costs of health care programmes and
treatments are not restricted to the

hospital, or even to the health sector v

Because secondary and tertiary care account for a high
proportion of health care expenditures, much of the effort
to increase efficiency has concentrated on these sectors
There have thus been many attempts to shorten hospital
inpatient stays and reduce the use of laboratory services.
Although such moves may be beneficial, reductions in
hospital stay may mean that extra resources are required
m the community care sector to aid in the rehabilitation
of patients. Similarly, in the long-term care sector the

IN A TREATMENT MODE) INESCAPABLY INVOLVE
VALUE JUDGMENTS

So far we have glibly talked about benefits from the use
of resources in health care. But who decides what is a
benefit and what is not? Obviously, the assessment of
benefits (and, through the logic explained previously
costs) can only be based on subjective valuation. There­
fore, choices in health care, which all require assessment
of costs and benefits of alternative programmes or therapies, involve value judgments.
In discussing health care issues, it is important first to
recognize this fact and also to make values explicit when
possible. For example, in prescribing therapy, the clini­
cian may be (unknowingly) making value judgments on
Drummond et al. • Health Economics

89

r

still acting in the wider social interest? We would not
necessarily expect clinicians to take on the broader role in
their day-to-day clinical work, but the answer lies in
i making a distinction between medical decisions made on
behalf of the one patient and those made on behalf of a
I group of patients . .(such as requests for expansion of serv­
ices). In our view it would be entirely consistent for the
r clinician to give each patient as much care asjiis or her.
condhion requires, yet also to participate in a decisiop* making process that, in evaluating competing claims., gor
K the development of services, considers the wider social
^^perspective.
The ideal form of such a decision-making process
would be one that enabled clinicians to retain an advocacy"rble for their own patients, yet brought various checks
and balances into play. In some health care systems this
policy is attempted by encouraging clinicians to take re­
sponsibility for a defined^Judget within the hospital (12)•
In "other systems it is considered more appropriate to en­
courage the adoption of guidelines for clinical practice
that take into account cost-effectiveness considerations
(6, 13). In the United States, the spread of HMOs and
prospective payment systems for hospitals is likely to
have a similar effect/)
f MOST CHOICES IN HEALTH CARE RELATE TO CHANGES
IN THE LEVEL OR EXTENT OF A GIVEN ACTIVITY; THE
RELEVANT EVALUATION CONCERNS THESE
MARGINAL CHANGES, NOT THE TOTAL ACTIVITY

Mt-

In the health care field there is often a mistaken ten­
dency to present choices on an all-or-nothing basis. For
example, the question is not usually whether we do X
(such as develop community programmes for the mental­
ly handicapped), but rather how much of X do we do
(that is, for which type of patient should such pro­
grammes be developed). Therefore, the relevant data for
making such decisions are the margmal costs and benefitsTjiot those of the whole activity. The marginal costs
and benefits strictly relate to one more (or one less) unit
of production, but are often used to refer to the incrementai costs and benefitTof the change in the .scale of the
_—--------------- - '
activity. The notion of “the margin’’ is very important in health
care decision making and there are numerous examples
reported where the marginal costs and benefits of expanding an activity fitter greatly from .the average costPand,
benefits of the activity as it stands.. One study showed
that the average cost (per case detected) of repeatedly
screening the same patient population for cancer of the
colon, up to a maximum of six times, was around $2500 J
(14). Yet the marginal cost of detecting a further case by
doing a sixth Jest, having already done five, was._oygr,
$47 000 000,
Although the sixth stool test for cancer of the colon is
the most well-known example of the importance of con­
sidering marginal cost, there are many other “how
much” decisioqs in diagnosis, treatment, and follow-up.
For example, Williams (15) has calculated that performing coronary artefy bypass grafting for mild angina with
two-vessel diseases is more than ten times the cost .(per

quality-adjusted life-year gained) of this procedure for
severe angina with left main disease. Levine and associates (16) calculated that a policy of comprehensive diag­
nosis for cancer of unknown primary origin (searching
all possible sites) would cost $7 million more per year in
Ontario than a limited diagnostic strategy (examining
only sites for which effective systematic therapy were
available). Sjason and Weinstein (17) have shown how
the cost effectiveness of strategies for the prevention, di­
agnosis, and treatment of hypertension depends on dia­
stolic blood pressure before treatment.
THE PROVISION OF HEALTH CARE IS BUT ONE WAY OF
IMPROVING THE HEALTH OF THE POPULATION

Many clinicians are familiar with the arguments of epi­
demiologists who have pointedout that recgqt technolog­
ic advances in health care have had little impact on life
expectancy, compared with improvements in nutrition,
sanitation, and general economic wealth. Economists
have done similar analyses, which attempt to estimate the
relative contributions of health care and other important
inputs, such as education, to the production of health.
More recently, thoreLhave been_studiesofdie relationship
between health status and general economic variables,
such as the level of unemployment.
Consideration of this notion does not lead to any obvi­
ous suggestions for the modification of clinical practice;
rather, it places all our efforts within the health care sys­
tem into perspective. Also, from the government’s point
of view it suggests that if improvements in health status
are desired, we should look not only to changes in health
policy buLalsp industrial,and ^ucational policy. There
arejoften conflicts here; many countries, have^agricultural
policies that are not conducive to _good Jiealth (such as
"'subsidies to farmers producing foods with high fat con­
tent). Perhaps we have to accept that health is traded, by
persons and governments, for-other benefits; otherwise
why do people drive fast cars, climb mountains, or
smoke?
AS A COMMUNITY WE PREFER TO POSTPONE COSTS
AND TO BRING FORWARD BENEFITS

Different investments in health care have different time
profiles of costs and benefits. A large health education
campaign aimed at reducing coronary risk factors may
require a sizeable resource outlay now, in return for bene­
fits in the future. Other investments, including most ther­
apeutic programmes, involve a steady stream of costs,
with a quick return in terms of improved health status for
the patients treated. It is usually argued that, as individu­
als and as a community, we are not indifferent to the
timing of costs and benefits. In facLwe prefer.to,postpone
costs _and to have benefits sooner rathei-than, latex. Of
course, one cannot have one’s bread buttered on both
sides and, as individuals, if we want to consume more
now we usually have to borrow money at a positive rate
of interest. The rate reflects not only inflation but the
compensation we have to pay to others for postponing
their consumption.
The main implication of this notion, the existence of a
Drummond et al. • Health Economics

91


tv '

$

kv

behalf of the patient. However, it may be possible to set

^^Second] even if consumers knew what they wanted, the

health Fare market is different in that, because of insur­
ance coverage or free provision of socialized health care,
cons^ers_do_nx2t_p^L^price that reflects the social op­
portunity cost of the resources used- Therefore there may
choices in health care iis complex. Economists tend to
be a tendehcyTor persons to consume more care than
believe (as a canon of faith, as a political tenet, or as an
they otherwise might. Economists would say that con­
act of expediency) that each person knows his or her own
sumers experience fgjoral hazard*’ a phenomenon cominterest best, although there is no such “rule” in econom­
insurance markets.
ic theory. Obviously, in some branches of health care,
^j^fhird,) he benefit from a person receiving care may not
such as mental illness, the proposition that the patient
°e confrn£d-tQ3hat_j)erson. The most obvious example is
(consumer) knows his or her own interest best may be
in the field of communicable diseases, whereby the.more
hard to defend. But what about health care more general­
people that become immunized the more protection is
ly? Certainly it is possible to find examples in the litera­
gj ven to othersTBecause of such externality relationships,
ture where health care providers’ values have been asthe valuation of the consumer alone~may understate the
sessed^and they dffigFWonFtKrvalues of the patients.
total value of care. Some economists extend such arguWith respect to larger planning choices, consideration of
Jt!ler ^OIins °f-Care, suggesting that there are
costs (as reflected by markets) incorporates an element
philanthropic (orFaring) externalities) that is, we care
of consumer judgment, because market prices reflect an
about Othertpeoples-healthraFTheKce the care they reamalgam of the valuations many consumers place on
*
~ ~ z ce*ve» *n a way that we do not care about whether they
goods and services and their alternative
Oui. wnoAhave recently been able to replace their automobile. The
should value^the l>effefiyffom the health servicw^dia©
_) main evidence in supporLof this argument is the tenden­
one might want' to give
cy, in most countries, to socialize health care to some
^he^gommunity a say in this, perhaps through their elect­
deffree./j£\
ed representatives. However, at the moment much pnori(Finally? there is a stronger version of the externality '
ty setting is done primarily Tperhaps"unknowingly) by
argument; namely that people ought to be encouraged to,^
clinicians, through their advocacy for ^hFdev^E^THFBF
consume niore_^care_jthan they Otherwise might This is vparticular services or through thei jjressu'resTtheirZactions
known as the |nerit good argument or patemalismjand is
placeon existmg resources, such as hospital laboratories.
most commonly applied to the consumption of education
Much of this may, m turn, be a reflection of patients’
Most of these arguments can be applied, to some de­
demands, however.
gree, to other commodities. However, the unique coinci­
dence of them in the case of the commodity known as
MANY OF THE SIMPLE RULES OF MARKET OPERATION
health care leads most economists to the conclusion that
DO NOT APPLY IN THE CASE OF HEALTH CARE
b |a market fo^ health Gare~would~nfhiif’—th^t is, not auto4 K
We have already mentioned that, if markets are work­
matically lead to an efficient allocation of resources. Fur-1
ing well, the prices of resources (and commodities) re­
^^hermore, m most health care systems there are 1 few (
flect their sqcial opportunity costs. A perfectly function­
k^jncentiygsj-to consumers or providers---for efficnFcy )
ing system of markets would have additional attributes
effi^y^eedsjpJ>ejmc^^^
1
too; for any given distribution of income and wealth, it
would guarantee that goods and services would be pro­
CONSIDERATION OF COSTS IS NOT NECESSARILY
duced in the most efficient way, because in the face _of
UNETHICAL
compg^rion ingffigignL producers would go out of busiIt is normally argued that the clinician’s responsibility
{technical efficiencyfr and that scarce resources
is to provjde_th^best_pQssible_care for his patients. Does
would be allocated so asTo satisfy the most highly valued
this mean that considering costs in clinical decision mak­
^n-ts (BJlocativg.etficieiLcyl^iis view of the world has
ing is unethical? If the^atient were reajly being treated in
led many to advocate a much wider role for the market in
isolation, it would be wrong to withhold care because of
health care delivery, with less government intervention
resource
considerations. However, a problem arises be­
(or interference).
cause (according to the arguments set out) once resourc­
There are doubts whether any markets function in the
es enter into the picture, then by definition the patient is
perfect way described above, owing to the existence of
noj being treated in isolation. According to_the opportu­
Monopoly Producers QI sellers who are relatively immune
nity
cost_priii£iple(64norg resources given tn one patient
to competition. However, in the case of health care there
meansjhat someone else will lose opt. This n^y perhaps
are a number of additional reasons that, taken together,
be a patient of a clinical colleague, or a persoi@hn rnuM
suggest that the market (if left to its own devices) would
potentially benefit from care but is not yet known to the
not lead to an efficient use of health care resources.! Firs|] 6?
health car^ystem- A third
.innsi possihUitFislhat increased
Ccomnypcrs may not have the knowledge? to make sensiblj
resources
used
in
the
health
care sector may mean that
choicesJwhich means that
x
___
that the
the providers
providers (especially
clini- Mbciety goes without other things, such aTFducati^,
cians) become key players in determining the demand for
which themselves may promote health/-------- ---------care, on behalf of patients. jEconomists call this the agen-J
Herein lies a dilemma: How does the clinician adhere
|cy relationship, J'fS?)
to generally accepted medical and ethical principles while
90
July 1987 • Annals of Internal Medicine • Volume 107 • Number 1
regimens and to let the patient make the choice. This is
partly what informed consent is about.

positive rate of time preferrncr;. is that invcstmenffi -tn
health care need to be compared on a common basis if^
their time profiles of costs and benefits_differ. This is ac­
complished by a procedure known as discounting to pres­
ent values, which is essentially a compound interest cal­
culation done in reverse. Discounting has the biggest
impact when one is comparing a preventive programme
with a curative one.

about health economics and that teachers respond by de­
vising more and better learning materials.
ACKNOWLEDGMENTS. The authors thank the members of the Health
Economics and Policy Analysis Group, Department of Clinical Epidemiolo­
gy and Biostatistics, and Dr. Vic Neufeld, Associate Dean (Education) of
Health Sciences, Faculty of Health Sciences, McMaster University.
► Requests for reprints should be addressed to Michael Drummond, Ph.D.;
Director, Health Services Management Centre, University of Birmingham,
Park House. 40 Edgbaston Park Road; Birmingham B15 2RT, United King­
dom.

EQUITY IN HEALTH CARE MAY BE DESIRABLE, BUT
REDUCING INEQUALITIES USUALLY COMES AT A

As noted,
noted. there are externality relationships in
health—that is, many of us care about the health of others. Therefore, in many countries there is concern about
the equity of health care provision, by exposure to risk,
income class, social class, geographical location, need,
and so on. Few people would openly argue against equity,
but it is worth pointing out that reduction of inequalities
may come at a price, in terms of other benefits foregone.
One common saying in economics is that “there is no
such thing as a free lunch.” In this case it is perhaps
easiest to see the validity of the argument in the context
of locating tertiary care facilities. Everyone would like a
specialist unit on their doorstep, but such a proliferation
may mean that units are underutilised or, more probably,
full with “inappropriate” cases—that is, patients who
could be treated perfectly well elsewhere at lower cost.

r

References

1. Relman AS. Cost control, doctors’ ethics and patient care. Issues in
Science and Technology. 1985; Winter: 103-11.
2. Etheredge L. Health care financing in 1990, or what will happen on
the San Andreas Fault? In: Ginzberg E. ed. The US Health Care
System: A Look to the 1990s. Totowa, New Jersey: Rowman and Allanheld; 1985: 89-107.
3. Herzlinger RE, Schwartz J. How companies tackle health care
costs: Part I. Harvard Business Review. 1985; July-August: 69-81.
4. Ginzberg E. What lies ahead for American physicians: one economist’s
views. 1985;253:2878-9.
5. Eisenberg JM. Physician utilization: the state of research about physi­
cians' practice patterns. Med Care. 1985;23:461-83.
6. Eisenberg JM, Williams SV. Cost containment and changing physi­
cians’ practice behavior: can the fox learn to guard the chicken coop?
JAMA. 1981;246^195-201.
7. Grossman RM. A review of physician cost-containment strategies for
laboratory testing. Med Care. 1983;21:783-802.
8. Sims PD, Cabral D, Daley W, Alfano L. The incentive plan: an
approach for modification of physician behavior. Am J Public Health.
1984;74:150-2.
9. Martin SG, Shwartz M. Whalen BJ, et al. Impact of a mandatory
second-opinion program on Medicaid surgery rates. Med Care.
1982;20:21-45.
10. Warner KE, Hutton RC. Cost-benefit and cost-effectiveness analysis
in health care: growth and composition of the literature. Med Care.
Discussion
1980;18:1069-84.
11. Doubilet P, Weinstein MC, McNeil BJ. Use and misuse of the term
This paper has contained a discussion of ten basic no­
’cost effective’ in medicine. N Engl J Med. 1986;314:253-6.
tions of health economics. These ten notions hardly rep­
12. Wickings I, Coles JM, Flux R, Howard L. jteyte* of clinical budresent the economist’s “ten commandments,” but we do
"geting and costing experiments. Br Afied Z 1983;286:575-8.
13. World Health Organization (Regional Office for Europe).
q
believe they provide a useful foundation for clinicians
Guidelines for Health Care Practice in Relation to Cost-EfFectiveness. toILwishing to understand this increasingly important field.
Euro Reports and^tudiesJJjCopenhagcn-. World Health Organization.
For those clinicians wishing to take matters further a (X\14. NeUBauser dTLewicki AM. What do we gain from the sixth stool
guaiac? N Engl J
;293:226-8.
number of textbooks exist (18-22), although in the main
X15. Williams A. Economics of coronary artery bypass grafting. Br Med J.
these are not ideal because they demand a higher level of
1985;291:326-9.
economics expertise than clinicians are likely to possess, <16. Levine MN, Drummond MF, Labelle RJ. Cost-effectiveness in the
' diagnosis and treatment of carcinoma of unknown primary origin. Can
or because they are structured around economics con­
Med Assoc J. 1985;133:977-87.
cepts (supply and demand) rather than around notions <17. Stason WB, Weinstein MC. Public health rounds at the Harvard
School of Public Health: allocation of resources to manage hypertension.
with which clinicians are more familiar.
N Engl J Med. 1977;296:732-9.
There is no space here to discuss teaching methods in
1 S^Cullis JG, WestlPA. The Economics of Health. Oxford: Martin Robdetail, but we have found that the most promising ap­
ertson; 1979.
19< Newhouse JP. The Economics of Medical Care. Reading, Massachuproach is to develop problem-based materials dealing
setts: Addison-Wesley; 1978.
with practical issues that clinicians feel are relevantx
20<Sorkin AL. Health Economics. Lexington, Massachusetts: D.C. Heath
These materials could concern the economic issues raised
and Co; 1975.
2F/ Jacobs P. The Economics of Health and Medical Care: An Introduc­
by the treatment of individual patients, the ways in which
tion. Baltimore: University Park Press; 1980.
health policy (such as the encouragement of prepaid
22^ Feldstein PJ. Health Care Economics. New York: John Wiley & Sons;
group practice) affects clinical practice, or the need to be
1979.
23. Department of Clinical Epidemiology and Biostatistics,
able to appraise economic evidence critically in order to
McMaster Health Sciences Centre. How to read clinical journals:
assess claims for the development of services (23). We
VII. To understand an economic evaluation (part B). Can Med Assoc J.
1984;130:1542-9.
hope that this article stimulates clinicians to learn more
PRICE

92

July 1987 • Annals of Internal Medicine • Volume 107 • Number 1

CLINICAL EPIDEMIOLOGY ROUNDS



How to read clinical journals:
YU. To understand an economic evaluation (part A)*
Department of clinical epidemi­
ology AND BIOSTATISTICS,

McMaster University Health
Sciences Centre
Those who plan, provide, receive or
pay for health services face an inces■ sant barrage of questions such as the
following: Should clinicians check
the blood pressure of each adult who
walks into their office? Should plan­
ners launch a scoliosis screening
program in .secondary schools?
Should patients be encouraged to
request annual check-ups? Should
local health departments free the
limited numbers of nursing person­
nel from well-baby clinics so that
they can make home visits to pa­
tients with hypertension who have
forgotten to present for their check­
ups? Should hospital administrators
purchase each and every piece of
new diagnostic equipment? In other
words, who should do what to
whom, with what health resources
and with what relation to other
health-services?
.....
The answers to these questions
are most strongly influenced by our
estimates of the relative merit or
value di the alternative courses of
action. This pair of clinical epidemiolcgy rounds is concerned with the
strategies and tactics whereby these
estimates of relative value can be
ascertained and interpreted; that is,
with the evaluation of health ser•Parts I through V of this series were pub­
lished in consecutive issues of the Journal,
Starting with the Jan. I. 1981 issue, and part
VI was published in the Feb. 15. 1984 issue.
Reprint requests to: Professor G.L. Stoddart
or Professor M.F; Drummond. McMastef
Universicy HeatflT Sciences Centre. Rm.
2C13. 1200 Main St. W, Hamilton. Oht.
LSN 3Z5

ft

Methodologic criteria for assess- {
vices. More specifically, the guides
we present here focus on one type of ing efficacy, effectiveness and avail- .
evaluation, sometimes referred to as ability evaluations have been des-a. <
economic evaluation or efficiency cribed in an article by Sackett, , ?
evaluation. In this type of evaluation from which the above questions have ; V.;;
we are asking Is this health proce­ been drawn. These questions were J- ft
dure, service or program worth also addressed in part V of this •’c
doing compared with other things series/ so they will not be reviewed t
?
we could do with the same re­ here.
This
pair
of
rounds
is
intended
for
••£
sources? Are we satisfied that the
health care resources (required to those who use, rather than those ■ *
make the procedure, sen-ice or pro­ who generate, evaluation data. Con- £
gram available to those who could sequently, it stresses data interpret!benefit from it) should be spent in tion rather than data acquisition. It
is meant to help the careful reader, g
this rather than some other way?
It is imperative to note that al­ acquire healthy scepticism regarding • xg
though an economic evaluation pro­ claims about the efficiency of health S
vides important information to deci­ procedures, services and programs. S
thisrpart
we describe the rationale
sion-makers, it addresses only oneIn
----------------------dimension of decision-making about for and clearly define the nature of
health programs. An economic eval- economic evaluation. We then idenidea- ;^
L
______ __is
____________

uation
most useful, and‘ appropritify the basic types of economic
evaluation, distinguishing
distinguishing the
the ptff-j®
ate, when it is preceded by three evaluation;
other types of evaluation, each, of pose and characteristics of each. IiL®r.
a -ig.
*
’’ , part ’3
heip you become a-ig
which addresses a different
question,
B we will 'help

more
critical
assessor
of
evidence
by
as follows:
«
’ --------io the uvaiui
• Can it work? Does
health identifying the wBvaMWMm.
elements wof -Ia sound]A
progfam
do
economic
evaluation,
procedure, service or
— —-------- ----------------- illustrated
more good than harm to people who through case presentations and tne,^
fully comply with' the associated current
------- x literature. We will also d*-*#'
recommendations or treatments? cuss the limitations, of economic^
This type of evaluation is concerned evaluation techniques.
with efficacy.
|
• Does it work? Does the proce­ Why do an economic evaluation?
dure, service or program do more .
good than harm to people to whom Case presentations
>v

••
i’-jj
it is offered? This form of health
A.
You
are
a
staff
surgeon
at
a
care evaluation, which considers
both the efficacy of a service and its busy community hospital. As the
acceptance by those to whom it is population served by your hospti"
offered,, is the evaluation of effec­ grows, so does the need for surgKal .beds. The scheduling of
4
tiveness or usefulness.
• [s it reaching those who need minor surgery has become chaotic, j
it? Is the procedure, service or pro­ but there is little prospect for any
gram accessible to all people who increase in the number of beds. You .
are^tne-that a “day surgery
couid benefit from it? Evaluation
this type is concerned with availabil- gram would
thatihe^i
ity.
... .pr-ssure, but you aiso know —

CA'i MED ASSCC I. VOL. ’ ’ri. JL'NE \ I



'

tyspital board will not approve the- due to chronic lung disease, then
,<iablishment of the program until prevention programs (e.g., related to
JdU give them some “hard data” on cigarette smoking) may represent a
whether it really will be less expen- more efficient avenue and should be
added to the programs competing in
jve and, if so, by how much.
B. As a member of your hospi- the evaluation.
j ;
• .
• The viewpoint assumed in an
rjfs medical advisory
committee
are
asked
to
approve
the
launchanalysis
is important. A program
\0U *
■ng of a renal transplant program. that looks unattractive from one
Your colleagues claim that trans­ viewpoint may look significantly
viewpoints are
plantation is “highly cost-effective” better when other...
jnd in support of their claim have „ considered. Analytic viewpoints may
jtzached an article1 to their request. include any or all of the following:
t
in­
> ju realize that you must wade the individual patient, a specific
stitution,
a
target
group
for
specific
the article in order to make
services, the ministry of health’s
j; izielligent decision.
C. As cs:e of the few physicians budget, the government’s overall
cn your ’ocal health council you are budget, and a focus on the com­
isked to comment on whether your munity or society.
• Without some attempt at mea­
:c**n •otJd be better served by a
ccmpreheasive blood pressure surement, the uncertainty surround­
screening program (that claims to ing orders of magnitude can be
••pent untimely death) or by an critical. For example, when the
•^..Juenza immunization program American Cancer Society endorsed
(that claims to prevent days of dis­ a. protocol of six sequential stool
ability). Because these programs tests for detecting cancer of the
seem so dissimilar you feel that .you large bowel, most analysts would
must come up with an organized have predicted that the cost per
i
> isx of comparing their costs and detected case would increase
markedly
with
each
test.
But
would
; benefits.
they have guessed that it would
reach $47 million for the sixth test?4
I Comment
i
While this is, admittedly, an ex­
In each of these increasingly fre­ treme example, it illustrates that
quent situations you are being asked without measurement and compari­
for an economic evaluation of alter­ son of outputs with inputs we have
native services. Why is economic little upon which to base any judge­
# In fact.
evaluation so important? To put it ment about value for money.
simply, resources — people, time, the real cost of any program is not
facilities, equipment and knowledge the number of dollars appearing in
i — are scarce. Choices must and will the program budget but, rather, the
be made concerning their deploy- health outcomes achievable with
ment, and. methods such as “what some other program that were forwe did last time”, “gut feelings” and gone when the resources were com.even “educated guesses” are not mitted to the first program. It is this
/ always better than an organized “opportunity cost” that an economic
consideration of the factors involved evaluation seeks to estimate and to
in a decision to commit resources to compare with program benefits.
one use instead of another. This is
What does economic eialuadon
true for at lea^t three reasons:
• Without systematic analysis it mean?
it difficult to clearly identify the
Two features characterize an eco­
^levant alternatives. For example,
>n deciding to introduce a new pro­ nomic analysis, regardless of the
gram (e.g., rehabilitation in a spe­ activities (including health services)
cial centre for patients with chronic to which it is applied. First, it deals
lung disease) too often little or no with both the inputs and outputs,
effort is made to describe the exist­ sometimes called the costs and coning activities (e.g., episodic care by sequences, of activities. Few of us
family physicians in their offices) would be prepared to pay a specific
for a package
whose contents
with which the new proposal must price
,
.
be compared. Furthermore, if the were unknown
«• •» until we could see
object is indeed to reduce morbidity what we were getting for our money.
L-

.

*

-



.



A

a

.9



_____

Conversely, few of us would accept
a package, even if its contents were
known and desired, until we knew its
price. In both cases, it is the linkage
of costs and consequences» that‘al­
lows, us to reach our decision.
Second, economic analysis con­
cerns itself with choices. The scarci­
ty of resources, and our consequent
inability to produce, all the desired
outputs (even efficacious therapies),
means that choices must, and will,
be made in all areas of human
activity. These choices are made on
the basis of many criteria, some­
times explicit but often implicit
Economic analysis seeks to identify
and make explicit one set of criteria
that may be useful in deciding
among different uses for limited
resources.
These two characteristics lead us
to define economic evaluation as the
comparative analysis of alternative
courses of action Or terms of both
their costs and their consequences.
Therefore, the basic tasks of any Aeconomic evaluation will be to identify, measure, value- and compare
the costs and consequences of
01 inc
the
r.
alternatives being considered. These ’
tasks characterize.all economic eval- ’
uations, including those concerned ;7
with health services,

These two characteristics of economic analysis may be used to distinguish and label several evaluation
situations commonly encountered in ".§■
the literature on health care. In
Table I the answers ta two questions
— Is there a. comparison of two or ~
s ■
more alternatives? and Are both
costs (inputs) and consequences
(outputs) of the alternatives exam­
ined? — define a six-cell matrix for
evaluation situations. Dr cells 1 A, IB 7
and 2 there us. no comparison of <
alternatives —that is; a. single ser­
vice or program ti being “evaluat- ‘J
ed”. To put it 'more* accurately, the
service or program- is being “de­
scribed*’, iince evaluation requires
comparison. In ceil 1A, only the
consequences of the service or pro­
gram are examined; thus, the evalu­
ation is called an outcome description. In ceil IB, since?only costs are - ?
examined, the evaluation is called a_
cost description^ Im ceil 2, both the1 r
outcomes and the? costs of' a? single
service or program are described;
thus, the evaluation is caBetia- cost—
outcome descriptions Am example of*'

_ _ —-------- A1— — A

^.1—

mA

A mrl

CAN MED ASSOC J. VOL. 130, JUNE!, 1984-

«5I«« If

’■**

1429

v.l;

r

fits.* For example, in .their cost­
benefit analysis of screening for
spina bifida cystica, Hagard and
coworkers12 assumed that therapy
would be given to children with
spina bifida with the “no-screening”
alternative.
Weisbrod and collaborators11 did
attempt to quantify and value a
wide range of costs and benefits in
their study of conventional hospitaloriented versus community-based
<
r
_o
... rpatients
_ ents with
programs
for
with mental
mental
illness. They found that although
the community-based program was
more expensive the costs were more
than offset, by the program’s value
in terms of the number of patients
who could become.' or stay, em­
ployed. (These investigators used
earnings as a dollar measure of the
benefits.)
A second measure of value, which
is more difficult to obtain but preferred by many analysts, is "utility”,
Utility refers to the value of a
specific level of, or improvement in,
health status and can be measured
by the preferences of individuals or
society for a particular set of health
outcomes. The notion that the utility
of an outcome, effect or level of
•Notice the different treatment of the “donothing’’ alternative. Cost-effectiveness analyses often implicitly assume at the outset that
•a tenable “do-nothing’’ alternative docs not
net benefit. While this may be quite a
realistic position for health care decision­
makers to adopt, cost-effectiveness analysis
may lead to a decision to undertake a pro­
gram that does not “pay for itself’; that is,
one that entails a net resource cost instead of
a benefit.

. <•.

--1

health status is different from the
An analysis that uses utilities aj
outcome, effect or level of health measure of the value of the efTectt "!
status itself can be illustrated by the of a program is termed a- cost^ L*
-..—.r!-: Suppose “t utility analysis^ The Yesults of t.’W'
following example:
that
twins, identical in all respects except cost-utility analysis are' expressed in
occupation, one being a sign painter ’terms of the cost per "healthy day” A
and the other a translator, both or per "quality-adjusted tife-year” 4
broke their right arm. While they gained by usmg one progranrinstead
would be equally disabled (or, con­ of another. Exam pies of tost-utflity f
versely, equally healthy), if we analyses mclude the study by Stasoo
w a and Weinstein”^qnjsfrategies forthe r ;’
asked them to rank their “having
broken arm” on a scale of 0 (dead) managcmeritlSF fend’d ^hypena^ ' l
to 10 (perfect health) their rankings sTon and thafby Boyle: and'associ^j;’
might differ considerably because of atesu;onmeonataF mteM^care^nf|
a
the significance, in this case based infants oFvery; low birtfi weight.
on occupation, that each twin at­ . The- different: cfiaractoistics^
tached to arm movement. We would the four types oFfiiil economic enkife
also expect that their assessment of uation^^^a^-mihfmmtfon^'aBt^^
the utility of treatment — that is, effecnveness, cw^-benefftand cost-^;'.^
the degree to which treatment of the utility -^arc^raminarized: in Table* g
fracture improved the quality of IL Two- further.gdnts. wanant g>phasis- Hrst.4±Le maiir
"
their life — would also differ.
Although a utility analysis is a classifyirig^ the' types’of full eco6oi^’^|’
relatively new technique in health ic evaluationJ‘isr^^fflustate.'
care evaluation, it is considered ex- jlifferenU^j^^^
j.
tremely promising because it allows completestudiM,
notJo prescribe
completed'studies,^notto
presdibea^
<4quality-of-life’”r adjustments to a porticularsUdyJQfterr

" r
"quality-of-life
at the
given set of treatment outcomes ^nm^o^5^nomic^^MdOTthrJ
while providing a common denomi- 'analyst* may^ noLbe able ta predict^ $
nator for a comparison of costs and .what
outcomes of different programs. The tak^’
common denominator is usually expressed as “healthy days or “qual
quaf-­
TTiijnminin MTr-j—.Jgjadv5ncc<^^^toa!
ity-adjusted life-years
life-yedrs ”i* arrived at ?Jarown;
known.jffijadvanc^
“at^diincal^^j
by adjusting the length of time af- ^TOFuationg^DQffio^ik,;eraIuatidn^nW^ov^ twnvtt
ealmytt:
*—
^5^ through
through the
the health
health outcome
outcome ja/hav^ltdentic^r^/effects^^ereby^
fected
according;
1 to the utility value (on a ^Tetiucmg^ a^ cSJ
according^
scale of 0 to 1) of the resulting level sis to as ajst-nijIinni^tioTr^aly^^
».•_
__
*<
__________
n__________
1 ' Ei^e^n^^o^^t^aMysaiW,'
ofr- «health
status.
Many
analyst^
find
this
method
of
valuing
the
cOnsethis method of valuing the cOnse- iarcsbm’etim^S^t^^^%.tad>^
quences of health care alternatives ; I^’ai'^tiCTfiH^thorn^'prcbleim^^jI^
nr^frrtihl*
1* .used, both i?
andTaksociat«^iBedpreferable tn
to vaInina
valuing them
them inrdnlim.Mok Boyle: andTaisociates
cos
t^
bene
fTtand-afoost^utifity
aM&jJt*
lars.
■ ‘ £
-



r

-

Table II—Measurement of costs and consequences in economic evaluations

W '

A
’ •

Type of economic
evaluation

Measurement/
valuation of
costs in both
alternatives

Cost-minimization
Cost-effectiveness

Dollars
Dollars

Identical In all relevant respects j
Single effect of Interest, common to the tw<r ^NatunfuniferJeg^ yeafroCHI^^i^;
alternatives but achieved to different
gainetLdayyordlisabillty saved^-i K
degrees
4
’V-W? umhidCbloo<rpiressurereduciiow^

Cost-benefit

Dollars

• Cost-utility

Dollars

Single or multiple effects, not necessarily .^Dbl&tsr;
common to the two alternatives; common
effects may be achieved to different
degrees
Single or multiple effects, not necessarily
common to the two alternatives; common- ;^"qvaiRy^adJistedlife^ea0?^^
effects may be achieved to different
.
‘’VrtWWffl&SuM
degrees

i

1432

---------- Consequences



CAN MED ASSOC J. VOL. 130, JUNE I, 1984

4■I
I

II

>861 ‘I 3Nnr *O£I 1OA T DOSSV Q3W NVD

qsAjm Ajnpn->«>3
siUfBuu lyauaq-jsoj
pU|«ub ssauaAp3ajj»*)S03
sisApnnr uopBziiuiuiui-jsoo

SlsX|«UB
1603

oopvnpMa 9iiiioiio3d ||nj



uopcnpAd

7

uopcnpAa
ssaudJipajja
JO X3C3IJJ3

uopvn|Bxa |B{MBd

>»□

z

flI

uopvnpA» |vpjcd

:

:<

■j

J

I
-J

°N

ON

■____

,i • '
GiA-.•
------ -----------•patujojjad XuBnpB uoiiBiqBAa jo adX) aqi
: ’ om soipnis qi°q nj •sowa osoouba •-iiopi
—ri oje
— soouonbosuoo oqi
. oseo siqi JO SJOlBOIpUl 91BJn33B UBMfB IOU 3JB SUOIJB
JO nnuaq fEuinSoi joj iuouheojj 2di uj iou soopi(XJo3jns XEp) Joqio
oqi
___ jo
r_sapp
n__
-n|BA3 oiiuouooa uo_________
S9pi)JB FMSifqnd
”--MOHOJ
’ sXeis* TBiidsoq oApojodoisod ojjqM ‘iq3iu' ] iseoj is joj uoissiiupE 3lJ1
xp^j| him oppjb sup jo siapraj
SB -pUE
uo ,‘SJoijOMOO puE JOHE^ jEiidsoq sajmboj (Xjo3jos luoiiBdui) - -paziu8o33j OABq• A1U1
/—• SUOpBn|BAS
jnoq-gp
! io pnE ‘spioquomaq pire SBjuwq . auo iBqi idaaxa ‘padsai lUBoyiuais 3.u.ou<x» jo npa »m j°
«i»to «y.

hum siuound
siuopod joj
joj Xjo
joSjus
jos Xnp uo
oo
• ‘sodSeohoo poE yossng Xq osoqi are
AjoSjos jounu oi sooioqo jo soipnjs
0M1 SJEU1 lEoiono'poiiOJiuoo 8uimo|
.pj XpiEipouiun jo qiiM uopspossB

b
IJ

-U3 3UI0S ttJinbM 'si^iBur uopra
-iniunu-w» B'q3nojqj uoiiBiqBAS ai
-iuouooo pnj e ‘JBAOMoq ‘o|dpuud nj
’sonbraqooi oqi nooMioq oouoiojjip
on si ojoqi ‘(sisX|eue uopEZiinraiui
—soo e m jEonuopf oq oi psiuns
•a oie Xoqi uoyo* ooi) ivouuspi
a® joaiipujoj/p 9i{2 Jo muojino
Ji oopoEJd uj ‘Qt poo ui sis
: -jpnt lSTO oqi mojj SKJJW>1 Wl
jo j, jpo 01 sisXjeue uopEZinnunn
■ -isao oqi Moq JopuoM Xeui noj^
_ -ajupsooid JEO,
-fims jod jsoo jo slseo oip no opsui
3q him nosuEdwao aouodijp pd

g
w'

joqio ou in Jojjip ‘BiEp ssouoaii
-oojjo jo uoiieuiuiexo ue uiojj ‘pun
isojoiui jo ouiooino oqi qsqduioooE
qiog -siinpB ui Xjo3jos jouiui JOJ
suiBjSojd omi pojopisuoo y

qiiM uosuBduioo pire uojiEnjEA
luouiojnsEoui Jioqi sioojje soouonb
-osuoo oqi jo ojniEU oqi Moq oos
oi j8jij» pquosop
suoijmiwsajd
OM 1J
□seo aajqi oqi Japjsuoo sn
sn wt X]qB
-jopisuoo jojjjp Xeui pouniiEXo 8uioq
soAiiEiuopE oqi uiQJj Suiuiuiois
soouonbosuoo oqi jo omjEU oqi ‘jo
-AOMoq ‘suoiieojbao Xouopijjo isom
ui JEpuiis si sJEpop oi juouiojns
-eoui luanbosqns Jioqi pnE sisoo jo
sodXi snoiJEA jo uorjEDijnuopi oqj.
4Kmonnpo ouies aqj
osn suopEnjBAO Aouapijja ije oq

‘*“
onnonooa [pij,, pqBj
M UopBn|BA9
.
..
oip Jopun p poo ui pwp sonbiuqooi

njiqs Xjagjns Asp iiqi jiwpw sip
*■^5) nsoo jo uopnqujsip oip in pa
-^u.^
jqS.w 3’A

.7

-

VI

uosusduioo v »jaip sj

^paiqiuixd sdjyvujdjp aqi jo
(sjndjno) sdaoMibasuoo puv (sjndui) spoo q>og ajy_________________________ •_____________ _____________
j
suopBnpjd dJB3 qpBdq jo sopsudpBJBqo 2ujqsin2upsiQ—] BjqBi

7
7

lSdAp«IU3)|B
9J0U1 JO OM) JO

saduanbdsuoa
X|uo sduiuiBxg

SJSOd

X|uo sauiuivxg
QQA______________________________



uopduasap
aiuo3>no

uopduosap

uoydposap
auloaino—JS03

OCfl

';

u0flD2llUlUlUi-^O0 B papEO UOJJO SI
W se qons sisXjeue uy -aApEuiopE
^Aisuodxo ;sEOi oip aoj qojsos b
____ ‘uoqj ‘si uopEnjBAO Xouop
^Ippuasso
•UP aq_£ -sisco piojojjip IE XpEiuns
•aid
**id q3noqj
qSnoqi ‘suiEiSojd
suiEJfiojd omi
omi oqi qpa
qpM
(suopBjodo jo joquinu jEopuapi)
uibs oqi
sojSap □3uiBS
oqi oi
oi psAOiqoE
poAOjqoE oq
oq
Pl1133 (poiojdwoo Xpnjssooons uoiie
l»wiu! jo omopino oqi -pap

s ■

-Wd jo ssolAJos qj^qI J0
-uoo pus sjsoo oqi jo SuipuEjsjopun
jno ui soSeis ojEipoiujoiui juEiJodmi
juosoidoj Xbuj Xoqi joj ‘juEyodmiun
mb sopsuoioEJEqo osoqi qpM soipnis
jEqj X|duii jou
iou soop siqi /^suop
j’^suop
lEqj
-EnpAO |EpjEd„ popuSisop uooq pE
OAEq Xoqi uoseoj siqi joj -uopEnjBAO
oituouooo joj suoippuoo oqi jo qioq
spjpj ^PJPM siP° 3S3lP J0 3U0M

*onssi snopuoiuoo b
iou SEM SUE0U1 oojqi oqi jo SSOUOAll
-oojjo oAiiEpj oqi osnEooq luopyjns
SEM sisXieue 1S00 e iBqi ponSJE Sjoqi

uuaj-Suoi

SUEa“ ’L?]

J U0 ,S3}B
jo sisoo aAijEJBdmoo aip
,
-ioossb puB U0SM03 Xq isq) si Xpms
b qons jo oiduiBxa jiwmj V 'sis^jdud
isod b popEO oq Xbuj pauuojjad
Xpnis aqj uopBnjis b qons ui :poui
-urexa ojb soApBUjapB aqi jo sjsoo
□qi Xpo g£ poo uj -pEj ppoM spuj
poiuip paZlIUOpUEJ ISOLD DO ajuj
-BJajp iDEUodini aqj qojq* u! IP3
SI
.
uof^n^
up hoiibd si

SSaU3A.llDdJfi JO ^DVDlffd UB p9||B0 SI
DonznfEAd jo adX) slip ‘snip ipajsd
-UI03 □JE S9APBIU9)|B oqi jo saousnb
-9SU03 oqj Xfiio Vf ipo uj X|snoauin
-puns pauiuiEXQ jou aJB dApEUJd)
—i sjsoo
-iE
JOP-B
oq) inq pajBdmoo ojb s9ApEUJ3}|B
3J’0UI
'EniEA3 U!B’U°?J£f“. nXE„uMW-t’dApEiuaip ub qjiM pun
9JE0 Xjeuojod aqi jo saouanbasuoo
------ oq) QjBdlUOO oi )dlU9)
puE sisoo^
poijdxa
ou sem ajaqi ‘joasmoh
-;b
■psAES saAii 'jo jaquinu Xp)|i| aqi
paiBiupsa pus pun qons ouo jo sisoo
oqi poiuosojd ’sJO)B3psoAUi osoqi
•sjiun ojbo Xjeuojod uo jpuXog puB
pouXog Xq pup’si Xpn;s jo odX] siqi

...........

short-stay alternative was compared ture. Ludbrook’ provided a more
with the traditional inpatient-treat­ recent estimate of the cost-effective­
ment. An example of a cost-minimi­ ness of treatment options for pa­
sation study of a different health tients with chronic renal failure.. In
care issue is that by Fenton and addition, a number of studies have
collaborators ’ on home versus hos­ compared the cost-effectiveness of
pital treatment for patients with actions that do not produce health
effects directly but that achieve
psychiatric problems.
ci clinical
vmnvu. objectives that can be
other
Our case B focused on the prolon­
linked to improvements
inVlCdllJ uimvM
------ -----------------gation of life after renal failure and clearly
f
vv tw
patient
outcome/
For
example,
Hull
compared the costs and conse­
compared diagnostic
quences of inpatient dialysis with and associates'’ <
strategies
for
deep-vein
thrombosis
those of kidney transplantation,. The
outcome of interest, years of life in terms of the cost per case detect­
gained, is common to the two pro­ ed. Similarly, Logan and col­
study of patients with
grams; however, the programs may leagues,11 in a
hypertension, comparedI care at the
have different degrees of success, as
well as different costs, in achieving worksite with care at a physician’s
this outcome. Consequently, we office in terms of the cost per 1-mm
would not automatically lean toward Hg drop in the diastolic blood pres­
the least expensive program, unless, sure.
reminds us that weof course, it also resulted in greater , Our
wu< case
— C
- ---------------------prolongation of life. In comparing cannot be assured, or assume, that
these alternatives we would usually the consequences of alternative procalculate the number of years of life grams are identical. In addition, it is
saved and compare the cost per unit frequently .....
not possible^
to_ consider
r____ _______
of effect (i.e., cost per year of life the outcomes of interest as a single
gained). Such an analysis, in which effect common to the two altemacosts are related to a single common lives. We may be interested in efeffect that may differ in magnitude fects that, although
_*i common to the
between the alternative programs, is two
two alternatives,
alternatives, are
are multiple,
multiple, or
or we
we
usually referred to as a cost-effec- may
may identify
identify single
single or
or multiple
multiple efef­
tiveness analysis. The results of fects that are not common to the
alternatives. The
The first
first case
case isis easy
easy to
to
such a comparison may be stated alternatives.
understand ifif we
we make
make two
two extenexteneither in terms of cost per unit of understand
----effect or in terms of effect per unit sions to our case B: include home
r

dialysis
in
addition
to
hospital
dialyof cost (years of life gained per
dollar spent). The latter°is particu- sis and kidney transplantation, and
larly useful when one is working include quality of life (perhaps meawithin a given budget, provided the sured by the occurrence of marital
alternativa are not of a radically disruption) and the frequency of
different scale.
medical complications as conseFurthermore, although the alter- quenca of interat in addition to
natives in this example are similar, years of life gained. To pursue x
in that both could be considered cost-effectiveness analysis we now
variants of a renal program, a cost­ have to compute cost-effectivenas
effectiveness analysis can be per- ratios for three effects. In the event
formed on any altemativa that have that one alternative was not clearly
a common effect. Thus, kidney superior on all three counts, we
transplantation could be compared would have to eitber daignate (imwith heart surgery (or even manda­ plicitly or explicitly) a primary ef­
tory seatbelt legislation) if the com­ fect on which to base the compari­
mon effect of interat was years of son or find a method whereby the
life saved. Similarly, an influenza multiple common effects could be
immunization program ccould
—— be combined into one common denomi­
compared with a home care pro­ nator.
The need, for a common denomi­
gram (or even a community safety
uatv.
w measure the consequences
i
nator
to
education program) if a common
effect of interat, perhaps days of of altemativa is even more apparent
disability avoided, could be identi­ in case C, where we attempt to
compare the effect of a hypertension
fied.
There are many exampla of cost- screening program (the prevention
effectivenas analysa in the liters- of premature death) with that of an

influenza immunizatibn program r
(the prevention of days of disabili- jx
ty). Here the outcome of interest
differs between the alternatives.
Consequently, a meaningful cost- .j,. geffectiveness comparison is impossible.
~
In situations like these, when we
|
need a common denominator to
compare
the outcomes, analysts
frei
- ,
____ ___ B**
quentiy attempt to go beyond
sidering the specific ^effects and^o
|
:attach
____ ' a measure of
-**--•
value —
to *the- .-~ \
effects resulting from a particular
““■ice or fprogram. One. measureof
service
value is dollars; the effects oF a: -gi
program wiil
will often
often be
be expressed
expressed inm | terms oF their dollar benefits to
facilitate a comparison with the pro- g,
gram costs. This, of course, means
that we have to translate the effects, * |
such as days of’disability avoided, jgi
----- • ei-r ' • -»
years of life gained, medical comph- e
cations avoids and evai nuntal
disruption avoided, into
mto their dollar
doUar

3

a

____

-

- . ..

p,'

depending on the type of effect, it is >
sometimes both appropriate and fea- j.’
sible to attempt it. An analysis that
measures'6otfa_the- costs an<£ theamswjuenccr oF alternatives m dollars is.
is. usually
usually called
calledIta:cost-benefit
cost-benefit
lars
tanafjim^te^eralts
analysis might
might'be
be stated
statedeither
either as
as a
analysis
¥
J
tn doll
dollar
hene*^1 p
ratio oFdollar costs to
ar te
ne- ^
fits or
.f
. as .at ample- stun (possibly
negative)'representtng theLnettene:^^.
fit z(or
las)' 6Fone program dyer ~
“- r
another.,
A cat-benefit
analysis, at least in
. ..
theory,
“c_:
absolute^
"x‘ benefit
“ bF a* program, in’
addition,
to
information
- -------------- ”7'7 eta
7^ itr rela- J:
tive performance^ that
r cost—
benefit analysis estimata the value.
of resources used by each program
compared: with the: value- of rc^
sources: the- program might save: or
create;. This view implicitly assumes" W
that each- program is. being com-^T
pared with ^ “do-nothing’*'aiterna^^ i
tive
tive that:
that? entails-no- costr andr
and-jo^
jrto:
benefits. However,, nr practice ar y
cost-benefit analysis usually comcat-benefit
:.y
para only catr and benefits ’that
can
'— easily be expressedindollarsrso
expressed, ir dollars; so .
very few analysa can aspire tor this. > g
wideri7 role
wider
rolc. Also; there
there: are very
very: few
instances: iitwhiclL absolutely_noth-r^J
ing is
ir done
done-ti
tackle a.
az giveir
givetr hgiithr
health: J
ing
to tackle
problem; ~sor
“sr firmest*
“t
fir mat^cat^benefit'-*t
analyses^ the
the: implicit “?do-nothihg
’^ jgf
it,
analyses
do-nothing^
alternative has some^catr
costs.aand
bene^J
nd^htaife^

I

■■

s

k-

CAN MED ASSOC J, VOL. 130; JUNEri, !9S^

....

.

MSP

■>

■-Nn^RQNG'SR
----------- -

THE 8 HOUR NITPOGL'iCERlN

Nilrogiycef"! sustained-release tablets
Therapeutic classification
Anti-angtnal Ageni

Warnings

'

Data on the safe use oi Nitrong SR during tha cany
phase of myocardial infarction (the period during
which climcai and laboratory findings are unsaotel
are insufficient to estaonsn safety.
The use ol Nilrong SR m patients with congestive
heart failure regures carenS dimcat anc/or nemo­
dynamic momtonng.
Nitrate dependence may occur in catieras with
chronic use. To avoid possbia withdrawal ertecrs.
the admmrstration ol Nitrong SR snoutd graouaify
be reduced over 4-6 weess in •ndustry •onera
continuously exposed :o ncate*. ohesx cam. acute
myocardial infarcoon and even sudden ceasi -aw
. occurred during temporary •rtnerawai & n.tnce
exposure.

Precautions

Headaches or symotoms ct typctenson. sucn as
weakness or dizziness, partcutarty when arising
suddenly from a recumoent position, may be cue
to overdosage. When trey occur, the dose sraxiid
be reduced or use ol Nitrong SR discontinued.
Nitroqtycenn is a potent vasodilator and causes a
slight decrease « mean Wood pressure (appraomatety 10-15 mm Hg) in some patients when used
in theraoeutc dosages. Caution snouid.be extcaed n usaig tre cnx; n panrrxs -no are prone
ta or who might be a‘*aed oy hyoctension.
Nitrong SR Tablets ve -nr mrwncecJ rr mmeeaie
rwiet ol acute anactt cr angma oeccxs. SUxmguai nitrogtycenn preparations should be used for _. ...
this purpose.
Tolerance to this drug and cross tolerance to other
nitrates or nitrites may occur.
Adverse Effects
Headache is the most common side effect espedafiy when higher dosages ol Nitrong SR are used.
Headache may be treated with concomrtant
adrmntstration ot mad analgesics. If headache is
unresponsive to such treatment, the dose of
Nitrong SR should be reduced or the use of the
product discontinued.
Lass frequently, postural hypotension, an increase
In heart rate, faintness, flushing, dizziness, nausea
and vomiting have been reported. __
Symptoms and treatment ol over doeage
Symptoms of overdosage are pnmaniy related to
vasodilation, including cutaneous Hushing,
headache, nausea, dizziness and hypotension.
Methemoglobinemia is also possible.
No specific antidote is available. Treatment should
primanty be symptomatic and supportive.
Dosage and administration
Adult: Recommended Initial dosagejs 1 tablet 3
times a cay before breakfast late afternoon
before meal and before retiring. Dosage may be
jncreased progressrvety up to 2 tablets 3 times
a day.
AvaAabfiity
Sustained-Release Tablets of 2.8 mg — Bottles of
100 and 100Q.

References:
1 HirsNeiier. L. Cuff. Ttier. R»s^ IS. 4. 158 (1973)
Z Elindw. S-. Curr. Ther. Ras.. 7. 12 (1965)
3. Klein. H.O.. and Berger. H.J.. Carrtotogy. 58.
313(1973)
4 Data on He. Rhdne-Pouienc Pharma Inc.
5. Wnsor. 7. and Berger. H. J„ Am. Haul J..
,ot. 90.611-612(1975)

fPriHONf-POUaNC

eWUlBta*.

|ccp" j
1434

we will describe a set of readerf^W^?' '
guides that will help you undent a nd
articles that present economic enlu.
.

ysis in their evaluation of neonatal
intensive care, since each explores a
’ different dimension of value.
Conclusions

Indications
Nitrong SR Tablets are indicated lor the prevention
o< attacks ot ang«a pectons associated weh
chronic angina ol effort
Contraindications
Nitrong SR Tablets are contraindicated m patients
with severe anerma. increased intraocular pres­
sure. increased intracranial pressure and hypoten­
sion. Nitrong SR is also contraindicated in patients
With known idiosyncrasy to organic nitrates.



t-4:

References
'■ • —TT&:- T7 “
From the user’s point of view, the
I. Sackett DL: Evaluation of health
most important consideration is
vices. In Last JM (ed): Maxey. ■
whether the complexity of the analy­
"Public Health and Preventive Mediei^
sis matches the breadth of the ques­ . J41lth ed. ACG New York. 1980: !80(U
1823
tion posed. Cost—benefit and cost—
E Department of clinical epidenrioiogv iwi 'i
’ utility analyses, since they address
hiostatistics. ^Mc MasterUniverxity
outcome valuation, shed more light
_Health science Centre. Hq~. i»
?
on whether the treatment concerned ^■Tdiniel jonnuls:
j
is “worth while compared with
fca nder or oti
other ’■’ratments. Cost-mimmization
Cea, iled Assoc J198Tz I«-c 1156-1 fix
and ctst-cffertivencss anah~^uciti
ly assume that the trcatmait is
I98i:
worth while. To assess whether a
Nelhalsex D. Le«icxj am- wW
.
particular evaluation is appropriate
r we
from the srah raci
to the question posed,
user needs
- J the------------''
to te aware of the differences in the S: Reynell PC.R£YNiu.MCi.TU«^^t^£^*^.
rf icar0Blir
... '---- 7.1
analyses.
n_ ___ i\om.
u- to7_om
BrHctat
J 1972:34:897-900
The power of these analytic tech­ 6. Lowson KV, Drummond MF. Buhot
niques aww***
should —
not- be
imjuw
— overstated.
------Costing- new^seiyicei; ai|-uu <k>
*
None of the approaches is intended '
oxygen th«w.
to be a magic formula for removing
Russeu. 117 DevonDI:
i
judgement, resixmsibilitv
responsi
y or risk ®7;
j^GUss-Nr£NEWEU.
i
from decision-making activities,
though each is capable of improving ^dhriciLsociai. andecomxnic^rdBtfioB^^^^^J
the quality and consistency of dedsion-making. They are, at^ mt, _
K
methods of critical thinking, f
proaching choices and often of placDept oTHealth tndS(xHUSecm^||fe^
ing difficult choices out in the open
hmsq, (jondon^.

for discussion. While quantitatively
i
they generate statements about, pro- ^Sorotnjs^Pc Ncuwc
gram costs and consequences, quailtratmentx raanaxlffl.^s^.
tatively they are simply frameworks 1

for comprehensive enumeraUon and
.}
display of economic factors involved l0: hoixtR^Himu JTSacxbtt^
' in decision-making. Whether, the

Sas :

analyses are, in fact, the dominant
concerns in a specific decision and
whether the limitations of am ecoSSSa
nomic evaluation (which we will J^ CXMraiulWPLHkWES
discuss in part B) significantly rc
re-. ^ti^ettoF^worktitehyperta^b^
— . x*2’*ment• P^iaram;^
strict its usefulness in a specific
situation are judgements that, quite
properly, remain the responsibility ^s<«eningi.f^spin^bifidx.cig^^:^^
of the final decision-maker. In this .
sense a , cost-minimization. Jcosteffectivenos, cost-benefit or cost»
utility analysis may represent only a. /^ir Econon^ast-benefiL-an*!^^^^
partial analysis of any specific
l98Or3n4CO^5.J^|
choice.
Of course, identifying an economresourcesta manage
.
ic evaluation is one thing; deciding
whether it has been soundly execut- •
spr
ed, and then whether it is potentially
iataBtve. care'
useful for a particular decision, is
birth-wrighr infant^'Mg^
^•^Sl
1 quite another. Therefore, in part B

’ CAN MED ASSOC J, VOL. 130. JUNE 1, 1984

.

. - ■' r'S-.- • *

CI.INICAL epidemiology rounds

■■ uu 'a
——

How to read clinical journals:
VII. To understand an economic
evaluation (part B)*
!

ly examining the methods used to
produce the evidence. Often it is not provide important information
ology AND BIOSTATISTICS, MChelpful to separate the various ele­ on accounting or management. They
M aster University Health ments of a method so that each can may, but alone they do not qualify
Sciences Centre
be scrutinized, a strategy we repeat- as efficiency statements.
edlv applied in the previous clinical
4
.
epidemiology rounds. Accordingly, the following: From the viewpoint of
The readers of clinical journals in­
we shall identify the key elements of (a) the budgets of both the Ministry
creasingly encounter articles on the
an economic' evaluation and, of Health and the Ministry of Comeconomic evaluation of one or more
through
the use of a set of
c. readers
-------- ’ munity and Social Services and (b)
clinical maneuvers or programs and
patients incurring out-of-pocket
are often faced with the task of guides (Table I), discuss the meth- costs, is a chronic home care pro­
results.
shown in * odologic standards that readers can
assessing their
t
_— .As
— —
gram preferable to the existing pro­
the case presentations in part A, the expect to find in a well executed gram of institutionalized extended
question that readers of such articles economic analysis. Of course, it is care in designated wards of general
are most likely to ask themselves is unrealistic to expect every study to hospitals? Note that the viewpoint
Are these results useful to me in my satisfy all of the guides: however, for an analysis may be that of a
setting? The answer is determined the systematic application of the
’ ‘ ■ or providing insti­
provider
by the answers to the following auides will allow readers to identify specific
tution,
a
patient
or groups of pa­
and assess the strengths and weak­
specific questions:
tients,
a
third-party
(public or pri­
• Are the methods employed in nesses of individual clinical studies.
vate) payer, or society (i.e.. all costs
the study appropriate, and are the Elements of a sound economic
and consequences to whomsoever
results valid?
they accrue). It may be that a
evaluation
® If the results are valid, would
program is preferable from the view­
they apply to my setting?
1. Was a well defined question
point of society but not from that of
This second in the pair of articles posed in answerable form?
the providing institution. In such a
on'understanding economic evalua­
Such a question will clearly iden­ case the Ministry of Health may
tion relates mainly to the former
’ ‘ giving
‘ j an incentive
being compared wish to consider
i
question. It is designed to assist tify the alternatives
the providing instituUon ^nsure
users of economic evaluations in and the viewpoint(s) from which the to
that
the socially preferred program
ins the validity of the results comparison is to be made. Questions
assessing
goes
ahead. The existence of differ­
such as Is a chronic home care
- they encounter.
ent
viewpoints
was highlighted y
When assessing the validity of program worth it? and Will a com­ Weisbrod and colleagues1 in their
evidence, whether in terms of effica­ munity hypertension screening pro­ study of community-oriented and
cy, effectiveness, availability or effi­ gram do any good? beg the further hospital-based treatments for pa­
ciency, we usually proceed by close- questions To whom? and Compared tients with mental illness.
with what? Similarly, questions such
*Pans I through V of this scries were pub­ as How much does it.cost to run our
lished in consecutive issues of the Journal
2. Was a comprehensive description
starting with the Jan. 1, 19S1 issue, and part intensive care unit? and What are of the competing alternatives given.
VI was published in the Feb. 15, 19S4 issue. the costs and outcomes of adolescent
Part VII(A) appeared in the June 1, 1984 counselling by social workers? are
A clear and specific statement^
issue.
not efficiency questions because
the primary objective of each alter­
Reprint requests to: Professor G.L. Stoddart they fail to specify the alternatives
or Professor M.F. Drummond, McMaster for comparison. (See part A for a native program, treatment or seiyi
University Health Sciences Centre. Rin. review on the nature of economic is critical in selecting the typ2C13. 1200 Main St. W, Hamilton, Ont.
evaluation — cost-effectivcnes ,
evaluation.) This is not to say that
LSN 3Z5

Department of clinical epidemi­

1 .

I-

’.V

h-

;r

1542

CAN MED ASSOC J. VOL. ISO. JUNE 15. 1984

$

Table I—Detailed readers’ guides for efficiency studies
I. Was a well defined question posed in answerable form’
3) pDrogram(s)?Udy
b°'h
!‘nd effeC'S °f 'he ServiCe<S) or

!

JT

i

i

I
*

i

]
?

i

i
J
7

T
‘ r

(b) Did the study inrohe a comparison of alternatives’
(c) Was a viewpoint for the analysis stated or was the study placed in a
particular decision-making context?
"
2 vnnVl|C7''aeaenSJ'e desc1riP,ion of '^competing alternatives given (i.e can
you tell who did what to whom where and how often)?
(a) Were any important alternatives omitted?
Was rhpS (ShOald) S “do‘n1othinS” a'fernative(have been) considered?
3.
Was this done^h6
6 Pr°grams’ eff«tiveness had been established?
as this done through a randomized, controlled clinical trial9 If nm
Strong was the evidence of effectiveness?
’ t,°W
4. idVenrtiifiled?,nPOr‘an< a"d "k'3"' C0S‘S and cons«I^nces for each alternative

(b) ’ DMSithenranBe "ide,enough for ,he "search question at hand?
socle t oatients ra ,bana viewpoints
th«^ of the community or
society, patients and third-partv payers)7
(c) Were capital costs as well as operating costs included?
units /° . and c°nsei,uences measured accuratelv in appropriate physical

(b)

Were 1^ ‘hey

in the^ub^uemTnalvsisV0’

'hiS

j
i

cost-benefit or cost-utility’— i0 be
undertaken. A full description of the
alternatives is essenthl for three
other reasons: (a) readers must be
able to judge the applicabilaty of the
programs to their own settings, (b)
readers should be able to assess
whether any casts or consequences
may have been omitted in tihe analy­
sis, and (c) readers my wish to
replicate the program procedures
being described. Therefore^ readers
should be provided with information
that allows ideniificatioz of {both the
costs (Who does whai to whom
where and how often?) and the
consequences (\\ ith what results?).

3. Was there evidence that rlhe
. ; programs ’ effectiveness cad been
established?
l

.



We are not imerested in tie
tic effi
effi-­
made
anJ’ SPeC‘al circun,SIances
joint use of resources! that
«o
cient
provision
oi
ineffective
services
Measurement difficult? Were these circumstances handled “pi
i (i.e., those that have been sfiiown to
6. \\ ere costs and consequences valued credibly7
1 do no more goed than harm by'
(a) Were the sources of all values (e.g., market values, patient or client
' themselves or compared w5ih no
ige«Xd: mXS3'6"’ VieWS and hea,‘h Care Professionals’
j treatment). In fact, we are non interested in the provision of sntrh ser(b) Whp1' Harker raiues used for changes involving resources gained or used’
(cl
i vices under any conditiors, efficient
not efiX a .
abSenI ,e g-’ "he" 'Olun,eers -re used) or did
: or otherwise. If something is not
were adiustn,
IT e’S”
SpaCe ’’aS dona,ed at a redueed rate)
ji worth doing it’s not worth, doing
Jre adjustments made to approximate market values?
■ well. Therefore, if the economic
(i eXi the a8"0" Of COnSequences appropriate for the question posed
evaluation assumes effectiveness,
en J h r. appropriate (fpe- or types, of analysis - cost-effectiveness
_
cost-benefit or cost-utility — selected)?
■■ecuveness,
some indication should be srx’cn of
'' ,",ere "s,s and consequences adjusted for differential timing’
i the prior validation of eiTectivcness.
IheiXXt :i:snrUenCeS ,hat 0CCUrred in the fU,Ure -d-ounted- to
It is also possible that the efficiency
evaluation may have been conducted
(bl was any justification given for the discount rate used?
■ simultaneously with the evahiation
perfornied?nCre,nental
,VSiS °f C°S‘S a"d ^sequences of alternatives
of efficacy or effectiveness. This is
Were the additional tincremental) costs generated by the use of one
the case in many randomized: Trials
alternative r—
“ ' eramd "
COmpared Hith ,he additional effects, benefits or
of
therapies that also include i com­
utilities generated?
parison
of the costs of the experi­
9. \Vasa<sensitivity
’ ' analysis performed?
.....
mental
program and49 a; ccmirol
(a) Was justification provided for
the
sensitivity analysis’
°f 'alUCS
parame,ersl
iused’ in
'

which may be a placebo ar & cur­
(b) assumed
Were therange)?
study results SenSlti’e
rently existing program. Note, .how­
ChangeS in ",e ''a,ues (wi,hin ,he
assumed range)?
ever,
that efficiency eralnrrnons
10. Did the presentation and discussion
;
alone
cannot
establish effectiveness,
of the results of the study include all
issues of concern to users?
i There are, after ail efficient nneth(a) Were the conclusions of the analysis based c-------on some overall index or ratio
.: ,ods of worsening the quality on lifet
of costs to consequences (e.g., cost-effectheness ratio)? '[f
so, was the
as well as improving it. (If you. want
.ndex interpreted intelligently or in a mechanistic fashion?

to
know more about .the methads of
'
"d'"
i determining whether a thernnv does
. more good than harm yoz’sfoould
»•«"»
, read part V of this series/)
(d)
Did the study allude to. or take account of. other important factors in
conswuT °r d<!C'Tn UndCr considera,ion ,e-S- distribution of costs and
consequences or relevant ethical issues)?
(e> a^oodnV'i'h/ d.iSC?S ‘T5 °f imPl'™n'a<io"- such as the feasibility of
oXaim a J" eTd Pr^ran1- gi'Cn eXiS'in’
31 - “‘her
wo^hwhik’progra'X?
C<,uid
for

s3

1
-—

- v-

\<SOC

'

j 4. Were all important and -eievnnt
! costs and consequerzas for
! alternative identifier?

i
... I

Even though it rr.iy not ~e zossihie or i'-'-----nccessary
:c measure’ -nd
value all of
.......thear.d cc’r.se-

F

\ oi._ isn. jt

15. ]c<_

1543

’G

quences of the alternatives being
compared, the important and rele­
vant ones should be identified. The
information in the viewpoint state­
ment and program description
should allow you to judge what
specific costs and consequences or
outcomes it is appropriate to include
in the analysis.
An overview of the types of costs
and consequences that may be rele­
vant to economic evaluation of
health services and programs is pro­
vided in Fig. 1. Three categories of
costs are shown. Since the costs of a
health care service or program are
best thought of in terms of the
resources used, category I contains
the costs of organizing and operat­
ing the program. The identification
of these costs often amounts to list­
ing the “ingredients" of the program
— both variable costs (such as those
of health care professionals’ time or
supplies) and fixed or overhead costs
(such as capital costs, rent and the
costs of light and heat). These costs
are often referred to by economists
as “direct costs’’.*
Category II contains costs that
are borne by patients and their fam-

iii
ft
iH
I 0

fl
PI'
& fh

R
M


I'

IJ


•Heallh care administrators sometimes re­
serve the term direct costs for variable costs
only and may refer to overhead costs as
;?ln economic evaluations, howindirect costs. ... -------- -—--;
ever, economists use the term indirect costs to
denote a separate and distinct type of cost, as
we will explain later. Users of evaluations
should be aware of this potential source of
confusion.

i

i4?

I
Itr'
■zc

ilies. These include any out-of-pock- gram mayj result in more costly
as the value of production processes, thereby raiset expenses as well

they
contribute
to the ing the price of, say, cars. In princiany resources
treatment process. Patients and their pie such instances should be identi­
families sometimes lose time from fied, though in practice they may
work while seeking treatment or rarely be significant. (Few economic
analyses of alternative health proparticipating in a health tprogram.
Such ‘‘production losses” are also a grams take them into account.)

Fig. 1 also shows three categories
cost of the health care service
or■
are------often referred to of consequences
of health care serprogram iand
-------by economists as “indirect costs”’’., vices and programs. Category I conHowever; care must be taken when tains therapeutic outcomes or effects
including this cost in an analysis, of the alternatives. These effects will
since its inclusion implies that the usually-include changes in the physicost was incurred as a result of cal, social or emotional functioning
participation in treatment and there- of individuals. In principle such
fore that the individual’s condition changes can be measured objectively
alone would not have prevented pro­ and refer only to an individual’s
ductive activity.! Finally, the anxi­ ability to function and not to the
r_
ety, and perhaps pain,
associated significance, preference or value atwith treatment constitutes a psychic tached to this ability by the individ­
cost frequently encountered by pa­ ual or by others.
The therapeutic effects of a ser­
tients and their families.
vice
or program give rise to two
While these two categories cover
other
important categories of consemost of the costs relevant to eco­
nomic evaluations of health care quences. First, the effects may result
services, a third category also war­ in changes in the use of resources in
rants mention. It may be that the the future (category II). Within the
operation of a health care service or health care sector, less use of re­
program changes the use of re­ sources may be required for treatsources in the broader economy out- ment of the condition and its sequeside the health sector. For example, lae than would otherwise have been
an occupational health or safety pro- the case. For example, an effective
hypertension screening program
+The complexity of the relation between lost averts the future cost of caring for
work time and the value of forgone output stroke victims. The saving in the use
places it beyond the scope and purpose of this
hea|th care resources attributable
article. However, for a discussion of its impli­
the screening« .program
is usually
cations for categories 11 and 111 you can read to t
referred
to
by
economists
as the
Stoddart’s article.'

I.



Bi
II.

Costs borne by patients and their families

Out-of-pocket expenses
Patient and family inputs into


I

I.

Organizing and operating costs within the
health care sector (e.g., health care
professionals' time, supplies, equipment,
power and capital costs)

treatment
Time lost from
work
Psychic costs

Changes in physical, social or emotional

functioning (effects)
II.

Direct
costs

III. Changes in the
quality of life
(benefits)
of patients and
For organizing and oper­
their families
ating services within
Changes in resource use

the health care sector

For the original
condition

Indirect costs

III. Costs borne externally to the health care
sector, patients and their families

i

I
j

I
t’’

r;I

9

(utility)

Direct benefits

For unrelated
conditions
Relating to activities

of patients and their
families
Savings in expenditure
or leisure time
Savings in lost work
time

Direct benefits
Indirect benefits

Fi-. 1-Tvpvs of costs and consequences relevant to economic evaluation of health care senices and programs (adapted,
permisbiun. from reference 3

I

Consequences

Costs

T ■

i

:I

I

n
v'
I.

!

direct benefit of the screening pro­ III). The change in quality of life
multaneously in economic analyses,,
gram. Notice, however, that if we produced by the therapeutic effects
you should view each as a separate
adopt the viewpoint of a health care is • distinguished from the effects
phase of the analysis. Once the
system the direct benefits are some­ themselves by the significance or
important and relevant costs and
times negative owing to the in­ value that patients and their families
consequences have been identified,
creased use of services for the treat­ attach to the effects. It is, of course,
they must be measured in appropri­
ment of conditions (e.g., arthritis) possible — and, in fact, likely —
ate physical or natural units. For
that may develop in patients during that different individuals place a
their newly prolonged lives. The different importance on the same example, measurement of the oper­
ating costs of a particular screening
therapeutic effects of a health care level of physical, social or emotional
program
may yield a partial list of
service or program may also affect functioning. (This was demonstrated
“ingredients” as follows: 500 physi­
the use of resources by patients and by the example of the impact of a
their families. Of particular interest broken arm on the sign painter and cal examinations performed by phy­
sicians, 10 weeks of salaried nursing
is the possibility that patients and the translator in part A.)
time, 10 weeks’ use of a
their families may gain working
With respect to both the costs and 100-m2 clinic, 20 hours of medical
time as a result of their participation the consequences we have described
librarian research time from an ad­
in treatment. These production gains it may be unrealistic to expect all
joining hospital and so forth. Simiare usually referred to by econo­ relevant items to be measured and larly,
,, costs borne by patients may be
mists as “indirect benefits”.
valued in an <economic analysis measured by the amount of medica­
The inclusion of indirect benefits owing to the small impact of some
tion purchased, by the number of
in economic evaluations is a source relative to the effort required to
times travel was required for treatof some controversy among analysts. measure or value them accurately; rnent or by the time lost from work
It is sometimes argued (rather nar­ however, it is helpful to users to while the'patient was being treat<
rowly it seems) that health care have as many of them as possible ed.
evaluation should confine itself to identified. It is particularly impor­
Situations in which resources are
i
changes in the use of resources in tant that the outcomes of interest be
used jointly by one or more pro­
the health care sector
<
only, rather identified clearly enough for you to grams present a particular challenge
than in the entire
re (economy. More judge the appropriateness of the to accurate measurement. How
serious is the assertion that changes type(s) of economic evaluation cho­
much use of a resource should be
in the output of individuals or sen. That is, it should be apparent
allocated to each program and on
groups are simply not the grounds (a) whether a single outcome is of
i
what basis? A common example is
upon which we usually make deci­ primary interest as opposed to a set found in every hospital, where nu­
sions about allocation of health care of outcomes that are each of some,
j
merous clinical sen-ices and proresources that will affect those indi­ it not equal, interest, (b) whether ^grams share overhead services (e.g.,
viduals or groups. Therefore, it is the outcomes are common to the two
power, cleaning and administration)
i
misleading to enter the value of such alternatives being compared, and (c)
that are provided centrally. In gen­
changes into a cost-benefit calcula­ to what degree each program is
eral, there is no nonarbitrary solu­
tion. A third criticism is that the successful in achieving each out­
tion to the measurement problem.
valuation of indirect benefits (usual
come of interest. Similarly, it is You should, however, satisfy your­
•’j—j0U?h increased earnings of important to know whether the con­ selves that “reasonable” criteria
individuals) makes a series of value sequence of primary interest is the
(e.g., number of square metres,
judgements and assumptions that therapeutic effect (which implies
number of employees or number of
may only be appropriate in a limited that a cost-effectiveness analysis cases) have been used in the distri­
number of cases. While it is not should be done if possible), the net bution of the common costs. You
possible to discuss and evaluate change
_____ ____
in the use of resources (cost- should definitely ascertain that such
these claims here, you should be benefit analysis)) or the quality of shared costs have, in fact, been
aware that the inclusion of indirect life of the patients and their families allocated to participating services or
benefits in a cost-benefit analysis (cost-utility analysis).
programs, as this is a common omis­
may not be straightforward.4-5*
sion in evaluations. Clinical service
The therapeutic effects of health 5. Were costs and consequences
directors often argue that small
care sen ices and programs also give measured accurately in appropriate
changes in the size of their programs
rise to another extremely important physical units?
.(up or down) do not affect the use of
category of consequences; namely,
central sen-ices. Sometimes it is
•j
changes in the quality of life of
While identification, measure­ even argued that overhead costs are
patients and their families (category ment and valuation often occur si- unaffected by the service itself.
'Those who criticize the inclusion of indirect services undoubtedly exist, by their very na­ However, though this argument may­
benetns. saving You can value a livelihood ture they are difficult to include in a cost­
be appealing from the viewpoint of a
but you can never value a life! appear to be benefit analysis, which expresses costs and particular program or service direc­
contusing indirect benefits with another type consequences in dollars. They presumably are
ol oenettt. Tats is the intangible value we. as taken into account, however, in cost-utility tor, the extension of this method to
individuals and as a society, place on life analyses, wherein program effects are trans­ each service in the hospital would
i'-sclf (regardless of earning potential) and on
lated into a measure of value based on imply that ail the services could be
ihc avoidance cf pain and suffering. Although rreferences rather than dollars, as discussed operated without light, heat, power
'ntanL’ible benefits and costs of health care in ujcs’.tan >'
and secretaries!

I

i

I

I

-

i

XX Mi'D \SSO<

\ ( >1 . : -f: JI \-

I M

ft

hi
SR

fe
B$

p i.
&

fe r
t?
b =

w4■

.

EK
;



ii.-'1- *

<;■

»

I

With respect to the measurement are usually valued in constant dol­ linking estimates of the value of
of consequences, if the outcomes of lars of a base year (usually the health care programs to the vagaries
interest have been clearly identified, present) to remove the effects of ■ of the market. They argue that
acceptance of existing wage rates
then the selection of appropriate inflation from the analysis.
coupled
with the inclusion of indi­
It
should
be
remembered
that
the
units of measurement for program
effects should be relatively straight­ object in valuing costs is to obtain rect benefits biases cost-benefit
forward. Effects might relate to an estimate of the worth of the studies against programs aimed at
mortality and be measured in years resources used by the program. This minority groups, housewives, the el­
of life gained or deaths averted, or . may necessitate adjustments to some derly, children and the unemployed.
they might relate to morbidity and apparent program costs, as in the Although it may be possible to ad­
be measured in reductions in the case of volunteer labour or subsi­ just some of the estimates to ac­
number of days of disability or im­ dized services received by one pro­ knowledge this problem (e.g., by
provements in health status accord­ gram instead of another. In addition, imputing a value to housewives’ ser­
ing to some index of physical, social valuation of the cost of a day of vices based on wages for similar
or emotional function. They may be institutional care for a specific con­ work) the indirect benefit issue re­
even more specific, depending on the dition is particularly troublesome, in mains controversial.
In valuation of preferences or util­
alternatives under consideration. that the use of an average cost per
Thus, “percentage ’increase in day (the widely quoted “per diem”), ities we are basically attempting to
weight-bearing ability” may be an calculated on the basis of the insti­ ascertain how much better, ail
appropriate natural measurement tution’s annual caseload, almost cer­ things considered, the quality of life
unit for an evaluation of a physio­ tainly overestimates or underesti­ is in one health situation or state
therapy program, and “the number mates (sometimes by a large (e.g., when dialysis is performed at
of correctly diagnosed cases” may amount) the actual cost for any home with help from a spouse or a
be appropriate for a comparison of specific condition. You should thus friend) compared with another (e.g.,
venography with leg scanning in the approach per-diem values with ex­ when dialysis is performed in hospi­
tal). Several techniques are avail­
treme caution.*
diagnosis of deep-vein thrombosis.
Valuation of direct benefits pro­ able for making the comparison;
Changes in the use of resources
resulting from the effects will be ceeds in the same fashion as that of each will produce a utility value
measured in physical units similar to costs and is subject to the same (mentioned in part A) with' which
those used for costs. Thus, the caveats since the benefits are usually one can increase or decrease the
changes in use resulting from any the expected future costs that are value of time spent in health situa­
particular program will likely be saved. Valuation of production gains tions resulting from the alternative
recorded in numbers of procedures or indirect benefits (i.e., changes in in question relative to a baseline.
or in amounts of time, space or the value of output of individuals or Usually the results of utility analy­
equipment. Changes in the use of groups who receive the health care ses will be expressed in “healthy
resources by patients will continue program or service) usually employs days” or “quality-adjusted lifeto be measured, for example, in the wage rates for individuals or years” resulting from the programs
amounts of medication purchased or groups to value the increased work­ being evaluated.
Two broad approaches to utility
ing time available. It is here that
number of trips taken for treatment.
While the nature of changes in critics of cost-benefit analyses point analysis can be found in the litera­
the quality of life may be described out the inequity associated with ture. The first approach, outlined by
Torrance/ emphasizes the develop
in an economic evaluation, measure­
’In principle, and with great effort in prac­
ment in objective, physical or natu­ tice. it is possible to identify, measure and ment of measurement methods and
ral units is difficult, although the value each resource (e.g.. drugs, nursing time, empirical testing in different popula­
consequence of some surgical inter­ light or food) used in treating a specific tions. The other approach, outlined
ventions may be quantified in “num­ patient or group of patients. While this yields by Weinstein.9 emphasizes the esti­
a relatively accurate cost estimate, the de­
ber of complications”. However, the tailed monitoring and data collection are mation of utility values by a quicx
adjustment of effects for quality of usually prohibitnely expensive. The other (and inexpensive) consensus-form­
ing exercise and then the perfor­
life is usually a matter of valuation.
broad costing strategy is to star, with the
mance of an extensive sensitivity
institution’s total costs for a particular pericd
and
improve
upon
the
method
oi
simplv
analysis
on the chosen values to see
6. H ere costs and consequences
dividing by the total number of patient days
whether
the results change T tiv.
valued credibly?
to produce an average cost per day. Quite
utility
estimates
are varied. We sc.
sophisticated methods of cost allocation to
The sources and methods of valu­ individual hospital departments or- wards a role for .both approaches. The
ation of costs, benefits and utilities have been developed.'’ /\n intermediate meth­ latter is useful in persuading deci­
involves acceptance of the components ot
sion-makers to think about problems
should be clearly stated in an eco­ od
the general per diem relating to “hotel” costs
nomic evaluation. Costs are usually (since these are relatively invariant for all in allocating resources and is, lfvalued in units of local currency on patients) combined with more precise calcula­ fact, relatively quick and inexpem
the basis of prevailing “prices” of tion of the costs of medical treatment of the sive. The measurement approach ••
personnel, commodities, services and individual patients.7 Of course, the effort useful in highlighting the fact t-im
devoted to estimating accurate per diems
so forth and can often be taken depends on their .•vernll importance in the different individuals (doctors. P°*
directly from program budgets. .All s’.udv, however, unthinking use per diems icymakers. patients and the
public is taxpayers) may have difcurrent and future progr!“: ccsts • r .ivcrm.’v co:.;?, .'hoiild h • '.voided.
'

' O!

L

i

J

I

ferent values, and it is clearly cru­ dcnccd by the existence of interest comes for the two alternatives on
cial ir. situations n which the result rates (as well as by the popular
should compare the incrementa
is sensitive to the utility values as­ wisdom about “a bird in the hand”).
costs with \hz.incremental outcomes,
signed.. (An examr'.e of such a situa­ Moreover, since time preference is since this will tell us how much wc
tion arose in the study by Stason not exclusively a financial concept,
are paying for each extra correc
and Weinstein10 or. the economics of outcomes should also be discounted
diagnosis in adding the extra diag­
hypertension therapy. The result of in cost-effectiveness and cost-utility
nostic test. In this case the relevant
their study was sensitive to whether studies."14
figure is S4781 per correct diagno­
it was assumed that the side effects
sis, not the average figure for the
of antihypertensive drugs constitut­ 8. Was an incremental analysis of
second program, which is $3003 per
ed a lx ora 2% reduction in health costs and consequences of
correct
diagnosis. It may be decided
status.)
alternatives performed?
that $4781 is still a price worth
Since the measurement of prefer­
paying; however, it is important to
ences in health is relatively new,
For a meaningful comparison it is be clear on the principle since, as we
there are naturally many unresolved necessary to examine the additional
pointed out in part A, in screening
issues in cost-utilim analyses. You costs imposed by the use of one
for cancer of the colon there was a
will probably wan: to know, at the service or program over another,
big difference between the average
least, ^hose preferences — pa
pa-­ compared with the additional ef­ cost per case detected of a protocol
tients’, providers’, taxpayers’ or bu­ fects, benefits or utilities it delivers.
of six sequential tests and the incre­
reaucrats'? — were used to con­ This “incremental” approach to the
mental cost of performing a sixth
struct the utility values. If pa­ analysis of costs and consequences
test, having already done five.15
tients’ preferences vere not used you can be illustrated by one of the
Obviously similar analyses could
may warn: lurther assurance that the examples cited in part A of this
be performed if the consequences
persons whose preferences were used article; namely, the strategies for
were effects in natural units (e.g.,
clearly understood me characterist­ diagnosing deep-vein thrombosis.7
years of life) or in utilities (e.g.,
ics of the health state, either Table II shows the costs and out­
quality-adjusted life-years).
through personal experience or comes (in terms of correct diagno­
through a cescripticr. of the state ses*) generated by two alternative 9. IVas a sensitivity analysis
presentee to them.
strategies: impedance plethysmogra- performed?
phy alone and impedance plethys­
7. Were costs and cc'sequences
mography plus outpatient venogra­
Every evaluation will contain
adjusted fo- differential timing?
phy if the former gives negative some degree of uncertainty, impreciresults.7 Impedance plethysmogra- sion or methodologic controversy.
Since 2 compariscz of programs phy is a noninvasive strategy, where­
What if the compliance 'rate for
or services must be made at one as venography, the diagnostic “gold
influenza vaccination was 10% high­
point in lime (usuaE> the present), standard” for deep-vein thrombosis,
er than that considered for the analthe timirg of progrim costs and can cause ]pain and other side ef- ysis? What if the hospital per diem
consequences that de not occur en­ fects. Although one could compare
still understated the true resource
tirely in ihe present must be taken the simple ratios of costs to outcost
of a treatment program by
into account. Different programs
*The study by Hull and associates' is .an $100? What if a discount rate of 6%
may have different time profiles of example of a cost-effectiveness analysis in
instead of 2% had been used? What
costs or consequences. For example, which the outcomes are not therapeutic ef­
if indirect costs and benefits had
the primary benefits c: an influenza fects- but. rather, intermediate diagnostic out­
immunization program are immedi­ comes with direct implications for therapeutic been excluded from the analysis?
You will often ask these and similar
ate. whereas those of a hypertension effects in that the failure to diagnose deep­
vein thrombosis leads directly to increased questions; therefore, careful analysts
screening program wil occur in the morbidity and mortality.
will identify critical methodologic
future. The time profile of costs and

consequences may als: differ within
a single program; the costs of the
hypertension screening program
would be incurred in the present.
Future doliar costs, and benefits are
therefore reduced or “discounted” to
reflect- the fact that wv
dedars
or
----- to spent U1
saved iin. the
’ future should not weigh
as heavily in program decisions as
dollars speui
sperm or savea-Ltaay.
saved-ttday. This
ir
uujiars
this is
primarily cue to the existence of
‘‘time preference” — that is, we, as
individuals and as a scciety, prefer
to have dollars or resources now as
opposed to Later because we can use
them in the interim. This is cvi-

: Table II—Economic evaluation of alternative diagnostic strategies for 516
; patients with clinicallv suspected deep-vein thrombosis7
I --------------- ----------------------------------------------------------- --------------

(SUS)

Outcome
(no. of
correct
diagnoses)

Ratio of cost to ‘
outcome
ji
(cost (SUSI per
correct diagnosis)

321 488

142

2264

603 552

201

3003

282 064

59

4781

Cost

i Program

| _____________

I IPG alone
i IPG and outpatient
j venography if
I results of IPG
are negative
i' Increment of second program
over first program

*IPG = impedance plethysmography.

I

rI
S?'1
■'i • J »

V

r I'
i

i
F

Ifh:
V

ft.
1

Ih
it
<-

■■

N

i

assumptions or areas of uncertainty.
Furthermore, they will often at­
tempt to rework the analysis (quali­
tatively at least, if not quantitative­
ly) with different assumptions or
estimates to test the sensitivity of
the results and the conclusions to
such changes. If large variations in
the assumptions or estimates under­
lying an analysis do not produce
significant alterations in the results,
then one would tend to have more
confidence in the original results. If
the converse occurs, then more ef­
fort is required to reduce the uncer­
tainty or improve the accuracy of
the critical variables*. In either case
a sensitivity analysis is an important
element of a sound economic evalua­
tion.

10. Did the presentation and
discussion of the results of the
study include all issues of concern
to users?
It will be clear from the preceding
discussion that economic analysts
have to make many methodologic
judgements when undertaking a
study. Faced with users who may be
mainly interested in the “bottom
line” (e.g., should they buy a com­
puterized tomography scanner?)
how should they present their re­
sults?
Decision indices, such as costeffectiveness and cost-benefit ratios,
arc useful in summarizing the re­
sults of a study. However, they
should be used with care, for the
reader, when interpreting them, may
not be completely clear on what has
gone into their construction. Some
analysts give a range of results. For
example, in an economic evaluation
of neonatal intensive care for verylow-birth-weight infants Boyle and
coworkers6 compared the results for
infants weighing less than 1000 g
and from 1000 to 1500 g in terms of
costs up to the time of hospital
discharge, costs and consequences to
age 15 years and costs and conse­
quences for a lifetime (Table III).
They left it to the reader to decide
which index (or indices) to use in
judging neonatal intensive care,
since the different measures incor­
porate different value judgements
and amounts of precision. (Far ex­
ample. the index of ret economic
benefit includes production gams nr

losses, and the index of cost per day-care surgery the full econom
quality-adjusted life-year incorpor­ payoff may not have been obtaine
ates 1the preferences for health states in at least one Canadian hospital.
of a sample of the local population.)
This leads to another general Limitations of economic evaluation
point; namely, it is important for techniques
analysts to be as explicit as possible
The main purpose of this pair c
about the various judgements they
have made in carrying out the study, articles is to make you more awar t.
. shouldi more
of the methodologic
of the methodologic
judgementsjudgements
ir. ?ir.
A good study
leave you
(ralher than'less) aware of the vari- volved
volved in
in an
an economic
economic evaluation
evaluation ii |
ous technical and value judgements the health care field. In Table I w |
necessary to arrive at decisions on have consolidated the points made i r!
the allocation of resources in health the text into a checklist of question
care
you should ask when critically as
Finally, a good study should begin sessing the results of an economic V
to help you interpret the results in (evaluation. Some of the questions f
the context of your own situation, signal limitations of economic evalu- jn-L!
— :rif the analysts n
This-----canu-bejdone
are at
: jon techniques. For example, ecoexplicit about the viewpoint for the nomic evaluation techniques assume.





’ ’’ L
7---------------------------- f£* :
analysis and indicate how particular rather than establish, program effec
costs and benefits might vary by tiveness. There are several other
location. For example, the costs of limitations of which you should be
instituting day-care surgery may aware.
Of primary concern from a policy
vary depending on whether a pur­
pose-built day-care unit already ex­ viewpoint is that economic evalua- v
ists or whether wards have to be tions do not usually incorporate into gj
converted. Similarly, the benefits of the analysis the importance of the fc
J------J
conseday-care surgery may vary dependdistribution
of costs and
ing on whether there is a shortage of quertces. Yet in some cases the idenbeds and on whether the beds will be tity of the recipient group (e.g.. the
_______ j ______
x
poor, the elderly or working moth- jleft empty.10 Obviously
it is impossible for the analysts to anticipate ers) may be an important factor in j
J_______________ ■
.
. .
,

i
C a
“j-.
every possibility In
every location,
assessing
the _social
desirability
of
but one limitation of economic eval- service
* or program. TIndeed,
" “ it may
uation techniques is that they as­ be the motivation for the program.
sume that freed resources will be put Although it is sometimes suggested
to other beneficial uses. Evans and that differential weights be attached g
Robinson17 argue that in the case of to the value of outcomes to special

IJ

I

I

I

II

---------- —--------------- --------------------------------- --—-- ------ ;
, i
; Table III—Measures of economic evaluation of neonatal intensive care, according
> to birth-weight class (5% discount rate)*

Birth-weight class: cost (S)t
Period
; To hospital discharge^

Cost/additional survivor at hospital
discharge
To age 15 years (projected)
Cost/life-year gained
Cost/QALY§ gained
I To death (projected)
Cost/life-year gained
Cost/QALY gained
Net economic benefit (lossi/live birth
Net economic cost/life-year gained
Net economic cost/QALY gained

1000-1499g

500-999g

59 500

102 500

6 100
7 700

12 200
40 100

2 900
3 200
(2 600)
900
1 000

9 300
22 400
(16 100)
7 300
17 500

• ’Adapted, with permission, from reference 6.
in 1978 ■
I ‘’’In 1978 Canadian dollars: multiply by 0.877 to calculate equivalent cost
i LS dollars.
1 ;A11 costs and effects occurred in year 1.
?;QALY -= quality-adjusted life-year.

I

I

i

1

i>_ £r
1.
I

T

5

i
f
1

I%

1

J
1

J

I
i

I

’ recipient groups, this is not usually
done in an economic evaluation.
Rather, an “equitable” distribution
of costs and consequences across
socioeconomic or other defined
groups is viewed as a competing
dimension, in addition to efficient
deployment of resources, upon
which decisions are made.
As we have already pointed out,
economic evaluation techniques as­
sume that resources that are freed
up or saved by preferred programs
will not, in fact, be wasted but will
be used in alternative worthwhile
programs. This assumption warrants
careful scrutiny, for if the freed
resources are consumed by ineffec­
tive or unevaluated programs, then
not only is there no saving, but the
overall costs of the health care sys­
tem will actually increase without
any assurance- of additional im­
provements in the health status of
the population.
Finally, an evaluation of any sort
is costly. If we bear in mind that
even a cost-benefit analysis should
be subject to a cost-benefit analysis,
it seems reasonable to suggest that
economic evaluation techniques will
prove most useful when program
objectives require clarification, when
the competing alternatives are sig­
nificantly different in nature and
when large resource commitments
are being considered.

--- -------- '

■'

Conclusion

In this pair of articles we have

4

tried to help potential users of eco­
nomic evaluations to understand
such studies and assess their validity
and usefulness. Our intent has not
been to create hypercritical users
who will be satisfied only by super­
lative studies. As we have empha­
sized, it is unlikely that^every study
will satisfy all the methoclologic crit­
eria we have discussed. However,
the readers’ guides should help you
to quickly identify the strengths and
weaknesses of any study.
We thank our colleagues and students
for their suggestions and criticisms of
earlier versions of this paper.

References

1983;308: 1330-1337
7. Hull R, Hirsch J, Sackett DL, Stoddart G: Cost effectiveness of clinical
diagnosis, venography and noninvasive
testing in patients with symptomatic
deep-vein thrombosis. N Engl J Med
1981;304:1561-1567
8. Torrance GW: Preferences for health
states: a review of measurement methods.
In Sinclair JC (ed): Clinical and Eco­
nomic Evaluation of Perinatal Pro­
grams. Proceedings of Mead Johnson
Symposium on Perinatal and Develop­
mental Medicine, Number 20. Vail, Col­
orado. June 6-10. 1982, Mead Johnson,
Evansville, Ind. 1982: 37-45
9. Weinstein MC: Economic assessments
of medical practices and technologies.
Med Decis Making 1981; 1: 309-330
10. Stason WB. Weinstein MC: Allocation
of resources to manage hypertension. N
Engl J Med 1977; 296“ 732-739

11. Weinstein MC. Stason WB: Founda­
tions of cost-effectiveness analysis for
1. Weisbrod BA. Test MA. Stein LI:.
health and medical practices. Ibid: 716—
Alternative to mental hospital treatment.
721
II. Economic cost-benefit analysis. Arch
12. Drummond MF: Principles of Economic
Gen Psychiatry 1980: 37: 400-405
Appraisal in Health Care, Oxford U Pr,
2. Department of clinical epidemiology and
Oxford. 1980
biostatistics. McMaster University
13. Idem: Studies in Economic Appraisal in
Health Sciences Centre: How to read
Health Care, Oxford U Pr, Oxford. 1981
clinical journals: V. To distinguish useful
from useless or even harmful therapy.
14. Warner KE. Luce BR: Cost-benefit
Can Med Assoc J 1981: 124: 1156-1162
and Cost-effectiveness Analysis in
Health Care: Principles, Practice and
3. Stoddart GL: Economic evaluation
Potential, Health Admin Pr, Ann Arbor,
methods and health policy. Eval Health
Mich. 1982
Prof\9Zl:S\2>92-4\A
15. Neuhauser D. Lewicki AM: What do
4. Drummond MF: Welfare economics and
wc gain from the sixth stool guaiac? .V
cost-benefit analysis in health care. Scott
Engl J Med 1975: 293: 226-228
J Polit Econ 1981; 28: 125-145
5. Williams AH: Welfare economics and
16. Russell IT. Devlin HB. Fell M.
health status measurement. In van der
Glass NJ, Newell DJ: Day-case sur­
Gaag J. Perlman M (eds): Health.
gery for hernias and haemorrhoids: a
Economics and Health Economics, El­
clinical, social, and economic evaluation.
Lancet 1977; I: 844-847
sevier. Amsterdam. 1981: 271-281
6. Boyle MH. Torrance GW. Sinclair
17. Evans RG. Robinson GC: Surgical day
JC. Horwood SP: Economic evaluation
care: measurements of the economic pay­
of neonatal intensive care of very-lowoff. Can Med Assoc J 1980; 123: 873birth-weight infants. .V Engl J Med
880

Index Medicus

I
1

J

Index Medicus is the basic tool for medical
literature searches. It is the National Library of
Medicine's major index to nearly 2 700 journal titles. It is published
monthly and is cumulated yearly. Review articles are included in the
subject section. It is subdivided into separate author and subject
sections. Medical literature, beginning in 1879 with the old Index
Medicus, is available through a series of indexes to the less current
literature. The original Index Medicus was edited for many years by
John Shaw Billings and Robert Fletcher. In 1960, the new, ongoing
Index Medicus was begun.

&

CLINICAL ECONOMICS

MODULE 2

THE COST OF ILLNESS
After completing this module you should understand:

a•

the concept of the cost of ilIness,

b.

how the cost of a particular disease or condition can be
calculated,

c.

how the approach can be used for policy purposes.

*

I

1
THE COST OF 1LI.NESS

Introduction
You will
be familiar with the concept of the burden of
i1Iness
from
the epi demi ology modules.
It uses indicators of
It.
prevalence,
inci denee,
mortali ty -- losses and
morbidity losses to
calculate the burden
imposed
on
i mposed
society by particular diseases.
The reference
by the Ghana
Health
Assessment Project Team will
refresh
your
memory.
Cl ini cal
economics
builds on these
foundations
to provide an extra dimensi on to the burden of illness,
k n o wn as
the c o st of______
i1Iness.
Two approaches can be used,, one
based on prevalence estimates, the other
-• * on estimates of incidence.
Prevalence Based Estimates of the Cost of Illness.
Prevalence is def ined
as
the number
of cases of a
particular disease or condi ti on
which exist
in
a given time
interval, usually a year.
TheJeconomTc co s r~o f t h ejsjT) cases i s :
(J)the value to societyv of aTT resources,consumed during
the period as a result of diagnosi
- s, treatment and
repercussions of
the disease or condition;

0th

PLUS
/^the value
of
lost production as a result of morbidity
during the period;
PLUS (5) the
value of
1 ost production
product ion as a result of premature
lost
mortality which occurs during the period.
Hence,
the prevalence approach to the cost of illness is
concerned
with
costs: accruing in
the specified
period and with costs
attributable
to premature mortality
in that period.
It includes
the
costsr incurred by both
fireexisting cases and new qases
identified 4during
<
the period.
The reference, by Coopej?_._A_Ri ce uses
t 111_m e t h o d .

Incidence Based Estimates,
Inc idence
iiss defined as
the number of new cases of a
disease or condition
that have
their onset
during a specified
period.
In
the
incidence based approach to the cost of i1Iness,
only the costs
incurred by new cases iidentified during the period
are considered.
A s i n the prevalence approach, direct health
--- ---- care
costs
and the
value of
lost production must
be
must
be
calculated,
However,
in this instance the expected future
associated w7ith
the new cases detected in the specified period costs
must be added to the
costs
iincurred during the period.
The reference by Hartunian et. al
uses this approach.

..'•c"—r
t
Most cost of illness studies divide costs into direct,
indi reel ano intangible costs.
Direct costs are the costs that the
heal th sector and the patient (or family) must bear because of the/?i1Iness.
Co^ts_to.. the health sector include the cost jof diagnosis,
treatment,
rehabi1i tat ion
and prevention of the disease and its
side effects,
However, other costs incurred by the patient, such
as
travel, special food, and home nursing,

should also be included,
Direct Costs

2
though they are often omitted due to problems with quantification.
For example, both the
exist, as well.
Other problems exist
prevalence
approaches
require
an estimate of the
and
incidence
costs
1
that
will
be
incurred
by
new
cases
of the disease
health care
time
period.
Often
the
only
available
information
current
in the
cost
of
treating
an
average
patient,
which
is not
current
i s the
to
the
cost
of
treating
a
new
patient.
necessarily equal

The
of hospitalization as an example,
Take the costs
information
is
derived
by
dividing
the
most readily available
most
recent
financial
year
by
the
costs
in
the
hospital’s total
that.
year.
This
produces
an
occupied
in
-number of bed days
'.
Using
of
the
average
cost
per
bed
day
in
the
recent
past.^
estimate <
additional
figure to estimate
the cost of hospitalizing an u—
this
patient poses at least two problems.

First1y,
the cost of additional patients in the future
than
the average cost of hospitalizing existing
will be lower
the
new
patients take up beds
that have been
if
patients
in
the
past.
This
is
because
new
buildings and
underutilized
equipment are
not
required
to
cater
for
the
new
patients.
On the
are
other hand,
the
cost
of
hospitalizing
new
patients
is
likely
to be
hand,
higher than
than the average cost of hospitalizing patients in the past
if new .buildings and equipment are necessary.
for all
Secondly^
the cost of a day in hospital
Using an average cost for all diseases
diseases is. 'no4r^"the same,
the cost of some diseases and
will result
in underestimating
overestimating the cost of others.

problem is that some patients may
Another practical
illnesses
at the same time.
It is then
suffer from a number of
to
each
disease.
difficult to allocate costs
Indirect Costs
Indirect costs are the losses in economic output due to
morbidity,
and disability.
For example,
a 20 year old
disability.
death,
in
Ghana
(see
reading
1)
could
be
expected
to live for a
person
42.5
years.
If,
however,
the
person
dies
of
malaria
at 20,
however,
further
has
lost
42.5
years
of
life,
and
42.5
years
of
work
society
retirement
at
65)•
The
number
of
lost
years
of
work
65 ) .
(assuming
multiplied by the yearly wage rate is usually taken to be the value
of the output lost as a result of the person’s death.
Three complications arise.
Firstly, some of the people
who died may have been unemployed had they lived,
liveo.
They would not
have produced while unemployed
unemployed..
If the unemployment rate is 10%,
it is generally assumed that only 90% of the total number of years
lost through premature mortality are productive years.
It ~is this
total
that
is multiplied by the wage rate to obtain the cost of
premature mortality.

3
Secondly,
some
of the people affected by a disease may
not be paid money wages - housewives and unpaid family-farm workers
for-- example.
To use a zero wage rate for these people implies that
their loss imposes no economic burden on society, and many studies
therefore impute a value for this labour.

Thirdly,
the extent to which morbidity and mortality
. result in a loss of production can be questioned when there is a
lar^e pool
of unemployed or underemployed labour.
People leaving
the workforce can be replaced with relatively little dislocation to
For this reason, indirect costs are often interpreted
production.
than losses in actual
potential
rather
as
losses of productive
production.
indirect costs of morbidity and disability can be
The
similar
manner to the costs of premature mortality,
estimated in a
generally
available
from life
tables and labour force
are
Data
Question
3
illustrates
the
procedure.
statistics.

Intangible Costs
In general,
the costs of pain,
suf£e_ring and grief
associated__with a disease are regarded as intangible (immeasurable)
costs
in the cost of illness literature.
This may not be important
if the aim is to compare the cost of different diseases and it can
be assumed that intangible costs are either proportional to the sum
Some
of direct and indirect costs, or constant across diseases.
attempts have been made to put a money value on the costs of pain
and suffering, but they are not widely used yet.
Time

Both
the
incidence and prevalence approaches involve
For example,
costs which affect society in different time periods.
costs
of
premature
mortality
must
be
estimated
in both
the
approaches.
Assume that
the average healthy bricklayer produces
output valued at $20,000 a year, and would retire from work at the
age of 60.
One bricklayer, however, dies this year at 40.
Through
this death,
society
loses
20
years
of
production.
The
cost
of this
death,
lost production
is
$20,006”this year, and $20,000 for each of the
Premature mortality, therefore, has resulted
following
19 years,
Economists argue that
in a stream ot
of costs over a number of years.
to simply add up costs which accrue..in different
i t is not possible
__ _
A technique known as d i scounting must be used for
time periods.
This is the topic of Module 3.
such comparisons.

Conclusions
In this module, two approaches to measuring the cost of
You will not be able to calculate
identified.
i1]ness have been
full
cost of illness until you have mastered Module 3, but at
the
stage you
should understand
the concept, be able to think
this
is preferable and the policy implicati ons of
about which approach
These issues provide the focus of questions 1 and 2 .
each .

4

REFERENCES

Cooper B S & D P Rice, "The economic cost of illness revisited" US
Dept of Health,
Educat ion and Welfare, Social Security Bulletin,
Feb, 1976.
Ghana Health Assessment Project Team,
"A quantitative method of
assessing
the health impact of different diseases in less developed
countries", International Journal of Epidemiology,
10, 1, 1981.

Hartunian N.S.,
C.N.
Smart,
& M.S. Thompson, "The incidence and
economic costs
of cancer,
motor vehicle injuries, coronary heart
disease and stroke: a comparative analysis", American Jnl of Public
Health t 70, 1980, ppl249-60.
Von Allmen S.D. et al,
"Epidemic dengue fever in Puerto Rico 1977:
a cost analysis", Am Jnl Trop Med Hyg, 1979, 28(6):1040-1044.

QUESTIONS

1.
The
following table is taken from Cooper
calculations are based on the prevalence approach.

&

Total economic costs of disease in the United States in 1972
Indirect Percentage
Direct
Total
distribution
costs
costs
cost
of total
(millions
(millions
(millions
costs
of dollars) of dollars) of dollars)
Disease type

Infective and parasitic
$ 1,822
1.8
$ 3,234
$ 1,412
diseases
11,769
8.9
3,872
15,641
Neoplasms
Endocrine, nutritional and
3.3
2,281
3,436
5,717
metabolic diseases
Diseases of the blood and
384
0.5
491
875
blood-forming organs
6,797.
7.9
6,985
13,782
Mental disorders
Diseases of the nervous
4,756
6.1
5,947
10,703
system and sense organs
Diseases of the circulatory
21.4
26,511
10,919
J system
37,430
Diseases of the respiratory
9.0
9,833
15,764
5,931
system
Diseases of the digestive
9.7
5,831
11,100
16,931
system
Diseases of the genito­
3.6
1,873
4,471
6,344
urinary system
Diseases of pregnancy,
childbirth, and the
1.7
307
2,607
2,914
puerperium
Diseases of the skin and
1.2
515
2,040
1,525
subcutaneous tissue
Diseases of the musculo­
skeletal system and
5,277
5.1
3,636
8,913
connective tissue
994
0.8
381
Congenital anomalies
1,375
Accidents, poisonings,
12.4
16,528
5,121
and violence
21,649
4,227
6.6
7,398
11,625
Other
100.0
$174,934
$99,703
$75,231
Total___________________
Datasource: Cooper, B. S., and Rice, D. P. 1976. The economic cost of illness revisited.
Social Security Bulletin, Table 7.

Rice.

The

5
1
(a)
Why
is
the cost of diseases of the circulatory system so
high?
(b)
(b)
Why
is the direct cost of neoplasms relatively low
while the indirect cost
is
relatively high?
(c)
What value is
this table to policy makers? Explain your answer.
a)
2*
<(a
Is Table 2 in the reference by the Ghana Health Assessment
Team based on
the
incidence or prevalence approach?
(b)
If you
calculated a cost of illness based on Table 2, how would you expect
the rankings of diseases to change?
(c)
What value is Table 2 to
policy makers?
(d)
What are
the strengths and weaknesses of
burden of illness and cost of illness estimates?

3.
Youi have; been asked to consider the comparative burden of 4
diseases
in your community based on
the incidence approach .
Disease A has an incidence of 7/1000 per annum and a case fatality
rate of 25%.
All deaths occur in the year of onset.
Disease B has
an
incidence
of
14/1000 p.a. and a case fatality rate of 5%.
The
incidence of Disease C is
15/1000 p.a. wi th a 17% case fatality
rate,
and
Disease D’s
incidence
is
10/1000 p.a. with 22% case
fatality.

Cases of A are hospitalised in regional hospitals for an average of
6 weeks (averages are determined from the days spent in hospital by
all
patients
including those who die),
Patients who recover are
then
free from disability.
Disease B requires an average of 3
weeks
treatment at a communi ty hospital,
but v<10% of survivors
developi comj)li cations —&- -years a£te.r the
onset of the disease,
These <complicat i ons
requi re 6
weeks o f
treatment in a regional
hospi tai.
Disease C requires an average of 12 weeks treatment at a community
hospi tai,
with no further complications.
Cases of disease D are
treated
in
regional
hospitals for an
average of 5 weeks,
Complications
occur
10 years after onset
in _25% of survivors,
requiring
6 weeks treatment in a community
hospital
_____________
- ,----- ,.
Costs are $20
per day at
the«regional hospital and $10 per day at the community
hospi tai.

a.
The country has a population of 100,000.
How many days of
hospital
treatment are
required for each disease in the year of
onset?
b.
What is the total cost of hospital
in the year of onset?

c.

treatment for each disease

How many cases will survive the first year of each disease?

d.
What are the future costs of treating each disease (after the
year of onset) and in what years do they occur?
The average age of onset of Disease A is 1 year, for Disease B is
25 years, Disease C is 30 years, and Disease I), 5 years.

6

e.
Using the life expectancy table (Table A) in the reading by the
Ghana Health Assessment Project,
calculate
the total number of
years of life lost from the mortality caused by each disease.
f.
What are the costs of premature mortality of each disease in
terms of lost earnings, assuming that the working life span is 15
to 65 years, average yearly earnings are $600, and the unemployment
rate is 12%?
The average numbers of ’’disability weeks” for each disease are as
follows, (including time spent obtaining hospital treatment):

Disease

Average no. of disability
weeks in year of onset

(year 1).
A
B
C
D

g.
What
total
disease? (1 year

12
6
24
10

In subsequent years
per person with
complications
yr . 11
yr. 6

0
26 v ‘

0
0

0
0
0
52

number of years of disability are caused by each
365 days)

h.
Calculate the cost of any production losses due to disability
and
specify
the years
in which they occur using the same
specif y
assumptions as part f.

!:.iernational Journal of Epidemiology
Oxford University Press 1931

Vol. 10, No. 1
Printed in Great Britain

A Quantriatavs Method of Assessing the
Health impact of Different Diseases in
Less Developed Countries
GHANA HEALTH ASSESSMENT PROJECT TEAM*
Ghana Health Assessment Project Team. A quantitative method of assessing the health impact of different diseases
in less developed countries. International Journal of Epidemiology 1981, 10: 73—80.
A method is described for assessing quantitatively the relative importance of different disease problems on the
health of a population. The impact of a disease on a community is measured by the number of healthy days of life
which are lost through illness, disability and death as a consequence of the disease. The measure is derived by com­
bining information on the incidence rate, the case fatality rate and the extent and duration of disability produced
by the disease. In Ghana, it is estimated that malaria, measles, childhood pneumonia, sickle cell disease and severe
malnutrition are the 5 most important causes of loss of healthy life and between them they account for 34% of
healthy life lost due to all diseases.
The methodology may be used to help determine the priorities for the allocation of resources to alternative health
improvement procedures by estimating the number of healthy days of life which are likely to be saved by different
procedures and by relating these savings to the costs of the procedures.

Ghana, like many l£ss developed countries (LDCs),
has put a high priority on health. In the 20 years
since independence, there has been a threefold
increase in the number of doctors, hospital beds,
and real per capita expenditure on health, (to more
than 5 cedis (about S5) per person in 1975). In

spite of this, vital statistics have changed little:
the infant mortality rate continues to average
about 130 per 1000 live births; the expectation of
life at birth is less than 50 years; in rural areas the
maternal mortality rate is about 14 per 1000 births;
and the birth rate remains at nearly 50 per 1000
population per year.
In order to analyse why there has been so little
apparent change in the health status of the population
of Ghana, we have developed a method whereby
the health impact of different disease problems
may be estimated quantitatively. The effects of
various intervention programmes can be examined
for the magnitude of the change which they might
be expected to produce in health status. The method
may be used as a tool to aid in the planning of a
health care system.
Ideally, priorities should be determined such
that the maximum benefit is obtained from a
given expenditure. In the health sector, however,
this approach has had little application anywhere,
primarily because of problems related to determining
health benefits. The difficulties of measuring the
monetary benefits of improved health in technically
advanced countries have been well reviewed in
Bunker et al.1 Additional difficulties arise in LDCs
because of theoretical problems associated with
assessing and valuing rural productivity. For these

Members of the Project Team arc as follows: KP
Nimo, V Agadzi, R Asantc, R Biritwum, C Jones,
RH Morrow, A Neill, K Nelson, J Owusu, K. SaakwaMante, PG Smith, F Wurapa.
Report prepared by: RUM,** PGS^ and KPN’l"!’
Senior Health Planning Associate to Ministry of
Health, Ghana, Kaiser Foundation International,
PO Box 1630, Accra, Ghana. On leave from Depart­
ment of Epidemiology, Harvard School of Public
Health, Boston, Mass, USA.
Present address: Special Programme for Research
and Training in Tropical Diseases, World Health
Organization, Geneva.
t
Consultant to Ministry of Health, Ghana, for Kaiser
Foundation International.
Present address: Tropical Epidemiology Unit,
London School of Hygiene and Tropical Medicine.
London WC1E 7HT.
tt
Senior Medical Officer, National Health Planning
Unit, Ministry of Health, Ghana.
Present address: Department of Community Health,
University of Ghana Medical School, Accra, Ghana.
Reprint requests to Dr R Morrow, TDR Programme,
World Health Organization, 1211 Geneva 27, Switzer­
land.

I

I9

J
i
t

I

73

"YtiT i in ihh1 in hi

hi

i

’’mi' 'in

Oil

74

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY
v-^--''

reasons we have developed a non-monetary measure
of benefit.
TheJLmost..important effects that a disease may
have in a community are as causcs-of-illness, .disabijity_and death. With some exceptions, other
social and economic effects of a disease are directly
related to its severity as measured by these 3 factors.
We have quantified each_Qf. these factors in terms
of the number- of days of healthy life Which are
lost due to_a _diseaser and we use the total days
lost in the community as a measure of the health
impact of the disease.
The benefits from various health improvement
procedures (HIPs) can be measured by the number
of healthy days of life that can be saved by application
of the procedures. By calculating the cost of each
set of HIPs that may go into a particular health
programme, the priority of that health programme
relative to other possible programmescan be evaluated
in terms of the benefit (healthy life saved) per unit
cost.
In this paper we describe the method we have
used to determine the relative impact of different
disease problemsjTQIhanauTthis js thcJjrsLstepJn
developing priorities for health care programme^..

Measuring Disease Impact
;
».*
We have considered all the diseases which are-major
causes of illness or death in Ghana (using a modified
version of the International Classification of Diseases
B list which categorises all disease problems into
about 50 groups2). By reviewing census data,
including derived estimates of age-specific death
rates and life tables,3 cause of death as recorded
on death certificates, inpatient and outpatient
statistics, and data from special surveys and studies,
we have estimated for each disease: the incidence,
the case fatality rate., the average ages at onset and
death from the disease and the_expectarions-ciT-life
at these ages, and the extent and duration of dic­
ability and illness among those attacked by the
disease. The information has been used to obtain,
estimates of the days of healthy life which each
disease costs the community. T|xr data fnr^niir
estimates were of variable reliability and, where
possible, we derived estimates from 2 or more
different sources^
Gastroenteritis may be used to illustrate the
methods employed. In 1975, SIX, of all certified
deaths in children aged_0—4 years were arnihured
to gastroenteritis. (Death certificates, which include
the cause of death, arc issued for about 12% of all
deaths in Ghana — most arc for deaths which
occur in hospitals and they are, therefore, a biased

■';

I

sample of all deaths: in some instances we have
attempted to correct for this bias, but quantification
of the bias is difficult). The annual mortality rate
from all causes of death .averages about 45 per 1000
per
year among children aged 0—4 years (based on
\
\ demographic....analysis from the national census3)
xand about^20% .of the population is in this age
group. Thus" the crude annual. jnortality rate from
gastroenteritis may be estimated as 0,09 x 45-xJL3= 0.81/1000 population. We have assumed that
there is negligible mortality from this cause over the
age of 4 years, that there is no significant residual
disability among those who recover and that each
episode of severe diarrhoea (requiring rehydration
therapy) completely disables a child for 2 weeks
(14 days). The incidence of such severe diarrhoea in
West Africa varies with the nutritional status of the
population. Morley4 reports that in the 0—4 year
age range. 22% of normal children experienced an
episode of diarrhojja^dryean^as compared to 62%
those^with firsVdegree;maJQJ4trition. Surveys in
Ghana indicate that about 30% of Ghanaian
children have first degree malnutrition. Thus we
estimate that -34%- (.22—x-7^Q%^t <62 x 30%) of
^-children have'a.fT'episode of severe^diarrhoca each
year and thus the crude annual [ncidencc is 68. per
) 1000 total population (0.34 x .2 (20% of population
in age group 0-4 years) x 1000). Tbc_case.fatality
rate is estimated,,to„be„QAlZ68_(mortality rate/ incidence rate) = 1.2%. The case fatality rate,
independently determined from hospital inpatient
and outpatient records, wa^_Ll%, in quite close
agreement with our estimate above. In the calculation
of the days of life lost through the disease we have
taken the case fatality rate' to be 1.0% and the
incidence rate to be 70 per 1000 per year.
"’We"have further assumed that among children
affected with severe gastroenteritis the average age
of onset is 2 years and that the expectation of
life at this age is 52.8 years.3 The days of healthy
life lost per 1000 population per year due to death
is .O1JL1K-X-Lse_fat.ality rate) x 70 (incidence per
1000) x 52.8.(expectation of life in years at age 2)
x 365.25 (average days in a year) = 13 500 days; the
days lost due to sickness per 1000 population per
year is 70 (incidence per 1000) x .99 (the proportion
surviving) x 14 days (average duration.of illness)
= 970 days. Thus a total of 14470 healthy days of
life per 1000 population is lost each year due to
gastroenteritis.
In the Appendix we give the algebraic details
of the metho. 1 used to estimate the days of healthy
life lost due to each disease problem. Table 1 shows
all of the ( isease problems considered togeth.c:-

'

tabu-: 1

Disease

Disr.isf prttblrnis <>! (ib.iihi

in

Ave.
age at
onset

Ci-’R

(Ao) +

(C)

1. Cholera
15
2. Typhoid
20
2
3. Gastroenteritis__
4. Tuberculosis
20
5. Diphtheria
3
6. Pertussis
1
7. Meningitis
10
8. Polio
3
9. Measles
2
10. Malaria
1
11. Venereal disease
20
12. Leprosy
20
13. Chicken pox
4
14. Schistosomiasis
5
15. Common Cold
15
16. Guinea Worm
7
17. Yaws
4
18. Onchocerciasis
5
19. Trachoma
3
20. Hepatitis
20
21. Trypanosomiasis
15
22. Tetanus (a) neonatal
0
(b) other
15
23. Malignant neoplasms
(a) child
6
(b) adult 50
24. Diabetes
40
25. Malnutrition (severe)
2
26. Sickle Cell Disease
0
27. Hookworm Anaemia
4
28. Rheumatic Heart Disease
25
- 29. Hypertension
40
30. Other Heart Disease
35
31. Congenital Heart Disease
0
32. Cerebrovascular Disease
50
33. Influenza
20
34. Pneumonia (a) child
2
(b) adult
30
35. Peptic ulcer
25
36. Other GI Disorders
25
37. Intestinal Obstruction
30
38. Cirrhosis
30
39. Chronic renal disease
30
40. Complications of Pregnancy 20
41. Birth Diseases
(a) Prematurity
0
(b) Pneumonia
0
(c) Birth injury (inc. asphyxia) 0
(d) Congenital malformations 0
(e) Others (inc. Umbilical
0
sepsis & Haemolytic
disease)
42. Skin Infections
4
43. Psychiatric Disorders
15
44. Other Eye Diseases
60
45. Dental Disease
10
46. Gynaecological Disorders
25
47. ENT Diseases
12
48. Accidents
15

<>/ the ./.M X t>l /.'< nli'. v

% Dis­
able­
ment to
death
(Hod)

Ave.
age at
death

%

<At!)

7.6 15
7.3 20
1.0
2
35.0 25
7.0
3
1.0
1
20.0 10
5.0
3
3.0
2
2.3
1
0.01 30
25.0 30
0.02 4
4.0 30
0.0
0.0
0.0
0.0
0.0
3.0 20
19.0 17
80.0
0
35.0 15
75.0
80.0
50.0
60.0
80.0
0.1
75.0
75.0
75.0
80.0
35.0
0.1
40.0
10.0
2.0
10.0
10.0
80.0
85.0
6.5

7
52
55
2
5
5
32
50
45
10
50
20
2
30
35
25
40
35
35
20

10.2
50.0
50.0
15.0
50.0

0
0
0
0
0

0.0
5.0
0.0
0.0
1.0
0.3
10.0

25

1.25
50

4

50

75
75
30
50
50
50
50
50
50

20

20
50
75

«n
0
0
0
0
0
0
0
95
0
97.7
0

50

40
25
15

10
20

D.;yx
of
temp.
Disab.

Im idencc

(D)

(t)

(1)

14
60

0.05
4.0
70.0
2.0 .
0.01
21.0
1.25
0.22
39.0
40.0
1.7
0.5
22.0
7.0
1000.0
2.4
6.0
2.8
1.6
8.87
0.05
0.5
.75

14
200
30
30
30

25

21
2

35

25
0
96
0
0
1
5
5
0
13.5
0
0
0
0
50
0
20
5
25
25
25
20
35
0
0
0
98
0
20
20
15
5
0
0
50
85
0

35

able
mrrt

% Perm. u'. Di'

Disab.

0
95
100
10
20
4
5

!■ i.i1, i

'■ i.

14

1

0.6
45
9O>

30
70
86

45
60
90
0
30

30

180
180

25

180
30
6
30
25
30
30
75

120
21
30
30

5
60
10
25
25
25

21

0
0
20
25

0

6

30
50
15
25
25
25

30
20
30
30

; I

Days
of

life

' dt;r to
premature
death

lo-.t

0.03
0.65
0.05
1.5
1.25
19.0
0.3
0.75
0.37
0.07

99.0
65
4 755
95.3
14-170
93.3
11005
94.6
14
98.0
4643
86.6
4650
99.3
1227
17.4
23358
96.6
32567
54.1
80
25.3
3167
46.8
394
21.8
4368
67.4
600
0.0
108
0.0
886
0.0
1926
0.0
1403
0.0
4647
88.9
195
79.1
6852 100.0
4473
99.7

50.0
2.4
7.0
3.88
2.8
4.0
0.65
0.31
4.8

436
3765
217
17465
17502
1482
3211
5071
2961
929
10477
1825
18557
9112
3558
4109
4950
6568
3387
5864

98.3
91.8
52.2
99.4
87.4
24.6
81.1
70.5
72.2
81.2
56.7
42.5
99.8
97.9
22.9
96.3
73.5
86.5
85.0
82.6

9.6
.46
1.6
0.96
0.54

16774
3940
16445
5961
4625

100.0
100.0
83.3
41.4
100.0

470.0
0.66
0.05
2.8
1.0
0.56

2820
3635
123
852
815
140
14909

0.0
8.8
0.0
0.0
10.6
15.8
87.7

2.3

For some disease problems for which there arc a small proportion of deaths occurring after the year of onset, average age
at death is given as that at onset and the effects of deaths at later ages are subsumed in the days of life lost under permanent
disability, e.g. congenital malformations. • Per 1000 persons per year. t See Appendix for abbreviation.

75

, .1

'I

■&V'- ‘

76

C’,-'.

"

;

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

with the estimates of case fatality rates, incidence
rates and other relevant estimates necessary for the
derivation of the number of days of life lost due to
each disease (this number is shown also in the Table).
For some diseases it was useful for our purposes to
consider that a disease with recurrent episodes was
a single life-time process. Thus, malaria was taken
to be a single life-time disease with high mortality
in late infancy and early childhood followed by a
marked reduction in mortality by age 5 due to the
development of immunity. Thereafter, everyone was
assumed to have recurring disability from clinical
attacks of malana~averaging 7 days of illness per
year. We have assume<1 th at~ everyone in Ghana
gets malaria first at age 1 year and therefore, the
incidence is equal to the number of 1 year olds,
that~is 40/1000 total population/year (4% of the
population are aged 1 year). The days of life lost
due to the disease are all those that will be lost
during the life of the individual, but these days
are attributed to the year of first onset of the
disease. A document detailing the basis on which
the estimates in Table 1 were made is available
on request from the-authors. Table 2, which is
derived from Table 1, lists the 25 most important
TABLE 2

disease problems in Ghana ranked in order of
importance as measured by the days of healthy
life that each costs the community each year.
Priorities among Health Improvement Procedures
Ranking disease problems as described above is
insufficient to determine priorities for health
services. Priorities should be determined on the
basis of which procedures most reduce the burden
of illness, disability and death for a given unit
cost. For example, cerebrovascular accidents and
cancer are important causes of disability and death,
but, with present knowledge and resources, little
can be done to prevent or cure them and their
priority for resource allocation in LDCs should
be low.
The major effects of health improvement pro­
grammes will be to reduce the incidence rate, the
case fatality rate, and/or the extent of disability
and sickness. A particular programme may have
an effect on more than one disease (e.g., improved
nutrition will not only reduce the incidence of
kwashiorkor but will also reduce the measles case
fatality rate), and a particular disease may be
affected by more than one procedure (e.g. the

Disease problems of Ghana — ranked in order of healthy days of life lost

Rank
ord er

Disease

No. in
Table 1

Days of Healthy
life lost*

Percent
of Total

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

Malaria
Measles
Pneumonia (child)
Sickle Cell Disease
Malnutrition (severe)
Prematurity
Birth Injury
Accidents
Gastroenteritis
Tuberculosis
Cerebrovascular Disease
Pneumonia (adult)
Tetanus (neonatal)
Cirrhosis
Congenital Malformations
Complications of Pregnancy
Hypertension
Intestinal Obstruction
Typhoid
Meningitis
Hepatitis
Pertussis

10
9
34 (a)
26
25
41 (a)
41 (c)
48
3
4
32
34 (b)
22(a)
38
41 (d)
40
29
37
2

32 600
23 400
18 600
17 500
17 500
16 800
16 400
14 900
14 500
11 000
10 400
9 100
6 900
6 600
6 000
5 900
5 100
4 900
4 800
4 600
4 600
4 600
4 600
4 500
4 400

10.2
7.3

270 200

84.9

Other Birth Diseases

Tetanus (adult)
Schistosomiasis

20
6
41 (e)
22(b)
14

5.8

5.5
5.5
5.2
5.2
4.7
4.5
3.5
3.3
2.9
2.2
2.1
1.9
1.8
1.6
1.6
1.5
1.5
1.5
1.5
1.5
1.4
1.4

I

Total of first 25 diseases

•Per 1000 persons per year



HEALTH ASSESSMENT IN DEVELOPING COUNTRIES

incidence of neonatal tetanus can be reduced
cither by maternal immunisation or by well managed
deliveries)*
To illustrate 2 different procedures which may
be directed against a particular disease, the effect
of an immunisation programme against measles may
be compared with the effect of outpatient care.
Measles vaccine is 95% effective in the non-immune,
immunocompetent child,5 and under conditions in
Ghana a well-managed immunisation programme
may reach 70% of susceptible children.6,7 Thus
administration of the vaccine would be expected to
reduce the incidence of measles in the country by
0.95 x 70% = 66.5%. but would make no change in
the case-fatality rate or in the duration of illness
for those who develop measles, in the absence of
any medical services we have estimated that measles
would ‘cost’ the community 27600 days of healthy
life per 1000 population each year and the number
of days of life which would be saved by an immuni­
sation programme is 27600 x .665^= J 8J54. If the
cost of the measles vaccine and its delivery are 0.5
cedis for each child immunised, the cost per thousand
population per year would bt^40 (the number of
1 year olds in a population of 1000) x 0.5, = 20
cedis. Thus the benefit would be about 918 health^
days’ of 'life^per cedi/ In fact, the cost of such
immunisation is highly dependent upon the in­
frastructure which exists for the administration of
the vaccine, and it is likely that measles immunisation
would be done in conjunction with other immuni­
sation procedures. In general, the immunisation
programme should be considered as a package and
the combined benefits and costs measured.
Outpatient therapy for measles is estimated to
reduce case-fatality rates by 50% (chiefly in those
with bacterial pneumonia and/or gasr enteritis).
About 30% of families in Ghana are abk io take
children with measles for outpatient care; thu;. he
days of healthy life saved by outpatient care would
be 26800 (the days lost attributable to mortality,
the 800 days lost due to morbidity is assumed
unaffected by treatment) x .30 (the proportion
of population utilising the health services) x .50
(the reduction in CFR) equals 4020 healthy days of
life per thousand population per year. The cost of
an outpatient visit, including drugs, is about 2 cedis.10
If each child with measles makes an average of 3
visits the cost would be 6 cedis per case. The cost
for outpatient care for measles would be 40 (the
incidence rate/1000) x .30 (the coverage) x 6 cedis,
that is 72 cedis per thousand population per year.
Under these assumptions^ outpatient care' would
save 56 days of healthy life per cedi expended.

1y

Thus compared io outpaticnr care an immunisation
programme against measles could save 16 times as
many days uU)calthy-lifc_pcr cedi.

j

DISCUSSION

fhe concept of healthy days of life lost is an extension
of similar measures that have been based on mortality
data alone. The development of these latter measures
and their potential use for health planning has
been ^reviewed by Romeder and AlcWhinnie11
but, to our knowledge, they have not been used
directly in the allocation of health resources.
Sullivan12,13 has proposed a similar index for measur­
ing the effects of morbidity and mortality in techni­
cally advanced countries, but he did not relate
this measure to specific diseases and thus, as presently
formulated, the measure is not likely to be of
direct value in determining the allocation of health
resources. The concept of ‘days of life lost’ has
been discussed for specific diseases, but a qualitative
method was used to determine resource allocation.14
Recently Shepard and Thompson?^ have proposed a
measure of benefit in terms of additional qualityadjusted life-years which is conceptually similar
to healthy days of life lost. They discuss four
possible types of health effects resulting from
health programmes — additional years of healthy
survival, additional years of disease, improved
health without additional survival, and additional
years but with restrictions — and address both the
value and the practical difficulties of adding these
effects together using a common unit of measure.
Other, more narrowly-defined approaches to assess
benefits such as estimating- the aggregate output
increases of labour as a _resuk.Df _controLofTropical
diseases have been reviewed by Prescott.16
The method we have proposed has a number of
advantages over qualitative approaches to health
planning and resource allocation:
(1) Quantitative assessment is made of the benefits
and costs of alternative health programmes.
To be able to state that for a fixed expendi­
ture an immunisation programme against
measles may save 16 times more healthy
life than would outpatient care is a more
persuasive argument for allocating resources
to the former than the mere assertion that
prevention is better than cure. Such calcu­
lations may be extended for all the procedures
which might be included in a primary health
care programme, and it would be possible to
compare the amount of life which is likely to
be saved by such a programme with that
saved by equal expenditure on other resources,

5

^fc;:.Jfc:
78

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

such as an expansion of hospital facilities.
With the exception of the extent of disability,
each of the variables required to estimate the
days of healthy life lost due to a disease
may be measured objectively.
(3) Attention is focused on the key information
which should be collected by health statis­
ticians. A distressing feature of much statistical
data collected routinely in LDCs is its irrele­
vance to health planning or care. To obtain
some of the required information special
studies may be needed, but it is clear what
data is required to be able to apply our
methodology.
(4) A framework is set up for the evaluation of
the planned programme after it has been
implemented since specific targets are provided
in terms of the changes in incidence, case
fatality rates, and disability which were
predicted.
(5) The assumptions that are used for setting
health priorities are made explicitly: if the
conclusions are challenged, the individual
steps for estimating the costs and benefits
in the computations can be examined for
their reasonableness and alternative assump­
tions can be readily considered.
(6) The health planner is forced to consider
the effects of each HIP on a disease by disease
and community by community basis to
estimate the expected amount of healthy
life that will be saved and the costs that will
be incurred. Examination of the amount of
disability and death due to each disease may
highlight those against which efforts may
most usefully be directed. Similarly, analysis
of the cost of programmes may indicate
those for which efforts towards cost reductions
should be focused.
It is important to consider the limitations and
approximations in the proposed approach. Many
of the data that are required to estimate the days
of healthy life lost due to a particular disease (c.g.
incidence rates, case fatality rates) are not available
from routine sources. In deriving estimates of
these rates for Ghana a variety of sources have
been used; in some instances the rates used have
been based upon little more than a consensus.
More work is required to assess the sensitivity
of the conclusions to changes in these estimates and
such studies will enable us to identify which esti­
mates are in need of further refinement. For sonic
diseases it seems likely that our estimates arc
reasonable, but for others we have made, at best,
(2)

a rough approximation. 2r. general, estimates of
death and disability rates Lre more reliable than are
estimates of rhe extent c: disability. It is apparent
from Table 1 that the mc<: important determinant
of loss of days of healthy life is premature death;
disability and illness play a relatively minor role
for most diseases. These laner 2 factors are difficult
to quantify objectively. We have equated disability
with death in the sense that a day of ‘total’ dis­
ability is equivalent to the loss of a day due to
death. For example, we have assumed that patients
with leprosy have an average disablement of 25%
from onset of the disease for the rest of their
lives. Such an assessment is subjective, but for all
diseases combined only 30% of all healthy life
lost is attributed to disability and illness. Thus
changes in estimates of disability are likely to
have a relatively small effect. However, for some
specific diseases this is not the case. For example,
if we assume that jnalaria-completely disables an
adult for 2 weeks a...year (4%) rather than for one
week (2% — as in Table 1) the estimated annual
number of days of life lost is 47 500 days per 1000
population rather than 32 600 days per 1000
population. Which of these 2 estimates of dis­
ability rate is most reasonable might be determined
through suitably designed special studies.
Our calculations use the average age at onset and
death as more detailed information on age specific
rates is not available in Ghana. It is unlikely, however,
that the use of age and sex specific disease rates
would make any substantial difference to the
conclusions as the onset of most of the important
diseases occurs over a small age range.
Often it is difficult to attribute sickness or
death to a single disease. Malnutrition, malaria,
diarrhoea and measles in childhood are very common,
but frequently it is the combination of these diseases
which is fatal rather than any of them individually.
We were careful to avoid double counting of deaths
when estimating death rates by cause, but the
attribution of days of healthy life lost between
causes is subject to some uncertainty. For many
purposes it is useful to consider a number of diseases
together — particularly when estimating the effects
of different health improvement programmes.
For example, a nutritional programme should
prevent deaths attributable to both malnutrition
ajid measles.
‘Days o£_Jiealthy life’ as a measure of benefit
may be useful for determining allocation of resources
within the health sector, but it cannot be used to
determine the appropriate allocation of resources
between the health and other sectors (such as

‘‘v.f.f.- i

HEALTH ASSESSMENT IN DEV ELOPING COUNTXll.S

J

education and agriculture). Without putting a
specific monetary valuc__on Jifc,_tlns is. an unavoid­
able limitation. However, the method is a con­
siderable. advance for purposes of allocation within
the health sector and can be used also to quantify
the effects on health of programmes in other
sectors.
The method we have proposed for measuring
the importance of different diseases values individuals
in direct proportion to their expectation of life
at their current age. Thus the death of a child is
regarded as costing the community more than
the death of an adult. This is appropriate if the
objective is to maximise the total amount of healthy
life of the community over time and if one ignores
any effect that the death or disablement of an
adult has on their dependents (e.g. the death of a
mother greatly increases the death rate among her
dependent children — the life lost by the children
should thus be attributed to the cause of the maternal
death). But not all would agreethatthisisappropriate.
We have discussed this issue at length with colleagues
in Ghana, where children are valued highly (as in
most societies) and have found substantial support
for our assumptions. We have tried a number of
alternative methods by giving different weights
to years of life at different ages and of discounting
future years of life. In most cases these alternatives
do not produce any very marked alteration in the
ranking of the relative importance of different
diseases. Of course, extreme assumptions (such as
totally disregarding deaths below the age of 15
years) do produce different rankings. It would be
straightforward to adopt different systems of
weightings if these were considered more appropriate
in particular settings.
It should be noted that reduction of the ’'•ath
rate from any disease increases the expectation of
life and consequently would increase the potential
benefits of any other health improvement procedure
aimed at other diseases. We have taken no account
of this effect, but this, and other refinements,
might be applied if the quality of data available
were better.

ACKNOWLEDGEMENTS
We thank SRA Dodu, EA Badoe, S Ofosu-Amaah,
AK Foli, KK Korsah, JOM Pobee, GA Ashitey, AG
Boohene, M Adibo, Paul Zukin and EG Beausoleil
for helpful advice and encouragement.
This work was supported in part by USAID
Project No. 641
641-0078
—0078 awarded to Kaiser
Foundation International, Oakland, California,
USA.

..._______

79

REIERENCES
Bunker J1‘, Barnes BA, Mostellcr F. (IMs). Costs, Risks
and Bcncf is of Surgery. New York: Oxford Univer­
sity Press, 1977.
“ World Health Organization, Eighth Revision of the
International Statistical Classification of Diseases,
Injuries and Causes of Death. Geneva, 1967.
3
Gaisie SK. Estimating Ghanaian Fertility, Mortality
and z\gc Structure. Population Dynamics Programme,
University of Ghana, L.egon, Accra, 1976.
4
Morley D. Paediatric Priorities in the Developing World,
180—182. London: Butterworths, 1973.
World Health Organization. Expanded Programme for
Immunisation Manual, Geneva, 1977.
6
' Danfa Comprehensive Rural Health and Family Planning
Project. Ghana, University of Ghana Medical School
and UCLA School of Public Health (unpublished
data).
7
Ghana Ministry of Health. Expanded Programme for
Immunisation. Division of Epidemiology, Accra,
(unpublished data).
8
Health Needs and Health Services in Rural Ghana,
Volume 1, A Report to the Government of Ghana,
Institute of Development Studies, University of
Sussex; Institute for Statistical, Social and Economic
Research, University of Ghana, Legon; National
Health Planning Unit, Ministry of Health, Ghana;
and Department of Community Health, University
of Ghana Medical School, Korle Bu. 88—89, 1978.
9
Korle Bu Outpatient Statistics. Centre for Health
Statistics, Ministry of Health, Ghana, 1975.
10
Brooks RG. Cost Analysis of Selected Medical Care
Institutions in the Eastern Region of Ghana, 1975.
Unpublished Report available from Department
of Economics, University of Legon, Ghana.
1 1
Romedcr JM and McWhinnie JR. Int J Epidemiol.1977; - *
6= 143-151.
- c
LI Sullivan DF. Conceptual problems in developing an
index of health. US Public Health Service Publication
Series No. 1000. Vital and Health Statistics Series
2, No. 17. National Centre for Health Statistics,
U

I

?!

?966-

Sullivan DF. Health Services and Mental Health Admini­
stration Health Reports 1971; 86: 347—354.
r^-Pan American Health Organization. Scientific Publication
No. 111,1965.
Shepard DS and Thompson MS. First principles of
cost-cffcctivencss analysis in health. Public Health
Rep 1979; 94: 535-543.
"
‘ Y
16
Prescott NM. On the benefits of tropical disease control.
Health Policies in Developing Countries. Royal
Society of Medicine International Congress and
Symposium Series No. 24. London: Academic Press,
1980.

H

APPENDIX

Estimation of the days of life lost due to a disease
Consider a particular disease problem.
= average age ar onset
Let Ao
= average age at death of those who die
Ad
of the disease

lb

«■

K'j ■fc- =
80.

•*h

international journal of EPIDEMIOLOGY

E(AO) =
C
^od

TABLE A

expectation of life (in years) at age Ao
(The values used for Ghana are given
in Table A)
case fatality rate (expressed as a percent)

Q

percent disablement in the period

Abridged Life Table for Ghana, 1968 (Gaisie
1976)

Age
Years

Expected years of life
remaining*

0

46.9

1

52.4

5

53.8

10

50.6

15

46.5

20

42.5

25

38.7

30

34.9

35

31.1

40

27.4

45

23.8

50

20.2

55

16.7

60

13.5 *

65

10.5

70

8.0

75

5.9

80

4.2

•Weighted average of males and females based upon sex
distribution in 1970 census population

D
t

from onset until death among those
who die of the disease (i.e. Doc} =
0 = no disablement, Docj = 100 =
disablement equivalent to death)
= percent of those affected by the
disease, who do not die of the
disease but who are permanently
disabled
:= percent disablement of those per­
manently disabled
s= average period of temporary disable­
ment (days) among those who are
affected but neither die nor are
permanently disabled, multiplied by
the proportion disablement of those
temporarily disabled

The average number of days of healthy life lost
to' the community by each patient with the disease
is given by: —
Days lost due to:
L = (C/100). [E(A0)-(Ad-A0)].365.25
: premature deaths
+ (C/100).(Ad-Ao).(DO(j/100).365.25
: disability before death
+ (Q/100).E(Ao).(D/100).365.25
: chronic disability
+ [(100-C-Q)/100].t
: acute illness
EcL-L = annual incidence of the disease (new cases/
1000 population/year)
Then the number of days lost by the community
which are attributable to the disease is:

JJ/1000 population
This is the method which was used to derive the
results in the last two columns of Table 1.

(Revised version received 13 October 1980)

i

Cost analysis

4. Cost analysis

I
I

I

4.1. SOME BASICS
The analysis of the comparative costs of alternative treatments or health
care programmes is common to all forms of economic evaluation and
therefore most of the methodological Issues discussed in this chapter are
likely to be of relevance to all analyses. Two particularly thorny issues, the
treatment of overhead costs (techniques for allocating shared overhead
costs to individual projects) and allowance for differential timing of costs
(the techniques of discounting and annuitization of capital expenditure),
will be discussed in some detail. However, the chapter begins by covering
some of the basic questions that an evaluator might have when embarking
on a costing study in the health field.

4.1.1. Which costs should be considered?
The main categories of costs df health care programmes or treatments
were identified in Fig. 3.1 of Chapter 3; these are the organizing and
operating costs within the health sector, costs borne by patients and their
families, and costs borne externally to the health sector, patients, and
their families. The particular range of costs included in a given study is
likely to be decided upon as a result of considering the following four
points.
1. What is the viewpoint for the analysis ?
It is essential to specify the viewpoint since an item may be a cost from one
point of view, but not a cost from another. (For example, patients’ travel
costs are a cost from the patients’ point of view and from the point of view
of society,, but not a cost from the Ministry of Health’s point of view.
Workmen’s compensation payments are a cost to the paying government,
a gain to the patient (recipient), and neither a cost nor a gain to society.
(These money transfers, which do not reflect resource consumption, are
called transfer payments by economists. Costs are involved in their
administration, but these are not measured by the amounts themselves.)
39

Possible points of view Include: society, Ministry of Health, other
provincial ministries, total provincial government, patient, employer,
federal government, the agency providing the programme, etc. if the
evaluation is being commissioned by a given body, this may give a clue Io
the relevant point(s) of view. However, when in doubt always adopt the
societal poiht of view, which is the broadest one and is always relevant.

2. Is the comparison restricted to the two or more programmes
immediately understudy?
If the comparison Is restricted to the programmes or treatments
immediately under study, costs common to both need not be considered
us they will not affect the choice between the given pfogramines.
(Elimination of such costs can save the evaluator a considerable amount
of work.) However, if it is thought that at some later stage a broader
comparison may be contemplated, including other alternatives not yet
specified, it might be prudent to consider all the costs of the programmes.
3. Are some costs merely likely to confirm a result that would be obtained
by consideration of a narrower range of costs?
Sometimes the consideration of patients' costs merely confirms a result
that might be obtained from, say, consideration of only operating costs
within the health sector. Therefore, if consideration of patients’ costs
requires extra effort and the choice of programme would not be changed,
it may not be worthwhile to complicate the analysis unnecessarily.
However, some justification for such an exclusion of a cost category
should be given.

4. What is the relative order of magnitude of costs ?
It is not worth investing a great deal of lime and effort considering costs
that, becmisc they tire small, me unlikely to make any difference to the
study result. I lowever, some justification should be given for the elimina­
tion of such costs, perhaps based on previous empirical work. Il is still
worthwhile identifying such cost categories in any event, although (he
estimation of them might not be pursued in any great detail.
Above all, the main point to remember when embarking on a costing
study isjhat, to an economist, cost refers to the sacrifice (of benefits)
made when a given resource is consumed in a programme or treatment.
Therefore, it is important not to confine one’s attention to expenditures,
but to consider also other resources, the consumption of which is not
adequately reflected in market prices, e.g., volunteer time, patients’
leisure lime, donated clinic space, etc.

40

Cost analysis

4.1.2. How should costs he estimated?

Once the relevant range of costs has been ideiHified (he individual items
must be measured and valued. In general, the programme ingredients
approach suggested in Chapter 3 should suffice and market prices will be
readily available for many of the cost items. Although the theoretical
proper price for a resource is its opportunity cost (i.e., the value of the
forgone benefits because the resource is not available for its best alterna­
tive use), the pragmatic approach to costing is to take existing market
prices unless there is some particular reason to do otherwise (e.g., the
price of some resources may be subsidized by a third party such as a
charitable institution).
Although the costing of most resource items is relatively unambiguous,
the following five issues commonly arise in costing studies.

1. How are values imputed for nonmarket items ?
'flic major nonmarket resource inputs Io health care programmes are
volunteer time and palient/family leisure lime. One approach to the
valuation of these would be to use market wage rates (e.g., for volunteer
time one might use unskilled wage rates). The market value of leisure lime
is harder to assess. One can argue for a value of lost leisure time of
anything from zero, through average earnings, to average overtime
earnings (time and a half or double time). The argument for the overtime
rate is that this is the price that an employer must pay, at the margin, to
buy some of the worker's leisure lime. 'Ute most common practice is Io
value lost leisure lime al zero in the base case analysis, and to investigate
the impact of the other assumptions through sensitivity analysis.
A slightly different approach is to identify and measure units of, say,
volunteer input and to document these alongside the other costs when
reporting results. This would enable the decision-maker to note those
programmes relying heavily on volunteers. Il would then be up to the
programme director Io demonstrate that such an input could be obtained
without an opportunity cost to other programmes arising from the
diversion of volunteers Io the new programme.
The valuation of nonmarket items is discussed further in Chapter 7 on
cost-benefit analysis.

2. How should capital outlays (on equipment, buildings and land) be
handled?
Capital costs are the costs to purchase the major capital assets required
by the programme: generally equipment, buildings and land. Capital costs
differ from operating costs in a number of ways. First, they represent
investments al a single point in time, often al the beginning of the

41

Cost analysis

programme, rather than annual sums like operating costs. Frequently, the
capital costs are not listed in the accounts or budgets of the organization
because they have been funded in advance, perhaps by a one-time grant,
while the budgets and accounts represent operating expenses only.
Sometimes, the annual budgets and accounts contain an item called
depreciation which relates to capital costs, as explained below.
Capital costs represent an investment in an asset which is used over
time. Most assets, such as equipment and buildings, wear out, or
depreciate, with time. On the other hand, land is a non-depreciable asset
because it maintains its value. There are two components of capital cost.
One is the opportunity cost of the funds tied up in the capital asset. This is
clearly seen in the case of land. Although an investment in non­
depreciable land will return the original capital sum when sold, there is
still a ‘cost’. This cost is the lost opportunity to invest the sum in some
other venture yielding positive benefits. This is called the opportunity
cost and is valued by applying an interest rate (equal to the discount rate
used in the study) to the amount of capital invested.
The second component of a capital cost represents the depreciation
over time of the asset itself. Various accounting procedures (straight line,
declining balance, double declining balance, etc.) are available for use in
the accounts of the organization. Often, accounting practices relate more
to the company tax laws governing the depreciation of assets than to the
real change in the value of the asset.
There are several methods of measuring and valuing capital costs in an
economic evaluation. The best method is to annuitize the initial capital
outlay over the useful life of the asset; that is, to calculate the ‘equivalent
annual cost*. 1 his method and its advantages are discussed in more detail
by Richardson and Gafni (1983). The method automatically incorpo­
rates both the depreciation aspect and the opportunity cost aspect of the •
capital cost. Il is our preferred approach and is described in Section 4.2
below. An alternative but less exact method is to detennine the deprecia­
tion cost each year using an accounting method and to determine the
opportunity cost on the undepreciated balance for each year (See Levin
1975, Boyle, Torrance, Horwood, and Sinclair 1982). Where market
rates exist for the rental of buildings or lease of equipment, these may be
used to estimate capital costs. This method also incorporates both the
depreciation and the opportunity components of the cost. (A series of
exercises illustrating the different methods of measuring and valuing
capital costs is given in Annex 4.1.)
If capital outlays relate to resources that are used by more than one
programme they may require allocation in a similar fashion to 'overhead'
costs. See the discussion of this point below.
42

,

I

i

I
!

Cost analysis

Cost analysis

3. What is the significance of the average cost/marginal cost distinction ?

Economists tend to emphasize this point, and the example of the sixth
stool guaiac in Chapter 2 illustrated the pitfalls in making decisions based
on average cost. In fact, marginal cost and average cost arc but two con­
cepts relating costs to quantity (Horngren 1982). A longer list would
comprise:

I

Total cost (TC)

“cost of producing a particular quantity of
output.
Fixed cost (FC)
“ costs which do not vary with the quantity of
output in the short run (about one year), e.g.
rent, equipment lease payments, some wages
and salaries. That is, costs which vary with time,
rather than quantity.
Variable cost (VC) “costs which vary with the level of output, eg.
supplies, food, fee for service.
Cost function (TC) “/(Q), total cost as a function of quantity.
Average cost (AC) “ TC/Q, the average cost per unit of output.
Marginal cost (MC) - (TC of x 4-1 units) - (TC of x units).
“ d(TC)/d(2 evaluated at x
“ the extra cost of producing one extra unit of
output.
1 he major significance of the averge-cost/marginal-cost distinction to
the evaluator is as follows. First, when making a comparison of two or
more programmes, it is worth asking independently of each, ‘What would
be the costs (and consequences) of having a little more or a little less?’
[e.g., suppose Ncuhauser and Lewickl (1975) had been comparing the
six-stool protocol for detecting colonic cancer with another diagnostic
lest. Perhaps the question of six- versus five-tests may never have been
asked!) Second, when examining the effects (on cost) of small changes in
output, it is likely that these will differ from average costs. For example,
the extra cost of keeping a patient in hospital for another day at the end of
his treatment might be less than the average daily cost for his whole stay.
(In fact, this issue usually arises in the opposite sense—the savings from a
reduction of one day’s stay are usually tower than the average daily cost.)

4. How should shared (or overhead) costs be handled?
The term ^overhead costs is an accounting term for those resources that
serve many different departments and programmes, e.g. general hospital
administration, central laundry, medical records, cleaning, porters,

43

I

iA-t

V J

i
O *

Ms.

A

1

'’A

'^4

•>

''—/

power, etc. If individual programmes are io be costed, these shared costs
may need to be attributed to programmes.
The main point to note at the outset is that there is no unambiguously
ng/i/ way to apportion such costs. The approach that is favoured by
economists is to employ marginal analysis. That is, to see which (if any) of
such costs would change if a given programme were added to, or
subtracted from; the overall activity. Whilst this is fine up Io a point, the
most common sltmition Is llial the Choice Is lidt such tin addition or
subtraction, but one between two programmes, each of which .would
consume the given central services (perhaps because they are competi­
tors for the same space in the hospital). For example, suppose the ques­
tion concerned space In the hospital that could be used either for
anticoagulant therapy for pulmonary embolism, or for renal dialysis. If
the economic evaluation concerned a choice between these two
programmes, then there would be no methodological problem, the costs
associated with use of the space would be common to both and could be
excluded from the analysis. However, typically the comparison might be
between the anticoagulant therapy and another programme in the same
field. This could be a programme of more definitive diagnosis of
pulmonary embolism, which would avert some hospitalization. In such ani
instance it would be relevant Io obtain an estimate of the value of the freed

resources (e.g. hospital door space) that could be diverted Io other uses.
Essentially, the Issue at stake here is that of accurately estimating all the
costs attributable to a given programme or treatment when, this is
delivered alongside other programmes, as in the acute hospital. In
Chapter 3 the reader was warned against the unthinking use of hospital
(or other institutional) per diems or average costs. Before the methods
available for apportioning institutional costs are described in more detail,
the dangers of using per diems require more elucidation.
Many institutions calculate a per diem or average cost of their
operations. This is essentially their total operating costs for the year
divided by their total patient utilization for the year. A common example
is a hospital's average cost per patient-day. It is tempting simply to
multiply this figure by the number of patients and their average length of
stay to determine the hospital cost of a programme. What Is wrong with
this procedure? First, it is only valid for truly ‘average’ patients—that is,
patients who use an average amount of radiological services, laboratory
services, operations, nursing attention, drugs, and so on. If patients in the
programme being costed are not average, the result will be in error.
Second, many per diem calculations include arbitrary adjustments. For
example, certain types of patients (outpatients, day patients, newborn
patients, etc.) may be excluded from (he denominator of (he calculation in

44

CJ
O

IU
,

Cost analysis
recognition that they are not typical. Then an estimate of tile costs of these
patients (often a very crude estimate) is subtracted from the numerator
before calculating the per diem. The result is that the per diem itself is
imprecise, even for the truly average patient.
Finally, typical per diem cost figures are incomplete, as they totally
ignore capital costs. In summary, per diem costs are only applicable to
average patients and even then are imprecise and incomplete.
A number of methods can be used Io determine a more accurate cost of
a programme in a hospital or other setting where shared (or overhead)
costs are involved. The methods are illustrated below in terms of a
hospital setting. The basic idea is Io determine the quantities of service
consumed by the patient (days of stay in ward A, B, or C, number of
laboratory tests of each type, number of radiological procedures number
of operations, etc.), to determine a full cost (including the proper share of
overhead, capital, etc.) for a unit of each type of service, and to multiply
these together and sum up the results. The allocation methods described
below are different ways Io determine the cost per unit for each type of
service. In these methods the overhead costs (e g., housekeeping) are
allocated to other departments (e;g., radiology) on the basis of some
measure, called an alloculion basis, judged Io be related to usage of the
ovci head item (e g., square feet of floor space in the radiology depart­
ment might be used to ullocale housekeeping costs to radiology).
In deciding which of the following approaches to use, the comments
made in Section 4.1.1 above, should be borne in mind. That is, the more
important the cost item is for the analysis, the greater the effort that
should be made Io estimate it accurately. There may conceivably be
evaluations for which simple per diem costs will suffice, since lhe result is
unlikely to change irrespective of lhe figure assumed for the cost of
hospital care. However, we suspect that such situations are in the
minority, given lhe relative order of magnitude of hospital costs
compared to other elements of health care expenditures.
Alternatively, lhe intermediate approach suggested by Hull Hirsh
Sackett, and Sloddart (1982) may suffice. Here lhe per diem cost is
puiged <>f any Hems i eluting Io medical cure costs, leaving just lhe 'hotel’
component of hospital expenditure. Il is then assumed that all patients are
average m respect of their hotel costs and that this expenditure can
therefore be apportioned on lhe basis of patient days. Thus, lhe hotel cost
can be calculated for the patients in lhe programme of interest and
combined with the medical care costs attributable to those patients Io give
the total costs of the programme. (The medical care costs would be
estimated separately, using data specifically relating to the patients in the
programme.)

45

Cost analysis
If a more detailed consideration of costs is required, various methods
for allocating shared (or overhead) costs are available, namely:

(a) Direct allocation (ignores interaction of overhead departments)..
Each overhead cost (e.g., central administration, housekeeping) is
allocated directly to final cost centres (e.g., programmes like day
surgery; departments like wards or radiology). Programme X’s
allocated share of central administration is equal to central
administration cost times Programme X’s proportion of the
allocation basis (say, paid hours). Note, Programme X’s propor­
tion is Programme X’s paid hours divided by total paid hours of all
final cost centres, not total paid hours for the whole organization.
The latter method would underestimate the costs in all final cost
centres.
(b) Step down allocation (partial adjustments for interaction of over­
head departments). The overhead departments are allocated in a
stepwise fashion to all of the remaining overhead departments and
to the final cost centres.
(c) Step down with iterations (full adjustment for interaction of
overhead departments). The overhead departments are allocated
in a stepwise fashion to all of the other overhead departments and
to the final cost centres. The procedure is repeated a number of
times (about three) to eliminate residual unallocated amounts.
(d) Simultaneous allocation (full adjustment for interaction of over­
head departments). This method uses the same data as (b) or (c)
but it solves a set of simultaneous linear equations to give the
allocations, it gives the same answer as method (c) but involves less
work. (The method is shown diagrammaticaliy in Fig. 4.1.)

I

An example showing the different approaches to the allocation of
overhead costs is presented in Section 4.3. Further details are available in
Horngren (1982), Clements (1974), Kaplan (1973), and Boyle, etal
(1982).,
The effort that one would put into overhead cost allocation would
depend on the likely importance of overhead costs (in quantitative terms)
for the whole analysis. A much simpler, but cruder, approach is to
I

(a) identify those hospital costs unambiguously attributable to the
treatment or programme in question (e.g., physicians* fees, labora­
tory tests, drugs). (These are known as the directly allocatable
costs.) Allocate these directly and immediately to the programme,
then;
(b) deduct, from total hospital operating expenses, the cost of depart-

46

I

Cost analysis

Cost analysis

(d) finally, undertake a sensitivity analysis.

Capital
coala

Whilst there is nothing to suppose that this method is anything but
crude, if the choice between programmes is fairly insensitive to the valiie
derived it may suffice.

2

1

z>—Support -----

5. How should indirect costs be estimated?

As was mentioned in Chapter 3, this Is a particularly contentious issue.
The discussion of this point will be postponed until Chapter 5 since
changes In productive output more often enter into the economic evalua­
tion as a consequence of health crire programnies, (hat is, (he therapy
often averts future production losses III (hnt It biudilcs the sick person to
return to work of work until later in life. Production losses occur less
often on the cost side of the equation since the patient Is already off work
because of his or her condition. Exceptions here would include popula­
tion screening or other preventive programmes and anyone considering
an evaluation of these should consult (he relevant section in Chapter 5.

' '—► Support -z v

~~TJ

r yf Support ----

Direct
costs

4
z '—►

Service

z x—-» Service

-►

*

4.2. ALLOWANCE FOR DIFFERENTIAL TIMING
OF COSTS (DISCOUNTING AND HIE
ANNUITIZATION OF CAPITAL EXPENDITURES)

Service

As was mentioned in Chapter 3, some allowance needs to be made for the
differential timing of costs and consequences. That is, even in a world
with zero innation and no bank interest, it would be an advantage to
receive a benefit earlier or to incur a cost later—it gives you more options.
Economists call this the notion of time preference.
Typically, economic evaluation texts discuss the situation where the
costs of the alternative programmes A and B can be identified by the year
in which they occur:
«

5

I

Pallent
care

Patient

r-.

Fl8’4'1’ 1982)lat,C lllUS,ration of cost avocations (from Boyle er al.

Year
I

ments already allocated above and departments known not to
service the programme being costed, then;
(C) al'OCatVhe rtema.inder of hospital operating expenses on the basis
of number of patient days, e.gj

1
2
3

Cost of Programme A
(SOOOs)

Cost of Programme B
(SOOOs)

5

15
10
4

10
15

In this example, B might be a preventive programme which requires more
outlay In Year 1 with the promise of lower cost in Year 3. 1 he crude
addition of the two cost streams shows B to be of lower cost, but (he
outlays under A occur more in the later years.
A comparison of A and B (adjusted for the differential timing of
resource outlays) would be made by discounting future costs to present

Net hospital
expenditure

Hospital
Hospital cost Directly
patient-days
of
the!
------— allocatable +
Total number X B,lr>butable
programme
costs
to the
of hospital
programme
patient-days

48
47

Cost analysis

Cost analysis

basis, obtaining an equivalent annual cost (E) for the capital outlay by an
amortization or annuitization procedure, lliis works as follows:

values. The calculation is performed as follows. If P — present value;
Fn — future cost at year n; and r — annual interest (discount) rate (e.g.,
0.05 or 5 per cent), then

If the capital outlay Is K, we need to find die annual sum E wliich over a
period of n years (the life of the facility), at an interest rate of r, will be
equivalent to K.

3

PI

This is expressed by the following formula:

-A-+-6+ pj
(1+r) + (l+r),T (1+r)’

K-

E

E

E

uT7>+
+(Fm'+--+ (l+rf

In our example lliis gives:
K-E 1- (l+r)-'
r

Present value of cost of A — 26.79
Present value of cost of B — 26.81

K — E|Annuity factor, n periods, interest r]

This assumes that the costs all occur at the end of each year. An
alternative assumption which is commonly used is to assume that the
costs all occur at the beginning of each year. Then, Year 1 costs need
not be discounted. Year 2 costs should be discounted by one year, etc.
Calculated in this way, the previous example is:

As before, the annuity factor is easily obtainable from Table 2 in
Annex 4.2. Fpr example, in the cost analysis of providing long term
oxygen therapy Lowson, Drummond, and Bishop (1981) found the total
capital (set up) costs (K) to be £2153.
Therefore, applying the formula given above:

2

p-

2153 -

n-0

“Fo +

_L_+
£ +
g
,
E
,
E
(1+r) (1+r)2 (1+r)3 (1 + r)4 (1+r)’

2153 - E (Annuity factor, 5 years, interest rate 7 per cent)

(1+r)

(1+r)2

2153 — E|4.1002| (from Table 2 in Annex 4.2)

Present value of A — 28.13

E —£525 (as shown in Table III of Lowson er o/., (1981).
Note that Lowson et al (1981) assumed that the annuity was in arrears,
that is, due at the end of the year. It might be argued that a more realistic
assumption would be that it were payable in advance, lliis is equivalent to
the formula:

Present value of Bi— 28.15
The factor (1 + r)~n is known as the discount factor and can be
obtained for a given n and r from Table 1 in Annex 4.2. For example, the
discount factor for three periods (years) at a discount rate of 5 per cent is
0.8638.
While this approach is the most convenient for a number of pro­
gramme comparisons, a more common situation is that where most of the
costs are easily expressed on an annual recurring basis and it is only
capital costs which differ from year to year (typically these will be at the
beginning of the programme, or Year 0).
I lere it might be more convenient to express all the cosls on an annual

F
!

i

a.
__
2153- ——— + —? +
(1 + r) (1+r)1 (1 + r)3 (1 + r)4

The value for E can still be obtained from Table 2 by taking one less
period and adding 1.000. This gives a lower value for E - £491. Tills is
logical since the repayments are being made earlier (at the beginning of
each year) rather than in arrears.
<n

Cost analysis
This approach can be generalized td
equipment or buildings linvc ri resale value U;
5the resale value;
nthe useful life of the equipment;
r—
discount (interest rate);
4(n, r) ~ the annuity factor (n years at interest rate»
K ■"
purchase price/inltial outlay;
equivalent annual cost;
then

This can be estimated empirically, although not without
controversy.
i
(b) r — the social rale of lime preference.

S
KE-------- (l+rf
zl(/!,r)

!
I

Choke 1 -Use the replacement cost of the equipment (or die original fH
Choice?
^^£u-H?!JLi9!lars)andafU||life'.-------- 1(1'

2“”"™, s:1"’ ’*“■»id “d""' - »■
Choice 1 is usually better as the results are more generalizable-less
situadonal. Note that using the undepreciated balance from die accounts
of the orgamzadon is never a method of choice.
accounts
to n O?" be Seen tha‘d,e et’ulvaIen'annuaI cost of buildings or equipment
K Pro8rnnimc depends on the values of n, r, and 5, all of which
Tval.
aSS-T 31 'e ,ime °f “,e evaluati°n- Practical points that
evaluators might care to note are:
,
P

i

I

The social rate of time preference is a measure of society’s willingness,
collectively, to forgo consumption (gratification) today in order to have
greater consumption (gratification) tomorrow. Frequently it Is assumed
that the interest rate on a risk free investment (e.g., long-term government
bonds) represents the individual investor’s willingness to forgo the
present for the future, and that this rate is the individual’s rate of time
preference. Thcrt If society’s collective rate of tline preference is simply
the aggregate of the Individual fates (a controversial assumption), the .
required rate is simply given by the real (adjusted for Inflation) rate of
return on long-term government bonds.
However, in practice it is usually admissible to select a central ‘best
estimate' of r, and then vary this systematically in a sensitivity analysis to
determine the impact on the study conclusions. The criteria to use in
selecting a central r and a range for sensitivity analysis are that these
should:

(a) be consistent with economic theory (2 per cent to 10 per cent);
(b) include (bracket) any government recommended rates (5 per cent,
7 per cent, 10 per cent);
(c) include (bracket) rates that have been used in other published
studies to which you might wish to compare results (3 per cent to
10 per cent);
(d) be consistent with ‘current practice’ (for example, 5 per cent has
been used recently In papers published in the New England
Journal of Medicine).

1. Useful life and resale value (n and S)
’POr.tan'^ make 8 distinct,°n between die physical life of a piece of
equipment and its useful clinical life. The latter is highly dependent on
Lsina diff6' tcha1"8e1Obvious|y °*'e can undertake a sensitivity analysis
us ng different values for n. but in general it is best to be conservative and
assume short lives (say, around five years) for clinical equipment.

3. How to handle inflation
If it is assumed that nil the items of cost in the programme will inflate al the
same rate and that this will be the same rate as inflation in general, there
are two equivalent choices:
(a) Inflate all future costs by this predicted inflation rate and then use a
larger discount rate that allows for the effect of general inflation
(the inflation adjusted discount rate*), or

2. Choice ofthe discount rate (r)

(a) r - the real rate of return (to society) forgone in the private sector.

51

* Calculation of inflation adjusted discount rate: if (lie real discount rate is 5 per cent
and general inflation is R per cent, then the inflation adjusted r — (1.05)(I .OR) — 1.134
or 13.4 per cent.

52

Cost analysis
(b) Do not inflate any future costs (i.e., use constant dollars) and use a
smaller discount rate that docs not allow for inflation (the real
discount rale).

Method (b) is the simpler and preferred approach.
If it is assumed that different items of cost in the programme will inflate
at different rates, there are also two equivalent choices:

(a) Inflate all future costs by their particular predicted inflation rates
and then use a larger discount rate that allows for the effect of
general inflation (the inflation adjusted discount rate*), or
(b) Do not inflate any future costs (i.e., use constant dollars) and use a
smaller discount rale that does not allow for inflation (the real
discount rate), but adjust the discount rate for each item to account
for the differential inflation rale between this item and the ‘general’
rale of inflation, e.g., if general inflation is 8 per cent, this item is
expected to inflate 10 per cent, and the real r is equal to 4 per cent,
then r adjusted for this item is

I 08
r — 1.04 x —- - | .021, i.e. 2.1 per cent.

Cost analysis

The first task is to determine a unit of output for those departments that
direcdy serve patients. We will be determining a cost per unit of output,
and multiplying this cost by the usage of each patient to determine the
cost per patient. Thus, the unit of output must be as homogeneous as
possible with respect to cost, and yet be available in the data for each
patient. We have selected a patient-day as the unit of output of the NICU,
and a DBS unit for the laboratory. A DBS (Dominion Bureau of
Statistics) unit is a standard laboratory work unit used in Canada; each
lab test is assigned a predetermined number of DBS units according to
the amount of work needed to perform the test.
An allocation basis must be determined for each overhead department.
For example, square feet of floor space hds been selected for house­
keeping. This means that housekeeping costs will be allocated to depart­
ments receiving housekeeping services in proportion to the square
footage of floor space in the department. Similarly, paid hours has been
selected as the allocation basis for administration costs, and pounds of
laundry for the laundry costs.
The data for this simplified example are given in Table 4.1. The
calculations, as performed by the different methods, are given in
Tables 4.2 to 4.7.

Method (b) is again the preferred approach. In general, however, most
studies perform the whole analysis in constant price terms and use a
single discount rale.

4.3. ALLOCATION OF OVERHEAD COSTS:
EXAMPLE
The following example demonstrates the various methods of handling
overhead costs discussed in Section 4.1.2 (4), p. 43. Suppose we wish Io
determine the cost pf neonatal intensive care (NIC) for a specific group of
patients. For each patient we have data on the length of stay in the
neonatal intensive care unit (NICU) and data on the number and type of
laboratory tests performed. For simplicity, let us assume that these were
the only services received by the patients—that is, the patients had no
operations, no radiological or nuclear medicine investigations, no social
work, etc. Furthermore, let us assume that there are only three overhead
departments tiiat serve the laboratory and the NICU: administration,
housekeeping and laundry. (In principle it would be possible to consider
other overhead departments, like plant operations and maintenance,
bioengineering, and materials management.)
• Sccfooinoicoiip. 52.

<14

I

o © o

_gg I

o © o

o o o o
o
o
_
o
o
o



© © ©
© © ©

apg 8

§ §

00

o o
o o

&

© © © o
O
00 co
CO
tn
i-

© © o
o © ©
© 00 Ol
tn
VO

o

2 8

ss §

o o
o o
o © oo
o o o o
8 tn 8 o
m

11

I

o o
o
©o
o
o oo

§

r-4

O
o
o
o
o
m

Table 4.2. Method I—ignore overhead

I

Lab cost/DBS unit - $4 000 000/8 000 000 - S0.50/DBS Unit
NICU cost/pt-day - $500 000/5000 - $ 100/pt-day

o
o

§_ o©

§ ©©

Table 4.3. Method 2—direct allocation ofoverhead
(Note: Allocation denominator — sum of Tinal’ department.)

CM

I

I

3

i

I
il

a

Ia

Cost analysis

Lab cost - direct cost + lab’s share of admin 4- lab’s share of
housekeeping 4- lab’s share of laundry

.9

.*3

5
&

g

2U I v>

■a

© —•

J &

I

o o
o o
o o
o m
o
o
oo

5
O

*o

o o o o
o o o o
o o o o
o ooo
o o o o

o o o
o o o
o o o
oo o
o
o o
O m
o
tn

§
o »n tn C'l §
T-< T-< o

s
3a

§
§Jo

©
©
©
©
©
©
in

3o

o
O
'’T

s

£

Lab cost/DBS unit - 4 350 000/8 000 00(1 - $0.54/DBS unit
NICU cost ™ direct cost + share of admin 4_ share of housekeeping +
share of laundry

3
"’o.
Q.
<71

-500 000 +

9
O8

1

o

EJ

- 4 000 000 4- 250 000 + 75 000 + 25 000 - 4 350 000

eg
o

£3
a
u

S

D-

73
o

E?

ao

*2

u

o

8 000
(1 500 000) +
600 000

- 500 000 + 50 000 + 20 000 + 75 000 + $645 000

If- .
g
11 3 g 5 3 ijlil
st
£
8 r-

(2 000 000) +

3
.3

S3

■g -S

50 000
2 000 000

75 000
(1 300 000)
1 300 000

*0

u M).

250 000
30 000
(2 000 000) +
(I 500 000) +
2 000 000
600 000

25 000
(1 300 000)
I 300 000

g
a
m
Q
.S
a
a

® o

I

-4 000 000 +

B

o

8 §

a “
5 3

to

o

NICU cost/pt-day - 645 000/5000 - $129/pt-day
-

.a
Q

55

56

-

i

>ie 4. Methods—Step down allocation of overhead
>te: Allocation
denominator
------------------- •' “ sum of remaining departments in the step down sequence.)
Admin
•t cost

2 000000

ire sxdmin

ate

ate

HK
1 500 000

3
TjT- 214 286

2 000 000 -

C

1 714 286-

2
28

8
766

undry

Laundry

Other

Lab

1 300 000

10 200 000

4 000 000

NICU
500 000

Other

7

30 500 000

50m

2S
28

178 571

158
— - 353 599

30
766

67 139

8
766

17 904

1 460 761---- Hl „ 123 445
1420 -----3

25
1420

25 718

75
1420

77 153

1^5 ^i-234 446

4 271 428

630 771

34 206 472

50m

+ 8 000 000

5 000

Other

J

30 500 000

50m

142 857

.

17 904

214286

OS

35 714

28

17
— - 1 214 286

Xo

562
— - 1 257 740
/oo

1200

ost

10 891330

nit

S0.53/DBS unit

S126.15/pt-dav

)ir \5 Method 4—Step down with iterations
->ti Allocation denominator — sum of all departments except the one being allocated.)
Admin

HK

Laundry

Other

I ab

1 500 000

1 300 000

10 200 000

4 000 000

NICU

it
tcost

at

dmin

ueHK
in

2 000 000

mndry

2 000 000 -

30
796
0
1500

Jtals

3
28

64 609

0

80
1500

64 609

2
214 286
28
8
1 714 286 —
796
77 871
77 871

142 857
\1129

- 214 286

1g -

120
1460 086-—1500

0

116
807

Mw7

2S _

28
30
796
25
1500 “

10 871 366

178 571
64 609
24 335

500 000

0.5
28
8
796
75
1500

4 267 515

35714

- 1214 286

17229

562
,,
796 - 1210 338

73 004

1200
--1 *68 069

625 947

34 092 693

or
te z-xdmin

te
e—jndry

64 609 -

30
796
0
1500

3
28

3 196

0

80
1500

2
28
8
84 793 796
6 922

292

3
28
158
852
796
5467--Hl
1500
4615 *

6 922
16 831
437

Z5

28
30
796
25
1500

5 769

3 196
91

0J
28
8
796
75
1500

1 154

852
273

17
28

562
796
1200
1500 ’

39 227
59 866

4 374

50m

«h

o
r-.

£
3

in

? I a

»n

P
CO

8

O\

I
I
I
t- I CO CM I O O IO

Table 4.6. Method 5—simultaneous allocation (reciprocal method)
(Note: Allocution denominator ~ sum of all departments).

s

s

3



i

Admin

2
30
C, -2000 000 + — C, +----- C2
30
800

O

HK

3
4
80
C2 — 1 500 000 + — C, +---- C2 + ------ C 3
30
800
1500

Laundry

2
8
Cj- I 300 000 + — Ci +--- C2
30
800

Lab

C,-40()0 000 + —C| +---- C1 + - -- Cj

NICU

C,-

»n IP 8 |»n

8
CM
8
«o

g



tn

'O

i

I

313

I
P

I

»n

8

"l^ll

+

i

3la«|s?

s

<M

s i i

£
8 10 JO

8
>o

o

£ s I
i §

oo

00

jq

»o
CM

O


CQ*

oo

•n
o
v>

CM

i



i

I

i

4-

30
800

2.5
30

25
1500

0.5
8
75
500 000 + — C, +----- C2 + ---- Ci
30
800
1500



jn

«n

CM

2

tn

jq

oj

5
o

o

oo
o

00

o

I

I

o

I

I

i

IS2|s§|§

5

8
CM
I

o

SI

2

i

18

l

*1

t
o
Qa

r*

i
CM

CM

o

t



O
tM

<M

p

- I 300 000

0.5

8

30

800

= 4 000 000

75

-c3 + c5
1500

500 000

Therefore, the Cost/unit of output is:

Lab:

J gg JI 1 ag |S
u

+ c3

C, — 2 215 531
C2— 1 808 772
C,- I 465 790
C4-4 276 886
C5- 628 303

”|2

b
3

/

- 1 500 000

The solution of this set of equations is:

o|S

“Is? o|§

1o
i
z

800 ‘

“ — Ci“— Ci“

I

'

30

80
--- C3
1500

2.5
30
25
------ C,---------C2- — C3 + C4
1500
30
800

I

I
xt

796

2
8
------ c,-------- Cj
30
800

a

i

CM I

.3

“ 2 000 000

“ —Cl+—C2“

I

"I8oo|8

CM

28
30
—c«-— C>
30
800

o
o

00

tn

o

tn

tn

- a

u

2

TZ

Z

F

L.

$4 276 886/8 000 000 - $0.53/1)118 iinil

NICU: $ 628 303/5 000

I

!

Jcod
i

- $125.66/pt-day

Cost analysis
Horngren, C.T. (1982). Cost accounting: a managerial cmnhasis (5lli
ciln). Prentice 1 lull, Englewood ClilTs, N. J.
Hull R Hirsh, J., Sackett, D. L., and Stoddart, G.L. (1982). Cost­
effect veness of primary and secondary prevention of fatal pulmonary
embolism in high-risk surgical patients. Can. Med. Assoc. J. 127,

Cost analysis
1 able 4.7. Method 6—patieiit-day allocation of overhead
This is the simple method described in the footnote of page 27 of
Chapter 3 and on page 46 of Chapter 4. It may be useful in some cases.

Kaplan, R S. (1973). Variable and self-service costs in reciprocal allocu­
tion models. Vie Accounting Review XLV111,738-48.
LeiVin:.HkM (1975)- Cos<-effectiveness analysis'in evaluation research
In M. Guttentag and E. L. Struening (eds) Handbook of evaluation
research. Vol. 2 pp. 89-122. Sage, London.
Lowspn, K. V., Drummond, M. F., and Bishop, J. M. (1981). Costing new
services: Long-term domiciliary oxygen dierapy. Lancet I, 1146-9
Neuhauser D and Lewicki, A. M. (1975). What do we gain from the sixth
stool guaiac? N. Engl. J. Med. 293(5), 226-8.
Richardson, A. W and Gafni, A. (1983). Treatment of capital costs in
evaluaung health care programmes. Cost and Management Nov-Dec*
26-30.

Laboratory costs would be charged without overhead: S0.50/DBS unit.
NICU costs would be the direct costs of $500 000 plus a share of all
relevant other departments (2.0m + 1.5m + 1.3m — 4.8m) in
proportion to patient-days (5 000/500 000 where the denominator is
total annual hospital patient-days). Thus,

NICU cost - $500 000 + $4 800 000 (5 000/500 000) - $548 000.
NICU cost/pl-day - $548 000/5 000 - $110/pt-day.

ANNEX4.1. METHODS OF MEASURING AND
VALUING CAPITAL COSTS

We are indebted to Morris Barer of the University of British Columbia
for producing these examples, which should clarify the treatment of
capital costs.
As a first note, we need to distinguish two classes of‘caphal ’-hnid and
equipment, lliis is an important consideration, because in costing
exercises we assume land docs not depreciate, while of course capital
equipment does. You can think of there being a continuum along which
materials and supplies ‘depreciate’ or are used up instantaneously and so
are costed fully in the year of use; capital equipment depreciates more
slowly, and may be handled in a variety of ways; land does not depreciate

REFERENCES
Boyle, M. H.» Torrance, G. W., Horwood, S. P., Sinclair, J. C. (1982). A
cost analysis of providing neonatal intensive care to 500-1499 gram
birth weight infants. Research Report #51, Programme for Quantita­
tive Studies in Economics and Population, McMaster University,
Hamilton, Canada.
Clements, R. M. (1974). The Canadian hospital accounting manual
supplement. Livingston Printing, Toronto.

I-

i

L
I

As a second note, recall that ‘capifal equipment costs have three
fi?mP?.n.entS“depreciation’ °PPorlunity cost, and actual operating costs.
We will ignore the last of these here.
First consider equipment, and let us use an example of a machine
costing $200 000 that, at the end of 5 years, has re-sale value of $20 000.
Assume straight-line depreciation and a discount rale of 4 per cent. There
are, then, four approaches to costing:

(i) one can assume all costs accrue at lime O.This amounts to treating
the equipment as one would less durable materials and supplies;

I.

Cost analysis
Cost analysis



I

lime



0

I'
Depreciation
Undepreciated balance at
beginning of period
Opportunity cost

200 000

I

2

3

4

5

0

0

0

0

(20 000)

0
0

0
0

0
0

0
0

0
0

Dep’n. 4- opp cost
Present value (PV)

200 000 0 0 0 0
200 000 0 0 0 0
Net present value (NPV) of equipment cost-$183 561

lime

(20 000)
(16 439)

Depreciation
Undepreciated balance
at beginning of period

I

2

36 000

36 000

3

4

5

36 000 36 000 36 000

Opportunity cost

200 000 164 000 128 000 92 000 56 000
8 000
5 120. j 680 2 240
6 560

Dep’n. 4- opp cost
PV

44 000
42 308

42 560
39 349

41 120 39 680 38 240
36 556 33 919 31 430

NPV of equipment cost — $ 183 562

Ahernatively, but equivalently, one can treat the machine as

wl ch lhenT y ^C,pr.ecl"'i1n8’
for "'<= S20 000 resale value,
wliicli then is maintained through the 5 years:

Time

0

Depreciation
Undepreciated balance
at beginning of period
Opportunity cost

Dep’n. 4- opp cost

1

2

3

4

5

180 000 -

20 000
800
180 000
180 000

NPV of equipment cost - $183 562

800
769

20 000 20 000 20 000
800
800
800
800
740

800
711

800
684

20 000
800

Time

I

2

40 000

40 000

3

4

5

Depreciation
Undepreciated balance
at beginning of period
Opportunity cost

200 000 160 000 120 000 80 000 40 000
8 000
6 400
4 800 3 200 I 600

Dep’n. + opp cost
PV

48 000
46 154

46 400
42 899

40 000 40 000 20 000

44 800 43 200 21 600
39 827 36 928 17 754

NPV of equipment cost — $ 183 562

800
658

(n) One can compute depreciation and opportunity costs separately

(iii) One can compute an equivalent annual cost. This may be useful in
a situation where other operating costs are the same each year,
making necessary the comparison of only a single year of cost data
for each alternative in the economic evaluation:

NPV
t

(Where AF5 is the annuity factor
for 5 years at an interest rate of 4 per
cent. See Table 2 in Annex 4.2)

183 562 -li- 4.4518 - E-$41 233
point in time. Hence, the higher the rate of depreciation, the lower
the opportunity cost, all else equal. Again, one has the choice of
but dmg the $20 000 resale in at the end, or just depreciating less
of (he machine. It works out the same:

In other words, an equal stream of costs amounting to $41 233 in
each of the five years of the program has a present value
equivalent to any of the unequal cost streams in (i) or (ii) above.
Note, therefore, that the equivalent annual cost embodies both
depreciation and opportunity cost.

68
69

c

o

o
5
co

ts>
03
M

GJ
GH

O\
GJ <

I I

5
O
TJ

n

o

3

co
co

Q.

O

z
•3
<

o“ ca

? ST £-8
2 <s.4 3
+

I
GJ
IA
O\

g’

2. □' 2. =r.

n

8 2-2."

eg

2
co
.a

w

o
•o

00

i

11
<0

S’

2

c

•5

03

3

3.

oj 00

CA

O\ o
o

CO

NJ C

n

2

2

m

o
00 o
o o
o c
o a

I

^J 00

GJ O
50 c
Q\ C

00

©
o
©

I

DJ

NJ C

o
00 o
o o
o o
o o

O\ 00

00 o

00

GJ C
00 Q

o\ oo

OO

co c

00
(0

LA C

'J c
<-n C

I

Q.

ri
= 0

Q

H -6- s S.

-2 3gI x2| »S2
co ~ ~= Z< S'g
£ 8£ =n
o
- o" £ s SO
CO

5co

3

2.

w
o co S
u S
H So »
•Q e « s
e.
o 5
n ©
2? o< »=-•
P n « Z
2. §
c 5
= °
□§
n 3 £ 9

g□.

“ S’
co o


o o
o a
o o

3 3

co c

2 =*

NJ

O

0

CO

54 **

o

~ 3—» CO

o s ja

o

o

03

co

£
s- £ 2 Z
cr

<0

I

Q

5S- ?
g5 I"

§ H
8
C0

GJ

NJ

o
o o
o o
o o

co
n £•=•
o
2 % 5 _ 43
2

© CO

I

n

WOW

c Q.

©

<0 g- g
cT

I
H o g. Z §

G CO

NJ

M

— c
— c

Tco

s„ “ - p =

c E.

a
oo o
o o
o o
o o

5. S O
-• coE
C-

2

S’

NJ

I

n n
w O
CD

g = COS=• >
E g 2
° - s.
3 a- S
“ “7 co
C 3 '
E
~
3 2. o
5 ZZ
8 S’
£, S.

£

E.

cn

- 2
2.

do _
g.ix
2 o’

O

w

o
o

GO
(0

O

*-

n* O

CO

Spj

o

Annex 4.2. Discount Table 1
Present value of 31
N
1%

2%

3%

4%

5%

6%

7%

8%

9%

10%

11%

12%

0.9174

0.9091
0^264

0.9009

0.8929
0.7972

1

0.9901

0.9804

0.9709

0.9615

0.9524

0.9434

0.9346

2

0.9803

0.9070

0.8734
0.8163
0.7629

5

0.9706
0.9610
0.9515

0.9426
0.9151
0.8885

0.9246

3

0.9612
0.9423
0.9238
0.9057

0.9259
0.8573
0.7938
0.7350

03626

6
7

0.9420
0.9327

03880
03706
0.8535

03375
03131
0.7894

0.8368

0.7664

03203

03963

03043

03874
03787

4

8

0.9235

9

0.9143
0.9053

10

0.4665
0.4241

0.4339

0.4039

03909

03855

03522

03606
03220

03505
03186
02897

02858

0.6806

0.7050
03651
0.6274

03663
03227

03302
03835

0.7307

0.7462
0.7107
0.6768

0.7441

0.7026
03756

0.6446
0.6139

03919
03584

0.72240.7014

0.6496
03246

03847

03268
0.4970

0.4751

0.4289

03568

0.4440

03971

03810
03611

03006
03775

03303

0.4688

03051

0.4423

0.4150
03878

03677
03405

03553

0.4810

0.4173

03624

03152

03936
03714

03387

02919

03166
02959
02765
02584

02415
0.2257

0.1987

0.1637

0.1339

0.1502

11
12
13
14

03700

0.7885
0.7730
0.7579

15

03613

0.7430

03419

0.4556

03066
0.4523

0.7130

0.4632

03718
0.4972

03346
0.4817

0.7473

03083

03921
03194

03645
03132

0.7835

03403

0.8696
0.7561
0.6575

03963

0.8638

0.8227

03002

0.8772
0.7695
03750

0.8850
0.7831
0.7118 0.6931
0.6355' 03133
03674 03428

0.8890

03820

15%

0.8116
0.7513. 0.7312
0.6830 0.6587
0.6209 03935

0.8548
0.8219

03439

14%

0.8417
0.7722
0.7084
0.6499

0.8900
0.8396
0.7921

0.7903
0.7599

13%

03470
03019
0.4604
0.4224

03875
03555

03173

0.4803
0.4251
03762

03506

0.4323
03759
03269

02946

03075
0^697

02843
02472

02607
02307
02042
0.1807

02366
02076
0.1821
0.1597

0.1599

0.1401

02149
0.1869
0.1625
0.1413
0.1229

03329

03996

02992
02745

02633

02575
02320

02394

02090

02875
02567
02292
02046
0.1827

02703

02519
02311

02176
0.1978

02502
02317

0.2120
0.1945

02145

0.1784

0.1799
0.1635
0.1486

0.1883
0.1696
0.1528
0.1377
0.1240

0.1631
0.1456
0.1300
0.1161
0.1037

0.1117
0.1007
0.0907

0.0926 0.0768
0.0826 . 0.0680
0.0738 0.0601

0.0491

03262

16

03528

0.7284

03232

03339

0.4581

17

03444

0.7142

03134

0.4363

18
19
20

0.8360

0.4936

03277

0.7002
0.6864

03050
03874
03703

0.4746

0.8195

0.6730

03537

0.4564

0.4155
0.3957
03769

03118

0.6598
0.6468

23

0.8114
0.8034
0.7954

0.6342

0.5375
03219
0.5067

0.4388
0.4220
0.4057

03589
0.3418
0.3256

0.2942
0.2775
02618

02109

0.1703

0.1378

0.1351
0.1228
0.1117

24

0.7876

0.6217

0.4919

0.3901

0.3101

0.1971

2S

0.2470

0.1577

0.1264

(I.77VH

0.1015

O.ftOWS

0.0817

O.37S 1

0.2953

0.0532

0.0431

0.0402 >
0.034*9

(>.4776

0.0659

0.2.13<»

<> 1 «42

O.l 4 AO

O.l 160

0.0923

O.O73A

(J.OSftN

0.04 71

O.O37M

0.0304

21
22

03503
03305

0.1415
0.1252
0.1108
0.0981
0.0868

0.1229

0.0829
0.0728

0.1069
0.0929
0.0808
0.0703
0.0611

0.0638
0.0560

0.0462

3.1078
0.0946

0.0531

>

•<
S2.
35’

Annualization Factors
1% '2%

0

1% 8% ^9%
0 935 0.926

0917 0.909 0.901

0893

0 MS 0877 0.870 0.862 0.855 0 847 0 840 0.833

1.808

1.783

1.759 1.736 1.713

1.690

1.668

1.647 1.626 1.605 1.585 1.566 1.547 1.528

2 775 2.723 2 673 2.624

2.577

2.531 2437 2 444

2 402 2 361

2.322 2.283 2.246 2210 2.174 2.140 2.106

1

0.990

0 980

0 971

0 962 0 952 0 943

2

1 970

1 942

1.913

1 M6

3

2 941

2884

2 82'

1 859 1 833

2-4 3 902 3 808 3717 3 630 3 546 3 465 3.387 3.312 3.240 3.170 3.102 3 037 2.974 2.914 2.855 2.798 2.743 2.690 2.639 2.589
3
4 853

4 713 4 580

4 452 4.329 4 212 4 100

3 993

3890 3.791 3 696

3 605

3517

3 433 3.352 3.274 3.199 3.127 3.058 2.991

S! 6- 5.795 5 601 5417 5 242 5 076 4 917 4.767 4.623 4 486 4.355 4.231 4 111 3 998 3.889 3.784 3.685 3589 3498 3.410 3.326
6 002 5 786 5 582, 5 389 5.206

5 033 4 868 4 712 4 564 4 423 4.288 4.160 4.039 3 922 3812 3.706 3.605

6.728

6 472 6230

8

7.652

7.325

7.020 6 733 6 463 6 210 5 971 5 747 5 535 5 335 5 146 4 968 4.799 4 639 4 487 4.344 4 207 4 078 3.954 3.837

9

8 566

8 162

7 876

7 435 7.108 6 802 6.515 6 247

5 995 5 759 5 537

8 530

8 111 7 722 7 360 7 024

6 710

6418 6 145 5M9 5 650 5 426 5 216 5.019 4 833 4 659 4.494 4 339 4 192

9 253

8 760 8 306 7 M7

7 499

7.139

6 805 6 495 6 207

5 938

5 687

5 453 5 234 5 029 4 836 4 656 4.486 4.327

12 11255 10 575 9 954

9 385 8 863 8 384

7 943 7 536

6814 6 492

6 194

5918

5 660 5421 5 197 4 988 4.793 4.611 4 439

13 12.134 11.348 10 635 9.986 9 394 8 853 8 358 7.904

7487 7.103 6.750

6 424

6.122 5.842 5583 5.342 5.118 4.910 4.715 4533

14 13004 12106 11.296 10563 9 899 9 295 8 745 8 244

7.786 7.367 6 982 6.628 6 302 6.002 5.724 5 468 5.229 5.008 4.802 4.611

co

i
LU

a.

5 328 5.132 4 946 4 772 4.607 4 451 ’4.303 4.163 4 031

X

10 pW71

O'

10368 9 787

7.161

15 13865 12 849 11 933 11 118 10 380 9.712 9 108 8 559 8 061 7606 7.191

6811

6 462 6.142 5.847 5.575 5.324 5 092 4 876 4.675

16 14.718 13 578 12561 11 652 10 838 10.106 9 447 8 851

8 313 7.824 7.379 6 974

17 15 562 14 292 13.166 12166 11.274 10477 9 763 9.122

8 544 8.022 7.549

7.120 6.729

18 16 398 14 992 13.754 12 659 11 690 10 828 10059 9 372

8.756 8 201 7.702

7.250 6 840 6.467 6.128 5.818 5.534 5.273 5.033 4 812

6604

6.265 5.954 5.668 5 405 5.162 4.938 4.73Q
6 373 6.047 5.749 5475

5.222 4.990 4.775

19 J17.226 15 678 14.324 13 134 12 085 11.158 10.336 9.604 8 950 8 365 7 839 7.366 6 938 6 550 6.198 5.877 5584 5.316 5.070 4 843
20 18 046 16351 14 877 13 5 90 12.46211.470 10594 9.818

9.129 8.514 7 963 7.469

21 18 857 17011 15415 14 029 12821 11 764 10836 10017 9 292 8 649 8 075

7 025 6.623 6.259 5.929 5.628 5.353 5.101 4.870

7.562 7.102 6 687 6.312 5.973 5.665 5.384 5.127 4.89.1

22 19.660 17 658 15 937 14 451 13.163 1 2 042 11061 10.201 9 442 8.772 8.176 7.645 7.170 6.743 6.539 6.011 5.696 5.410 5.149 4.909
23 20 456 18 292 16444 14 857 13489 12303 11272 10.371 9.580 8. M3 8 266 7.718 7.230 6.792 6.399 6.044 5.723 5.432 5.167 4.925

241

7.784

25 22 023 19 523 17413 15622 14 094 12.7M 11.654 10.675 9.823 9.077 8 422

7 843 7.330 6.873 6464 6.097 5.766 5.467 5.195 4 948

21.243 18914 16936 15247 13 799 12 550 11.469 10529 9.707 8 985 8 348

7.283 6.835 6.434 6.073 5.746 5.451 5.182 4.937

26 22.795 20 121 17.877 15 9M 14 375 13 003 11.826 10810 9 929 9.161 8.488 7.896 7.372 6.906 6.491 6.118 5.783 5.480 5.206 4.956

27 23 560 20 707 18.327 16 330 14 643 13 211 11.987 10.935 10 027 9237 8 548 7.943 7409 6.935 6.514 6.136 5.798 5.492 5.215 4.964
28 24 316 21 281 18 764 16 663 14 898 13406 12.137 11.051 10.116 9 307 8.602 7.984 7 441 6 961 6.534 6.152 5810 5.502 5.223 4 970

29 25 066 21 844 19 IM 16 984 15 141 13 591 12 278 11.158 10 198 9.370 8.650 8 022 7 470 6.983 6.551 6.166 5 820 5.510 5.229 4.975
30 25 808 22.396 19600 17 292 15 372 13 765 12409 11 258 10 274 9 427 8 694 8 055 7 496 7.003 6.566 6.177 5.829 5517 5.235 4 979

136

THE LANCET, MAY 23,198 ]

National Health Service
COSTING NEW SERVICES: LONG-TERM
DOMICILIARY OXYGEN THERAPY
Karin V. Lowson

Operational Research Unity
West Midlands Regional Health Authority, U.K.
M. F. Drummond

J. M. Bishop

Health Services Management Centre and Department of Medicine,
University of Birmingham

Summary

An economic appraisal ofdifferent methods
of long-term treatment with oxygen in the
home has shown that the oxygen concentrator is the cheapest
and most convenient one. The only method at present
generally available in the National Health Service, the use of
small oxygen cylinders, is the most expensive and least
convenient ofthose studied. There is need for a more flexible
administrative system which will allow patients and the
Health Service to benefit from the economies which are
offered by technical advances.

INTRODUCTION



Research over the. past 15 years1’6 has indicated that
oxygen given for 12-24 h a day to patients with chronic
bronchitis and chronic hypoxaemia reduces pulmonary
arterial pressure and red cell mass. More recently, clinical
trials in the U.S.A.7 and a Medical Research Council
(M.R.C.) multicentre controlled trial8 have confirmed that
long-term domiciliary oxygen therapy reduces mortality and
improves general quality of life. The M.R.C. trial employed
three methods of delivery: cylinder oxygen, liquid oxygen,
and oxygen from concentrators, and all seemed to be equally
effective.
~ ’
Although medical research inevitably leads to demands for
new services, the resource consequences of new treatments
are typically underexplored. In the case ofoxygen therapy the
resource consequences take .on extra significance, since the



4. Almy T. The Role ofthe Primiry Phyjjcian in the Health Care Induitry. NEngHMtd
1981; 304: 225-28.
5. Flexner A. Medical Education in the U.S. & Canada. Bui! Carmgit Fdn Advanc
Tracking 1910; 4: Bouon: Merrymont Pre».
6. Van Steenwyck J, Fink R. Implications of Hawaii's Mandatory Health Insurance LawAnnual Meeting, American Public Health Association, 1979. G.LS. Associates,
1616 Walnut Street, Philadelphia.
7. Enthoven A. The Competition Strategy: Status and Prospeos. N Engl J Med 1981;

]
I

I

PROFESSOR HATCHER: REFERENCES—continued

8. Royal Commiuion on Health Services (chairman. Chief Justice Emmett Hall)
(Canada), vol. I. Queen’s Printer, Ottawa, Ontario, 1981.
9. Hall E. Canada's National-Provincial Health Program for the 1980’s: A commitment
i°rgQ new’1' ReP°n 10 Ministers of Health. Saskatchewan: Universal Bindery Ltd,

'
|
i

,

10. Hatcher GH. Canadian utilization review programs: their implications for contain­
ment of health care costs; tables A-12 and A-l 3. In: Policies for the containment of
health care costs and expenditures. Bethesda: Fogarty International Center for
Advanced Study in the Health Sciences, 1978. DHEW Publication (NIH 78-184).
U.S. Government Printing Offlce. Washington, D.C.
11. Nova Scotia Health Services and Insurance Commission Annual Report 1979. Tables 6
and 8. P.O. Box 760, Halifax, N.S.
12. Personal communication from Dr J. R. Carlisle, Assistant Registrar, College of
Physicians and Surgeons of Ontario, Sept. 3, 1980.
13. Graduate Medical Education National Advisory Committee, September, 1980. Report
to the Secretary, Department of Health and Human Services. (7 Volumes) U.S.
Government Printing Office, Washington, D.C. 20402. U.S.A. 1981.

1147

THE LANCET, MAY 23,1981

three treatment methods have similar medical attributes, yet
vary considerably in cost. Furthermore, detailed examination
ofthe relative costs ofthe treatments reveals firstly that one of
the methods (oxygen concentrators) is more capital intensive,
and is therefore sensitive, in terms of cost per patient, to the
size of the patient population; and secondly that only one of
the treatment methods examined in the trial is currently
funded on a regular basis within the National Health Service
(N.H.S.).
This paper describes the use of economic appraisal9 in
examining the resource consequences of providing long-term
oxygen therapy. The costing exercise also attempts to
illustrate some of the problems facing those researchers and
clinicians who wish to examine the resource consequences of
introducing new services.
METHODS OF OXYGEN ADMINISTRATION

Cylinder oxygen comes in large (120 cu ft, 3400 litres) or
small (48 cu ft, 1360 litres) containers. If large cylinders are
used for 15 h per day, the patient consumes ’A-1 cylinder of
oxygen daily, necessitating approximately one delivery of 6
cylinders a week. Small cylinders are supplied and delivered
by the local pharmacist; 15 cylinders are used per week,
necessitating approximately 2 deliveries per week. The
cylinders have to be stored, for example, in a cupboard or the
hall, and the oxygen is piped through tubing to, say, the
lounge and bedroom.
Liquid oxygen provides a large volume of gaseous oxygen
from a small storage space. Each patient has a reservoir
weighing 32 kg when full which contains the equivalent of
13 800 litres of gaseous oxygen. This reservoir, which is
refilled twice weekly from a delivery service ofliquid oxygen,
stands in a corner of the bedroom or living room; oxygen
evaporates through coils in the top ofthe reservoir and is then
fed via the flow meter and plastic tubing to the patient’s nasal
prongs. Most patients also receive a patient-carried liquid
oxygen reservoir (Union Carbide ‘Walker’ System) weighing
4 • 5 kg when full ofoxygen (1026 equivalent litres ofgaseous
oxygen). The portable container can be refilled with liquid
oxygen from the larger reservoir by the patient.
The oxygen concentrator10 is an electrically driven machine
which separates oxygen from the other gases in the air. The
concentrator output is 95% oxygen and the machine
incorporates a small storage cylinder. Concentrators, once
installed into, for example, a corner of a bedroom or an
outhouse from which the oxygen is piped as before, are easy
for the patient to use, compact, quiet, easy to maintain, and
reliable.
COSTING METHODOLOGY

The costs considered here are those falling on the N.H.S. and
patients as a result ofthe incremental resource use in providing long­
term oxygen therapy. Other costs incurred in treating patients with
chronic bronchitis regardless of the method of oxygen provision,
such as the cost of drugs and visits by the general practitioner, are

number of patients served (such as the provision of a workshop to
maintain concentrators) and those which vary with the patient
population (such as the electricity required to run the
concentrators). This distinction becomes important in this study
because concentrators have a higher element of fixed costs than the
other two treatment methods.
Secondly, assumptions have to be made about the length oflife of
the capital cost items, such as buildings, workshop facilities, and
equipment. The assumptions made here of30 years, 10 years, and 5
years, respectively, are conservative but the assumption of longer
lives does not change the results very much.
Thirdly, it is necessary to allow for the fact that the alternatives
with a larger capital component require a larger proponion of the
resource outlays earlier in the life of the project. In industry the
interest a company would have to pay on the capital would be
reflected in investment appraisals. Although the N.H.S. does not
pay for its capital in the same way, one could still argue that as a
community we are not indifferent to the timing of resource
commitments; that is, we prefer to incur costs later rather than
sooner. The most widely accepted way ofincorporating this notion
into public sector appraisals is to apply a public sector discount rate
to costs and benefits occuring in future years.9 Since most of the
costs of oxygen therapy can be expressed on an annual basis, the
discount rate is used to conven the capital outlays to an annual
charge which reflects not only the actual sums involved but also the
fact that they occur sooner rather than later.
Several models of oxygen concentrator are commercially
available, so costs may vary. Firstly, the type of concentrator
determines not only its purchase price, but also the running costs,
because the more recent models are smaller and require less power.
Secondly, the maintenance costs vary according to whether this
service is provided by technicians operating from purpose-built
workshops, or technicians attached to an existing unit.
Thus, two methods of providing oxygen by concentrators have
been costed, the purchase of a new machine being assumed in both
cases:
Alternative A.—To maintain the concentrators effectively
hospitals or health authorities with very.few or no specialised
workshop facilities have to be provided with a workshop. This must
be equipped with furniture and tools and a vehicle in which to make
home visits, and it has to be staffed by two technicians to allow for
illness and holidays. If more than 60 concentrators are serviced,
three technicians and an extra vehicle may be required.

Alternative B.—Where hospital workshops are currently
operating below full capacity the additional cost of providing basic
servicing facilities for the concentrators through existing
workshops would not be very large. It would consist of the costs of
providing extra equipment, tools, furniture, and a vehicle; and only
one technician need be employed for servicing 30 concentrators
since cover could be provided by the existing staff. For up to 60
concentrators, two technicians are assumed to be employed, and
three, plus an extra vehicle for more than 60. Finally, only a
proportion of the running costs of the whole workshop need be
apportioned to the concentrator servicing.
Since the number of patients served is a key factor in determining
the relative costs of the treatment alternatives, the figures for total
cost and cost per patient are given below for varying population
sizes. Some of the assumptions made in deriving the costs are given
in the accompanying discussion. A background paper giving fuller
details of the cost calculations can be obtained from the authors.

not considered since there is no evidence that they differ greatly

among the three treatment methods. Furthermore, the question
being posed here is “what is the most efficient way to administer
long-term oxygen therapy?”, not “is treatment worthwhile perse?'
given the other priorities facing the N.H.S. This latter question
would also require estimates of the benefits to patients and to the
N.H.S. of improvements in morbidity and mortality resulting from
treatment.
Points to note in the cost calculations include, firstly, the
distinction between costs which are relatively fixed regardless ofthe

RESULTS

The set-up costs and running costs (tables I-IV) are based on
15 h of oxygen use per day at 1980 prices. Some of the set-up
costs are common to all methods, for example, the cost ofthe
piping; whilst others differ between methods, such as the cost
of the base for the liquid oxygen reservoir, the stands for the
cylinders, and the electricity sockets for. the concentrators.
The running costs differ between methods, for example, for

THE LANCET, MAY 23,1981

1148
9000-|

TABLE 1—COST PER ANNUM PER PATIENT FOR SMALL CYLINDERS
Amount (£)

Item

8000-

Running costs
780 cylinders at £3-78*
104 deliveries at £6-00+
Rental for sets and stands at £3-60 per month
Total running cost, including value added tax
Set up costs
Per annum installation cost discounted at 7%

2948
624
43
3615

i
i

400024
3640

_____________

Total cost

5000-

IS



^ 3000-

* Rental cost of cylinders includes cost of oxygen and the maintenance charge,
although there can be a separate cost for out-of-hours calls.
+Charge fixed per delivery regardless of number of cylinders per delivery, but
may vary according to geographical location or patient.

Jr..

5MU CrUMOtf

I
I

\
\
\ \

LAUGE CTUNOEK

2000-

IIQUIO OXrCFM
Conctntrotor (Alt.A)

1000-

Concentrotor (Alt. Bl

TABLE II—COST PER PATIENT PER ANNUM FOR LARGE CYLINDERS
Amount (jQ^

hem

Running com
365 cylinders at £3 • 78*
60 deliveries at £9-50+
2 call-out charges at £8-32
Total running cost, including value added tax
Set up com
Per annum installation cost discounted at 7%
Total cost
_____________

10

1380
570
16
2261

TABLE III—COST PER ANNUM PER PATIENT FOR LIQUID OXYGEN
Amount (£)

662
2000
153
2153
525
1486+

Total con

80

70

90

Too

Cost in £.
Discount rate of 7% assumed.
Kinks in graphs for concentrators are due to sharp increases in workshop
costs (see text) after 60 patients for alternative A, and after 30 and 60 patients
for alternative B.

•Rental cost of cylinders includes cost of oxygen and the maintanance charge.
fAs for table I.
+There is no general N.H.S. tariff for large cylinders, and costs are calculated
from typical costs incurred by hospitals.

Running com
48 refills at £13-80*
Set-up com
Costs of main and portable reservoir
Costs of installation, flow meters and fire extinguisher
Total set up costs
Discounted per annum set-up cost

40
50
60
No. of patients

Fig. 1—Cost per patient for all methods of providing oxygen.

24
2286

hem

30

20

* Includes delivery charge and cost of routine maintenance.
flndividual cost items at 1978 price levels; total cost adjusted to 1980 price
level.

costs for delivery of the oxygen, or for electricity for the
concentrators. Costs per patient and costs in relation to
number of patients are shown in figs. 1 and 2, respectively.
DISCUSSION

The figures indicate a cost advantage for concentrators for
all but small numbers of patients. However, all cost figures
are to some extent the product of the assumptions made in

their calculation. Also they must be interpreted in the light of
local circumstances! what is cost effective in one location may
not necessarily be cost effective in another. We discuss here
how the consideration of some economic issues may aid
decision making.

...
The choice of discount rate is often a contentious issue in
economic appraisals, but although concentrators and liquid
oxygen need larger resource outlays in the form of
equipment, the choice of discount rate has little effect on the
results. In practice the capital outlays on equipment are much
more likely to present financial problems.
The results show that size ofpatient population is one ofthe
key influences on the relative cost (per patient) of the various
treatment modes, largely because of the high set-up costs of
concentrators. Maintenance facilities and technicians must
be provided to service the concentrators irrespective of the
number of patients, and the level of such provision only
increases slightly with larger numbers. The least expensive
treatment modes are large cylinders for less than 8 patients,
concentrators (serviced by alternative B) for 8-13 patients,
concentrators (and irrespective of the method for servicing)
for any number above 13. As numbers ofpatients increase the
analysis of costs with increasing numbers ofpatients becomes
more complex—e.g. larger patient numbers may mean that-

TABLE 1V-COST PER ANNUM AND COST PER PATIENT FOR OXYGEN CONCENTRATOR (ALTERNATIVES A AND B)

Workshop set up cost*

No. of patients
1
5
10
20
50
100

A
3062
3062
3062
3062
3062
4428

B
1557
1557
1557
1557
1667
2996

‘Discounted at 7%.
j-Includes electricity and maintenance.

Workshop running
cost
A

12593
12593
12593
12593
12593
17583

B
7102
7102
7102
7102
12042
17032

Capital cost
(concentrators,
installation)*
A

287
1433
2865
5731
14328
28657

B
287
1433
2865
5731
14328
28657

Running costs
per concentrator+

A
127
635
1270
2540
6350
12700

B
127
635
1270
2540
6350
12700

Cost per patient

Total cost
A

16069
17723
19822
23927
43600
63368

B

A

B

9072
10726
12794.
16930
34386
61384

16069
3545
1982
1196
727
634

9072
2145
1279
846
688
614

THE LANCET, MA Y 23,1981

1149

701
6050-

4030-

2010-

10

20

30

40 50 6(
No. of patients

80

90

100

Fig. 2—Total costs of providing oxygen for all methods.
Cost in £x 1000.
Discount rate of 7% assumed.

technicians could be given geographical zones, thereby
reducing the amount of travelling between patients.
The most cost-effective method for a given health authority
may depend upon the facilities already available. Whereas the
costs of both cylinder and liquid oxygen include the costs of
maintenance and repair by the supplier, concentrators would
probably be maintained by health authority employees and
therefore the costs of this must also be included in the cost of
providing concentrators. In addition to the two methods of
maintenance which have been costed, rental schemes (see
below) or servicing by the concentrator manufacturer or the
provision ofa basic service by the use ofexisting facilities and
staff should be considered. These ma£ be most appropriate
for servicing small numbers ofconcentrators but one must be
sure that methods which appear “cheap” do not result in an
opportunity cost to the authority, in that staff and facilities
are diverted from other important activities. Even in such
circumstances a charge should be imputed to the use of
facilities even if they have “already been paid for” by the
authority.
All of the methods of treatment require very little
instruction to patients. We have no reason to believe that
training costs differ greatly between treatment methods or
that they are large for any method. If the oxygen systems are
regularly checked or maintained by hospital technicians who
are also in contact with the hospital doctors, patients gain
confidence m their oxygen system, and know that they have a
point of contact in case of difficulty.
Some companies are now offering rental or leasing
schemes. The costs, including maintenance and running
costs, vary somewhat, but are similar to those determined for
machines purchased and maintained. Although these
schemes are fairly new and untried, they seem to be worthy of
consideration when small numbers of patients are to be
treated, and when capital to purchase machines is not
available. They also provide some assurance that up-to-date
equipment will be available for use-an important
consideration given the current rapidity of technological
change in this field.
Economic appraisals attempt to assess the opportunity
costs, in terms of changes in resource use, brought about by
treatment alternatives. However, given the present

budgetary restrictions, one also needs to look at the purely
financial aspects—-that is, does the authority have the cash
available and from which budgets would it come?
The financial aspects of the treatment choices examined
here are of particular interest for two reasons. Firstly, the
concentrator method, although less costly overall, requires a
larger capital outlay on buildings and equipment.
Paradoxically, there may not be money in the capital building
or equipment budgets to enable the authority to launch such a
service. Therefore, this study illustrates again the need for
increased flexibility (or virement) between budgets, a point
made recently by researchers appointed by the Royal
Commission on the N.H.S.11
Secondly, the only existing funded service is the provision
of small ‘F’ sized cylinders. Under the present arrangement
the Family Practitioner Committee (F.P.C.) will pay for these
if they are prescribed by a general practitioner and delivered
by the pharmacist. Our study shows this to be the most
expensive and least convenient method of providing oxygen.
Although the F.P.C. budget is not cash-limited, the budget
from which concentrators would be provided should an
authority launch such a service, is almost certainly cash­
limited.
In order that patients and the N.H.S., should benefit from
the use of the most effective, convenient, and economical
method, administrative change is necessary. It seems
reasonable that once the need for long-term domiciliary
treatment has been established for a patient by appropriate
investigation in hospital, the cost ofthe continuing treatment
should be met from the F.P.C. budget, in the same way as that
for long term drug treatment for a chronic disorder. This
paper is not the place for discussing the ways in which the cost
of treatment could be funded from the F.P.C. budget, but
none of them would be necessary in a health service which
was unified so that all care, primary and secondary, were
provided by one authority. Meanwhile the question remains,
is it sensible for the N.H.S. to finance only the most expensive
and inconvenient treatment method? There is little point in
encouraging clinicians and managers to become more cost
effective in their actions, if some of the budgetary incentives
operate in the opposite direction.

We thank our many technical colleagues for their help. The study began
when K. V. L. was a member of the M.Sc course in Health Economics at the
University of York. Thanks are due to Mr A. K. Maynard the course Director,
and Mr Ken Wnght for helpful comments during the course of study.
Requests for reprints should be addressed to J. M. B., Depanment of
Medicine, Queen Elizabeth Hospital, Edgbaston, Birmingham BIS 2TH.
REFERENCES

1. Levine BE. Bigelow DB. Homstra RD, et al. The role oflong term continuous oxygen
administration in patients with chronic airway obstruction with hypoxaemia. An*
Inttrn Med 1967; 61: 6J9-50.
2. Abraham AS, Cole RB, Bishop JM. Reversal of pulmonary hypertension by prolonged
oxygen^ administration to patients with chronic bronchitis. Cin Ret 1968; 23:
3. Stark RD, Finnegan P, Bishop JM. Long term domiciliary oxygen therapy in chronic
bronchitis with pulmonary hypenension. BrMed] 1973; iii: 467-70.
4. NelT TA. Petty TL. Long term continuous oxygen therapy in chronic airway
obstruction. Ann intern Med 1970; 72: 621-26.
5. Anderson PB. Cayton RM, Holt PJ, Howard P. Long term oxygen therapy in cor
pulmonale. QJ Med 1973; 42: 563-73.
6. Leggett RJE, Kirby B J, Cooke NJ, Flenley DC. Longterm domiciliary oxygen therapy
in chronic cor pulmonale, non* 1976; 31: 414-18.
7. Nocturnal Oxygen Therapy Trial Group. Ann Intern Med 1980; 93t 391-98.
8. Long-term domiciliary oxygen therapy in chronic hypoxic cor pulmonale complicating
chronic bronchitis and emphysema. Lancet 1981; I: 681-85.
9. Drummond MF. Principles of economic appraisal in health care. Oxford: Oxford
Medical Publications, 1980.
10. Stark RD, Bishop JM. New method for oxygen therapy in the home using an oxygen
concentrator. Br Med J 1973; ii: 105-06.
11. Perrin f. Management of financial resources in the N.H.S. London: H.M. Stationery
Office, 1980.


•TH

K

t
t

Patients: 257 patients with deep venous thrombosis and
241 age- and sex-matched controls without deep venous
thrombosis.

Ithough information on prevention of, risk fac­
tors for, and treatment of complications of ve­
nous thromboembolism has increased substantially,
epidemiologic data and data on characteristics of
patients with previously verified thrombosis are
largely lacking. Despite treatment, long-term com­
plications of thrombosis are a major problem. In
older patients, leg ulcers are prevalent and place
great demands on the health care system (1). How­
ever, no information is available on the use of
health care resources by affected patients and the
costs of treating complications of thrombosis.
We therefore used an incidence approach (2) to
retrospectively collect patient data. Our objectives
were to document the extent and timing of long­
term complications and recurrent thromboembolism
and to estimate the health care costs of treating
these conditions. Eight types of complications were
defined (for controls, these complications are called
events): superficial venous thrombosis, deep venous
thrombosis, cellulitis, venous ulcer, varicose veins,
stasis dermatitis, deep venous insufficiency, and pul­
monary embolism.

Measurements: Data on use of health care resources and
costs of inpatient and outpatient care, pharmaceutical
agents, and treatment of complications.

Methods

Cost of Long-Term Complications
of Deep Venous Thrombosis of the
Lower Extremities: An Analysis of
a Defined Patient Population in
Sweden
David Bergqvist, PhD. MD; Stefan Jendteg, BSc;
Lars Johansen, MD; Ulf Persson, PhL, MPolSci;
and Knut Odegaard, PhD
Background: Little information is available on the epide­
miology and economic effect of long-term complications
developing after deep venous thrombosis.

Objective: To determine the extent of, timing of, and
treatment costs associated with long-term complications
developing after deep venous thrombosis of the lower
extremities.
Design: 15-year retrospective cohort study.
Setting: County hospital in Sweden.

Results: After 15 years of follow-up, 35% of the patients
with thrombosis and 57% of the controls were alive. Two
hundred forty-two complications were reported among
the patients with thrombosis, and 25 similar events were
reported among the controls. The average expected
present value of the health care cost of treating complica­
tions of thrombosis was estimated to be about $4659 in
the patients with thrombosis and $375 in the controls. In
controls, primary deep venous thrombosis cost about
$6000; thus, the additional long-term health care cost of
post-thrombotic complications is about 75% of the cost of
primary deep venous thrombosis.

Conclusions: The economic effect of post-thrombotic
complications is considerable. The use of measures to pre­
vent thromboembolism and its long-term complications
are justified on both clinical and economic grounds.

Ann Intern Med. 1997;126:454-457.
From University Hospital. Uppsala, Sweden; Institute of Health Econom­
ics. Lund. Sweden; and Skdvde Hospital. Skovde. Sweden. For current
author addresses, see end of text.

454

The medical records of 257 patients with a his­
tory of deep venous thrombosis of a lower limb
(verified by phlebography) and 241 controls without
a history of thromboembolic disease were reviewed
for clinical outcomes and use of health care re­
sources. By using hospital diagnosis registries, we
identified consecutive patients with deep venous
thrombosis whose condition was diagnosed between
1970 and 1985 at Skdvde County Hospital, Sweden,
or 1 of its 18 associated outpatient settings. Patients
who were registered in error, had thrombosis in a
part of the body other than lower limb, or had
thrombosis that had not been verified by phlebog­
raphy were excluded. The 257 study patients were
selected for thrombosis only. Controls were selected
from among all persons living in Skdvde, Sweden,
on 31 December 1979; this date was chosen to allow
a duration of follow-up similar to that for the pa­
tients with thrombosis. Controls were matched to
the patients by age (the closest date of birth) and
sex.
All patients who survived the follow-up period
had been followed for at least 10 but no more than
15 years. Patients who died were followed until the
year of death. Controls were followed for 15 years
or until the year of death. At study entry, mean
ages were 64 years in the thrombosis group (60%
men) and 66 years in the control group (59% men).
The following complications or events were re­
corded: superficial venous thrombosis (verified by

15 March 1997 • Annals of Internal Medicine • Volume 126 • Number 6

.■ I
phlebography or clinical diagnosis); deep venous
time of the primary thrombosis, we aclded all pr
thrombosis (verified by phlebography); cellulitis, ve­
ent and expected future costs according to the f
nous ulcer that was not caused by arterial insuffi­
lowing formula:
ciency or diabetes, varicose veins, and stasis derma­
titis (each verified by clinical diagnosis); deep venous
15
insufficiency (verified by Doppler ultrasonography);
q7’Gj/(1 + rf I Ni
and pulmonary embolism (verified by scintigraphy
or autopsy). Total use of medical care for each re­
where TC is the total treatment cost for cohort i
corded complication or event was ascertained from
year t after primary thrombosis, r is the discotic
records of primary care and hospital care from six
rate, N is the sample population of cohort z, and i is
hospital departments. About 3000 medical records
1 for patients with thrombosis and 2 for controls
were surveyed to determine total number of outpa­
None of the funding parties had any role in tf
tient visits, days in the hospital, diagnostic measures,
collection, analysis, or interpretation of study data.
surgical procedures, and pharmaceutical agents.
Costs of the health care resources were based on
reported unit prices of inpatient and outpatient care
Results
and pharmaceutical agents. In Sweden, health care
purchasers use pricing data to reimburse providers
The results of our analysis are presented in Ti
for cross-boundary health services. Thus, such bill­
hies
1 and 2. Table 1 shows the differences in the
ing data are available from local, regional, and na­
frequency of complications or events and the sur­
tional health services. From an economic point of
vival pattern for the two groups. At the end of th.
2CW’PnCing dat5 Cannot £enerally be considered a
follow-up period, 242 post-thrombotic complication,
true measure of actual
<
------1 costs.
However, true cost
were recorded among the patients with thrombosis
measures that reflect full-opportunityj costs are not
and 25 events were recorded among the controls ('
regularly produced in the health care system. In our
of these events were cases of primary deep venou:
study, we assumed that reported unit prices arc
are
thrombosis). In the patients with thrombosis, ap­
reasonable approximations of costs. All estimated
proximately two thirds of the complications oc­
costs were calculated in Swedish kronors (SEK) in
curred within 5 years after primary deep venouj
fixed 1990-1991 prices and then converted to US
thrombosis
had developed. The survival pattern also
dollars (in 1991, 1 U.S. dollar = SEK 5.5). A 5%
differed
substantially
between groups: Thirty-five
discount rate was used to adjust for the different
percent
of
patients
with
thrombosis and 57% of
timings of complications or events.
controls
were
alive
at
the
end
of the period.
To estimate the average present value of expect­
Because
of
differences
in
the
frequency of com­
ed lifetime costs of treating post-thrombotic compli­
plications or events and survival, the differences in
cations per patient of cohort z, discounted to the
the frequency of complications or events per patient

Year of Follow-up

___________________ Thrombosis Group
Control Group
CompEcations

Probability
of Survival

n

0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

Total

89
26
22
14

18
10
10
14
16
5
4
5
4
5
0
242

1
0.8872
0.8171
0.7665
0.7121
0.6848
0.6342
0.6186
0.5754
0.5556
0.5355
0.4861
0.4388
0.3921
0.3671
0.3516

Complications per
Exposed Patient

Total Cost

Events

n

SEK

n

3 050 422
922 780
334 777
334 028
1 189 057
354 273
263 806
572 053
515 598
86 114

5
2
0
6

0.0000
0.3903
0.5141
0.6258
0.7023
0.8046
0.8660
0.9289
1.0235
1.1356
1.1719
1.2039
1.2483
1.2880
1.3410
1.3410

43 053
78 433
57015
49 287
0

1
2
3
1
0
0
2
0
0
2
1

7 850 696

25

Probability
of Survival

1
0.9627
0.9295
0.8963
0.8548
0.8299
0.7884
0.7801
0.722
0.6971
0.6846
0.6473
0.6349
0.5975
0.5892
0.5726

Events per
Exposed Patient

Total Cost

n

SEK

0.0000
0.0216
0.0305
0.0305
0.0596
0.0646
0.0751
0.0911
0.0968
0.0968
0.0968
0.1097
0.1097
0.1097
0.1237
0.1310

220 264
14 300
0
154 226
8 872
78 602
38 386
8 189
0
0
75 706
0
0
1200
7359
607 104

* SEK = Swedish kronors.

15 March 1997 • Annals of Internal Medicine • Volume 126 • Number 6

455

■.WIL

•i-.

Table 2.

Average Costs per Complication or Event in Fixed 1990-1991 Prices by Post-Thrombotic Complications or Events*

Superficial venous thrombosis
Recurrent deep venous thrombosis
Primary deep venous thrombosis
Cellulitis
Venous ulcer
Varicose veins
Stasis dermatitis
Deep venous insufficiency
Pulmonary embolism
Combinations of several conditions
Overall average

Control Group

Thrombosis Group

Complication or Event

Average Cost per
Complication

Complications

Average Cost
per Event

Events

SEK

n

SEK

n

15 638
36 877

58
74

600

2

14 907
43 630
11 420
7300
6413
33 932
90 262
32 441

22
19
11
3
3
32
20
242

33 455
28 017
38 433
6105

7
3
5
4

17 771

4

24 284

25

f

* SEK = Swedish kronors.

exposed to risk were even larger. At the end of the
follow-up period, a surviving patient in the throm­
bosis group had had, on average, 10 times more
complications than a surviving control. The total
costs of treating complications or events were SEK
7 850 696 ($1 427 399) for patients with thrombosis
and SEK 607 104 ($110 383) for controls.
For the entire follow-up period, the average costs
per complication or event were SEK 32 441 ($5898)
for patients with thrombosis and SEK 24 284 ($4415)
for controls. Venous ulcer was the most expensive
type of complication, followed by deep venous throm­
bosis and pulmonary embolism (Table 2). For both
groups, however, the treatment cost per complica­
tion or event varied greatly, indicating similar vari­
ation in the severity of the complications or events.
Among the patients with thrombosis, more than
one third of the treatment cost was attributable to
recurrent deep venous thrombosis. Among the con­
trols, primary deep venous thrombosis accounted
for 38.6% of the total treatment cost. The estimated
average cost of primary deep venous thrombosis was
SEK 33 455 ($6083) (Table 2).
At the end of the follow-up period, the dis­
counted average present values of treatment cost
were SEK 25 625 ($4659) for patients with throm­
bosis and SEK 2060 ($375) for controls. The differ­
ence in the discounted present values of treatment
cost between the thrombosis and control groups
(that is, SEK 25 625 - SEK 2060 = SEK 23 565
[$4285]) can be interpreted as the expected cost of
treating post-thrombotic complications. Thus, an es­
timate of the present and expected cost of treating
both primary deep venous thrombosis and related
post-thrombotic complications is SEK 57 020 ($10 368)
(that is, SEK 33 455 + SEK 23 565). According to this
estimate, approximately 60% of the cost is attribut­
able to primary deep venous thrombosis and 40% is
attributable to post-thrombotic complications.
456

Discussion

In our study of long-term complications of deep
venous thrombosis of the lower extremity, we com­
pared patients who had phlebography-verified throm­
bosis with age- and sex-matched controls who served
as a baseline benchmark. We considered the inclu­
sion of controls to be important because the prev­
alence of nonthrombotic deep venous insufficiency
increases with age (3). One limitation of our study
could be that we matched the groups for age and
sex only; we were unable to match them for health
conditions or factors that predispose patients to
deep venous thrombosis or venous insufficiency. ,
Nonetheless, even controls matched only for age
and sex provide the incidence of deep venous insuf­
ficiency in an unselected population.
In our study, the post-thrombotic complications
that occurred over 10 to 15 years of follow-up could
be corrected for similar events among controls: In
patients with thrombosis, the risk for a thrombotic
complication was 10 times greater and the cost of such
a complication was 12 times greater compared with
controls. To our knowledge, ours is the first study to
compare clinical and economic post-thrombotic com­
plications with what could be considered the natural
occurrence in an age- and sex-matched population.
In the control group, all patients were followed
for 15 years or until death. In the thrombosis group,
however, 30 patients (12%) were censored because
they were followed for 10 to 14 years. These pa­
tients represent a loss of 3.7% of all years of follow­
up. The number of complications per exposed pa­
tient decreased with time (Table 1). Through the 10
to 15 years of follow-up, the annual risk for a com­
plication is about one third of the risk during the
first 10 years. Thus, the number of complications in
the thrombosis group is underestimated by about
1.2% (3.7% X one third) as a result of the censored

15 March 1997 • Annals of Inienuil Medicine • Volume 126 • Number 6

I

Ai
&
’T?'-

I

&
■#

4


years of follow-up. Because of the discounting prin­
ciple, however, costs that are incurred after 10 years
and are discounted by 5% will result in a present
value of only about 50% of the original. Because of
censored patients, therefore, we may have underes­
timated the present value of post-thrombotic com­
plications by less than 1%. We did not adjust for
these censored patients; thus, our estimate of the costs
of post-thrombotic complications is conservative.
We obtained data on complications from patient
records. This suggests that the frequency of compli­
cations was lower than that seen in Lindhagen and
colleagues’ study (3), in which each patient was
investigated at follow-up. However, because our
goal was to assess clinically important long-term
consequences, our approach seems reasonable.
We estimated that the average cost per complica­
tion or event was approximately SEK 33 000 ($6000)
for primary deep venous thrombosis and SEK
34 000 ($6182) for pulmonary embolism. In an ear­
lier study of patients from Malmo General Hospital
in southern Sweden (4), these estimates (converted
to 1991 prices) were approximately SEK 20 000
($3600) for deep venous thrombosis and SEK 24 000
($4400) for pulmonary embolism. Because use of
hospital resources and costs are local, our cost es­
timates (obtained from one hospital) may not be
generalizable to other settings. One of our more
general results is the ratio between the expected
costs of long-term treatment of post-thrombotic
complications and primary deep venous thrombosis.
This ratio was approximately 0.75 (that is, the ex­
pected long-term costs of treating post-thrombotic
complications are about 75% of the costs of treating
primary deep venous thrombosis).
In the classic study by Bauer (5), the incidence of
complications increased during the first several
years of follow-up and decreased after about 5
years. However, decrease in incidence may start ear­
lier when venous function is measured objectively
(6). During the first 5 years of our study period,
complications developed in about two thirds of the
patients with thrombosis and in 56% of the controls.
Pulmonary embolism and recurrent thrombosis, how­
ever, usually occurred within the first year.
A high mortality rate in patients with thrombosis
could be partially explained by concomitant cancer
or cardiovascular disease (7, 8). In our study, the
mortality rate after 15 years was 65% in the throm­
bosis group and 43% in the control group.
Medical records contained little data on use of
oral anticoagulation therapy, ulcer dressings, and
supportive stockings. Thus, the costs of these mea­
sures are not included. Indirect costs resulting from
loss of productivity were not estimated because no
appropriate data were available. However, because
two thirds of the patients were younger than 70

years of age at the time of thrombosis and ,most of
them could be considered to be employed, the in­
clusion of indirect costs would further increase the
long-term cost. This suggests that the estimated cost
difference between the groups is minimal.
The estimated incidence of deep venous throm­
bosis in Nordic countries is 1.5 to 2 cases per 1000
persons per year; surgery without prophylaxis is an
important risk factor (9, 10). Untreated venous
thromboembolism is associated with considerable
risk for death and chronic venous insufficiency. More­
over, our findings show that long-term complications
have a notable economic effect. Economic evaluations
(4, 11, 12) have shown that prophylactic measures are
cost-effective compared with no prophylaxis or surveil­
lance and selective treatment of venous thromboem­
bolism. We conclude that the use of measures to
prevent thromboembolism and its long-term complica­
tions is justified on both clinical and economic
grounds.
Acknowledgment: The authors thank Eva-Karin Johansson, RN,
for compiling patient data and providing medical judgments.
Grant Support: In part by Ciba-Geigy Limited, grant 00759 from
the Swedish Medical Research Council, and the Swedish Heart
and Lung Foundation.

Requests for Reprints: Lars Johansen, MD, Department of Sur­
gery, Skdvde Hospital, S-541 85 Skovde, Sweden.
Current Author Addresses: Dr. Bergqvist: Department of Surgery,
University Hospital, S-751 85 Uppsala, Sweden.
Mr. Jendteg: Institute of Health Economics, PO Box 2127, S-220
02 Lund, Sweden.
Dr. Johansen: Department of Surgery, Skdvde Hospital, S-541 85
Skdvde, Sweden.
Mr. Persson and Dr. Odegaard: Institute of Health Economics,
PO Box 2127, S-220 02 Lund, Sweden.

References
1. Nelzen O, Bergqvist D. Chronic venous insufficiency and leg ulcers: how big
is the problem? Perspectives in Vascular Surgery. 1994;7:33-42.
2. Hartunian MS, Smart CN, Thompson MS. The Incidence and Economic
Costs of Major Health Impairments. Lexington, MA: Lexington Books; 1981.
3. Lindhagen A, Bergqvist D, Hallbook T, Efsing HO. Venous function five
to eight years after clinically suspected deep venous thrombosis. Acta Med
Scand. 1985;217:389-95.
4. Bergqvist D, Jendteg S, Lindgren B, Matzsch T, Persson U. The eco­
nomics of general thromboembolic prophylaxis. World J Surg. 1988;12:349-55.
5. Bauer G. A roentgenological and clinical study of the sequels of thrombosis.
Acta Chir Scand. 1942;86(Suppl 74): 1-126.
6. Kakkar W, Lawrence D. Hemodynamic and clinical assessment after therapy
for acute deep vein thrombosis. A prospective study. Am J Surg. 1985;6:54-63.
7. Beyth RJ, Cohen AM, Landefeld CS. Long-term outcomes of deep-vein
thrombosis. Arch Intern Med. 1995;155:1031-7.
8. Nordstrom M, Lindblad B, Anderson H, Bergqvist D, Kjellstrom T. Deep
venous thrombosis and occult malignancy; an epidemiological study. BMJ.
1994;308:891-4.
9. Nordstrom M, Lindblad B, Bergqvist D, Kjellstrom T. A prospective study
of the incidence of deep-vein thrombosis within a defined urban population.
J Intern Med. 1992;232:155-60.
10. Bergqvist D. Postoperative Thromboembolism: Frequency, Etiology, Prophy­
laxis. Berlin: Springer-Verlag; 1983.
11. Bergqvist D, Matzsch T. Cost/benefit aspects on thromboprophylaxis. Hae­
mostasis. 1993;23(Suppl 1):15-9.
12. Oster G, Tuden RL. Colditz GA. A cost-effectiveness analysis of prophylaxis
against deep-vein thrombosis in major orthopedic surgery. JAMA. 1987;257:
203-8.

© 1997 American College of Physicians

15 March 1997 • Annals of Internal Medicine • Volume 126 • Number 6

457

f

i
i

THE NATIONAL MEDICAL JOURNAL OF INDIA

I

VOL. 9, NO. 6, 1996

259

Original Articles
I

Morbidity pattern, health care utilization and per capita health
expenditure in a rural population of Tamil Nadu
J. RAJARATNAM, R. ABEL, S. DURAISAMY, K. R. JOHN

ABSTRACT
Background. Information on the existing morbidity pattern,
pattern of health care utilization and the per capita health
f 'enditure is essential to provide need-based health care
Gw.ivery to a rural population. To obtain this information we
performed a study in the K.V. Kuppam Block, North Arcot
Ambedkar District, Tamil Nadu.
Methods. We did a cross-sectional study, interviewing
respondents from 300 households, from 3 panchayats using a
multistage sampling technique. Information relating to 1440
persons was collected. The morbidity data was obtained
initially for the week prior to the day of Interview, followed by
one week to one month and then for two months to one year.
Results. During 1990-91,825 of the 1440 persons (57.3%)
did not have any illness. Sex had no bearing on the number of
illnesses. Of the 60 children less than 2 years of age, 42 (70%)
had one or two illnesses. The period prevalence of infective
and parasitic diseases was found to be 21.9% with an average
of 3 episodes. Services rendered by private practitioners
(registered, non-registered and Indigenous) were utilized by
59% of the households and 79% of the households had used
allopathic treatment at some time. The average per capita per
annum health expenditure was Rs 89.9 (Rs 449 per housed). This increased significantly with increase in the house­
hold size (p<0.001) and per capita income (p<0.01).
Conclusion. The health-seeking behaviour of this popula­
tion can be changed if efficient senrices are rendered through
government primary health centres and subcentres. This
would allow the existing voluntary agency to withdraw without
much change in the per capita health expenditure.
Natl Med J India 1996;9:259-62

I
J

INTRODUCTION
Health surveys are a useful tool for assessing a population’s health
care needs reliably. Morbidity and health care utilization surveys
ideally go together. It is useful to know the pattern of health
problems suffered by a population, what was done about them as
well as the expenditure incurred on treatment
An important factor in providing primary health care is the
Collegc HosPital’ RUHSA Campus 632209, Tamil

v RDAJT^utTNAM’ R ABEL* S* DURAISAMY RUHSA Department
K.. K. JOHN Department of Community Health
Correspondence to J. RAJARATNAM

© The National Medical Journal of India 1996

cost. Cost studies provide valuable information on health
expenses, help develop cost-consciousness and are a means of
budgetary control. Such information includes treatment costs,
patterns of health expenditure in specific settings and the preva­
lence of diseases in the area.1
This study was carried out in arural area of south India to study
the morbidity pattern, the pattern of utilization of health services
and the mean expenditure on health per household and per person.

SUBJECTS AND METHODS
A cross-sectional study was carried outinJuly 1991 in3 panchayats
of K.V. Kuppam Block, North Arcot Ambedkar District, Tamil
Nadu. The subjects were chosen using a multistage sampling
technique. A stratified random sampling method was adopted at
the first stage to select 3 of the 39 rural panchayats based on their
distance from the main road with <3.3 km, 3.3-6.6 km and >6.6
km as cut-offs. The selected panchayats represented three major
population groups with different socio-economic characteristics.
A systematic sampling method was used at the second stage to
choose 23 % ofthe households (a total of331) from each panchayat.
Of these, 31 (2%) could not be studied, since the houses were
found locked or were vacant during the survey period. Thus,
respondents from 300 (21 %) households ofthe 3 panchayats were
interviewed. There was no instance of non-cooperation.
A pre-tested interview schedule was used by trained interview­
ers to collect the data. Information relating to a total of 1440
persons was obtained from these 300 households. The morbidity
pattern was obtained by asking the respondent first about the one
week period prior to the day of interview, followed by the one
week to one month period and then the 2-12 months period. This
was done to minimize the relapse bias. Subsequently, the place of
treatment for each episode of each illness was obtained and the
type of treatment given recorded. Health expenditure was defined
as the expenditure incurred for preventive and curative health
care; but money spent on home treatment (use ofbalm and buying
medicines from shops) was excluded. The health expenditure
incurred included treatment cost, travel, food and wages lost
during illness; other expenses were obtained corresponding to
each episode of each illness.
The other variables recorded were the age, sex, caste, income,
education, illness and number of episodes. The income through
different sources and through wages of different persons were
carefully obtained. The respondent was the head of the household
or any responsible person within that household. The morbidity
was classified according to ICD codes.2
The terms ‘person’, ‘illness’ and ‘episodes’ were defined as

Him

Ill I

260

THE NATIONAL MEDICAL JOURNAL OF INDIA

recommended by the Expert Committee on Health Statistics ofthe
World Health Organization (WHO).3 If a person within a given
period of observation (e.g. a year) suffered from illness A twice
and from illness B three times and if the episodes of illness A and
B occurred at different times, this person contributed one unit to
statistics of persons for each illness, two units to statistics of
illness, and 5 units to episodes (2 'episodes* to illness A and 3 to
B). Period prevalence (persons) was defined as the number of ill
persons during a defined period (a year divided by the average
number of persons). Period prevalence (episodes) was the number
of episodes of illness which were current at some time during a
defined period divided by the average number of persons exposed
to risk during the periock
Means, percentages, period prevalence (persons), period preva­
lence (episodes). Chi-square test, correlation in univariate analy­
sis and step-wise regression in multivariate analysis were the
statistical measures used.
’SULTS
Of the 300 households studied, 48 belonged to forward commu­
nities, 144 to backward communities and 108 to scheduled castes.
Half the households had a thatched roof, 29% a tiled roof and 21 %
a terraced roof. No morbidity was reported in 7.7% (23) of house­
holds. Ofthe 1440persons from300households, 615 (42.7%) had
one or more illness during the year. The sex of the person had no
bearing on the number of illnesses (Table I).
Among children below 2 years of age, 70% had one or two
illnesses. This group also had the highest proportion of two
illnesses (21.7%) and 63% of the under-fives had been ill once or
twice. As a group, school-going and adolescent children (6-19
years) had the least (31.7%) instances of illness while 45% of the
20-44 years group had had an illness. Three illnesses were
reported only in the age group £45 years (Table II).
Sixty per cent of the sampled households had infective and
parasitic diseases, 34% had diseases related to the nervous system
and sense organs, 15.3% had diseases of the digestive system,
10% had injuries and poisoning, 10% disease of the skin and 10%
—eded preventive care.
Table III details the morbidity pattern. Period prevalence

VOL. 9, NO. 6, 1996

Table I. Sex distribution of the population related to the number
of illnesses
Number

Male
(n=731)

Female
(n=709)

Total
(n=1440)

One
Two
Three

261 (35.7)
33 (4.5)
1 (0.1)
295 (40.4)

272 (38.4)
44 (6.2)
4 (0.6)
320 (45.1)

533 (37.0)
77 (5.3)

Total

5 (03)

615 (42.7)

Figures in parentheses are percentages

Table II. Age distribution of the population related to the number
of illnesses
Age

n

0-2
3-5
&-19
20-44
£45
Total

60
67
372
533
308
1440

Number of illnesses
One
Two
Three

29 (48.3)
36 (53.7)
139 (29.2)
210 (39.4)
119 (38.7)
533 (37.0)

13 (21.7)
2 (3.0)
12 (2.7)
30 (5.6)
20 (6.5)
77 (5.3)

Sick persons

42 (70.0)
38 (56.7)
151 (31.7)
240 (45.0)
5 (1.7) 139 (45.2)
5 (0.3) 615 (42.7)

Figures in parentheses are percentages

(persons) of infective and parasitic diseases (including fever,
cold, cough and headache) was found to be 21.9% and that for
diarrhoea and dysentery was 2.8%; the period prevalence (epi­
sodes) were 74.7% and 8.3%, respectively. Illnesses related to
the nervous system and sense organs accounted for a period
prevalence of 8.6% (persons) and 75.5% (episodes) followed by
diseases of the digestive system.
Less than 1% of the population had chronic diseases such as
asthma (0.7%), ulcer (0.6%), tuberculosis (0.3%), diabetes (0.3%),
leprosy (0.2%), mental disorders (0.1%), cancer (0.06%) and
acute diseases such as typhoid (0.2%), measles (0.2%) and jaun­
dice (0.1%).
Of the total population less than 1 % had problems related to
pregnancy, childbirth and puerperium. However, the period preva­
lence among women only for menstrual problems was 1.4%, for

Table HI. Period prevalence of illness (persons and episodes) during 1990-91

ICD codes
I
II

in
IV
V
VI

vn
vni
IX
X
XI
XII

xni
XVII

Type of illness

No. of persons

Infective and parasitic diseases
Neoplasm-cancer*
Diabetes mellitus*
Anaemia
Mental disorders*
Nervous system and sense organs
Circulatory system
Asthma
Digestive system
Genito-urinary system*
Complications of pregnancy,
childbirth and puerperium
Skin and subcutaneous tissues
Musculoskeletal system and connective
tissue
Injuries and poisonings
Dosham/Sevappu/MedicaA checkup

351 (179)
1

5 (4)
2
2
124 (103)
5 (5)
10 (8)
67 (46)
6
42 (32)
46 (30)

15 (13)
43 (32)
5_____

"Episodes not applicable
Figures in parentheses are number of households
Sevappu local term used to describe a child who turns blue and dies

Period prevalence (persons)
per 100(n=1440)
21.9
0.06
0.3
0.1
0.1
8.6
0.3
0.7
4.7
0.4
2.9

3.2
1.0

No. of episodes Period prevalence (episodes)
per 100(n=l440)
1076

74.7

25

1.7

1087
13
110
229

75.5
0.9
7.6
15.9

202

14

343
136

23.8
9.4

142
9.9
3.0
9
0.7
_________ 03
Dosham local term used to describe children suffering from fever, diarrhoea and dysentry

RAJARATNAM et al. : MORBIDITY PATTERN, HEALTH CARE UTILIZATION AND EXPENDITURE

antenatal checkup 1.8% and for child immunization 1.6%.
More than half the households (59%) preferred to go to private
practitioners (registered, non-registered or indigenous) and only
28% used the services provided by a voluntary agency serving the
block (Table IV). Only 2% had adopted home treatment and 3%
had gone to a medical shop. This was either for pain-relieving
medicines or just for simple cold, cough or fever. While 79% of
the households had used allopathic treatment at some time, 33%
had used homoeopathic medicines (Table IV).
The annual health expenditure was Rs 89.9 per person and
7.7% of households had not incurred any expenditure. Consider­
ing an average household size of 4.8 persons, the annual health
expenditure worked out to Rs 469 per household. As the per capita
income (PCI) increased, the per capita health expenditure (PCHE)
also increased significantly (p<0.001; Table V). Similarly, as the
family size increased the PCHE also increased. However, caste
had no association with PCHE.
In the correlation analysis the PCHE correlated positively with
PCI (p<0.01) and household size (pcO.OO 1). However, the PCHE
f' was not related to the social status of the family (p>0.01).
In the step-wise regression analysis the estimated coefficient
indicated that when the PCI increased by Rs 1.00, the average
increase in PCHE was Rs 0.03. When the household size in­
creased by one unit, the PCHE increased by Rs 8.58 and when the
PCI was Rs 1200 step-wise regression showed that on an average
every individual spent Rs 36 per annum. The R2 value of PCHE
with household size (0.11) was higher than the R2 value of PCHE
with PCI (0.06), thereby implying that the regression fit of the
PCHE and household size was comparatively better than that of
PCHE and PCI.

DISCUSSION
Considering the methodological issues in a morbidity survey, the
use of tracer conditions has been found to significantly increase
completeness of reporting. This consists of the use of a checklist
of specific symptoms associated with a given health problem in
the questionnaire and asking the respondent whether she/he had
Table IV. Place and type of treatment (n=277)
Place/Type of treatment

n(%)

Place
Voluntary organization (Primary and secondary care)
Christian Medical College Hospital (Voluntary tertiary care)
Government (Primary, secondary and tertiary)
Private practitioners (Registered, non-registered, indigenous)

85 (28)
12 (4)
75 (25)
177 (59)

Type
Allopathy
Homoeopathy
Medical shop
Native doctor
Petty shop
Home treatment

238 (79)
98 (33)
10 (3)
7 (2)
4 (1)
6 (2)

Ii
Table V. Per capita income v. per capita health expenditure
Per capita income (Rs) ' Households
<600

601-1200
1201-2400
>2400
Total

109
96
59
36
300

Mean (SD) health expenditure
40.0
68.0
140.1
216.9
89.9

(94.1)
(140.5)
(408)
(407.6)
(254.4)

261

any of these during the period in question. The WHO has pro­
duced a select list of symptoms associated with various health
problems that may be used by a lay interviewer in a health survey.4
Tekle-Haimanot Makonnen picked up all morbidity by using
tracer conditions in a rural health survey in Ethiopia.3 We did not
use any systematic list of tracer conditions, but spent time by
probing into various morbid conditions to increase completeness
of reporting.
The duration of the recall period also influences the complete­
ness and reliability of reporting. The longer the recall period, the
less likely a person will remember an illness. The health survey
conducted in 1954-55 in Califomia,USA6 established this fact A
morbidity survey in Japan confirmed that recall lapse affects the
not-so-serious health conditions much more. We asked the
respondents for information with regard to the incidence of illness
for each of the four calendar weeks preceding the interview. The
total samples were then randomized over a period of 52 weeks so
as to accurately reflect the prevalence. We adopted a systematic
recall of one week prior to the interview, two weeks to one month
and then one month to 1 year which, to some extent, assures the
completeness and reliability of reporting. This method also
allows the seasonal variations in morbidity to be inchided and
estimation of the PCHE per year.
Rao et aL1 have stated that a longitudinal study overcomes the
problem of a recall bias. However, longitudinal studies are expen­
sive and hence can cover only a short duration. A combination of
retrospective and longitudinal studies are considered ideal.
Another methodological issue relates to not including overthe-counter purchases and self-care. The proportion reporting
self-care was low and these tended to be symptomatic treatment
of one or two doses which would not contribute substantially to
the total cost With the information available from this study and
with our own experience of this area avc do not anticipate any
major bias in the total costs incurred by the households.
The findings of a previous study have shown that for one visit
to the RUHSA Health Centre each person on an average paid only
Rs 8.80.1 This fact, as also the information available from this
study, suggest that it is unlikely that any significant bias in the
total costs incurred has been introduced.
In Mumbai,9 75% of urban poor households suffered from
infective diseases while we found that 60% of rural households
had the same diseases. In rural Nigeria10 in 1991 the prevalence of
fever, gastroenteritis and chest infection was estimated at 50%,
37% and 10%, respectively. We found that 60% of the households
studied had had infective and parasitic illnesses. A longitudinal
study carried out in the same population between 1990-92 had
indicated that the incidence (number of episodes/child-year) of
diarrhoea and respiratory diseases among children less than 3
years of age was 1.77 and 2.56 respectively.11 The low incidence
may be related to the existence of a rural health project providing
primary and secondary health care in this area for fifteen years.12
For the same reason, 30% of the less than 2-year-olds in this study
had no disease compared to only 16.3% in another study from the
same state.”
The positive relationship, between PCI and PCHE is highly
significant*with the chi-square test and correlation analysis.
However, in the multivariate analysis the relationship between
PCI and PCHE is not as strong.
In 1971-72 the annual health expenditure in the same district1
was estimated at Rs 13.09 per person and Rs 75 per family. In
1983-84 Scheer et aL14 reported an average health expenditure of
Rs 250-300 per year per family in the same area. The present

262

THE NATIONAL MEDICAL JOURNAL OF INDIA

study showed the annual health expenditure to be Rs 89.9 per
person and Rs 449 per household. The increase in the health
expenditure over this period is possibly due to inflation. In 1975
SxiiivcHai et aL13 estimated the annual PCHE in a rural area of
Tamil Nadu to be Rs 24 whereas a study from Mumbai9 among the
urban poor, found the annual per household health expenditure to
be Rs 300 or more.
The morbidity pattern of this rural community, the health care
utilization pattern and the PCHE indicate that strengthening of the
government primary health Centres and subcentres in this area and
changing the health-seeking behaviour of the population may
allow the voluntary agency to withdraw. Also the PCHE can be
kept at an affordable level if sufficient and efficient health care
services are provided.
ACKNOWLEDGEMENTS
We sincerely acknowledge the help of Mr Govindaraj, Mr Parandaman and
Mr Sundaramoorthy in date collection and Mr Anbalagan and Ms Mercy in
date analysis. The consultation of Dr P. S. S. Rao, Head of Biostatistics and
Mr Sampathkumar, Training Officer in RUHSA Department, Christian
Medical College, Vellore, are gratefully acknowledged. Ms V. Mahalakshmi’s
d processing service is sincerely acknowledged.

REFERENCES
1 Venn* BL, Srivutava RN. Measurement of the personal cost of illness due to some
major water related diseases in an Indian rural population. Int J Epidemiol
1990; 19:169-76.

VOL. 9, NO. 6, 1996

2 World Health Organization. Inumational Classification of Diseases. Volumes i
and 11. Geneva:World Health Organization, 1975.
3 Rao PSS.JesudinG, Richard J. An introduction to biostatistics. Vellore: Prestografic,
1987:191-8.
4 World Health Organization. Lay reporting of health urformation. Geneva:World
Health Organization, 1978.
5 Tekle-Haitnanoc Makonnen. Two rounds of a raral health survey in Ethiopia in
1982-83. Need, objectives, preparations, date collections and evaluation ofquality.
World Health Stat Q 198538:107-21.
6 Mooney W. Methodology in two California Health-Surveys: Public health
monographs. Washington:Goveniment Printing Office, 1962:70.
7 Rao PSS, Rajamaniciom C, Fernandez SR. Personal health expenses among rural
communities of North Arcot District Indian J Med Res 1973:61:110-11.
8 Kunivilla A, George K. Rajaratnam A, John KR. Prescription pattern and cost
analysis of drugs in a base hospital in Sooth India. Natl Med J India 19943:
167-8.
9 Yesudian CAK Utilization ofhealth services by the urban poor. Mumbai:World
Health Organization and Tata Institute of Social Services, 1988.
10 Adekoht John ED. A health expenditure survey of morbidity in raral Nigeria. AfrJ
MedSci 199130:15-22.
11 Ramakrishnan U, Latham MC, Rajaratnam A. Vitamin A supplementation and
morbidity among pre-school children in South India. Am J Clin Nutr 1995;67:
1295-303.
12 Abel R. RUHSA—A model Primary Health Care Programme. J Trap Pediatr
199238:270-2.
13 Srinivasan V. Nair PS, Devi RK, Dutt PR. Report on a morbidity survey of
Gandhigram community, Tamil Nadu. Bull Gandhigram Inst Rural Health and
Family Planning 1975:10:43-63.
14 Scheer P. Pinto A, Tuqa S, Ebrahim GJ, Abel R, Mukherjee DS. Perspectives in
Primary Health Care: Does health intervention ameliorate the effects of poverty
related diseases ? Experience in rural south India. J Trap Paediatr 198531:219-22.

Growth hormone-producing pituitary tumours:
Clinical profile and results of surgery
K. D. MODI, A. MITHAL, D. BANERJI, D. KUMAR, P. SHAH, V. K. JAIN, D. K. CHHABRA

ABSTRACT
Background. Growth hormone-producing pituitary tumours
present with a wide variety of manifestations. The optimum
diagnostic work up, management and follow up of such
patients is complex and involves a multidisciplinary approach.
There is paucity of data from India with regard to the clinical
presentation and results of surgery for growth hormoneproducing tumours.
Sanjay Gandhi Postgraduate Institute of Medical Sciences,
Lucknow 226014, Utter Pradesh, India
K. D. MODI, A. MITHAL, P. SHAH Department of Endocrinology
D. BANERJI, D. KUMAR. V. K. JAIN, D. K. CHHABRA
Department of Neurosurgery

Correspondence to A. MITHAL

© The National Medical Journal of India 1996

Methods. We studied the first 50 patients presenting during
1989-94 with growth hormone-producing pituitary tumours to
our centre. The work up included detailed endocrine and radio­
logical assessment The surgical outcome was analysed for 35
patients who were operated (trans-sphenoidal 29, transcranial
6) at our centre.
Results. All the patients had macroadenomas [mean (SD)
diameter 3.12 (0.87) cm]. Seventy-five per cent of the patients
had supra- and/or parasellar extension and 57% had visual
field defects. Tumour size correlated with the preoperative
basal (r=0.57) and glucose-suppressed (r=0.54) growth hor­
mone levels. Thirty-three of the 35 patients operated at our
centre (trans-sphenoidal 28, transcranial 5) were available for
follow up (median duration 34 months). After trans-sphenoidal
surgery alone, 12 of the 28 (43%) patients had normalization
of growth hormone levels (post-glucose growth hormone

.....
i |ic New England Journal of Medicine

f

tv.’

1
\

Special Articles
*
$

i

I

COSTS AND CHARGES ASSOCIATED WITH THREE ALTERNATIVE TECHNIQUES
OF HYSTERECTOMY
James H. Dorsey, M.D., Patrice M. Holtz, R.N., M.S., Robert I. Griffiths, Sc.D., Margaret M. McGrath, M.S.,
and Earl P. Steinberg, M.D.

Abstract
Background Many hysterectomies are now per­
formed by a laparoscopically assisted vaginal tech­
nique. This procedure is controversial, partly be­
cause of concern about cost. We studied hospital
charges and costs for the procedure as compared
with those for total abdominal hysterectomy and to­
tal vaginal hysterectomy in clinically similar groups
of patients.
Methods From hospital-discharge data and pa­
tients' charts, we identified hysterectomies per­
formed in 1993 and 1994 by 96 surgeons at a com­
munity teaching hospital to treat benign conditions.
The patients were grouped according to the surgical
procedures performed in conjunction with the hys­
terectomy. Data on hospital charges and cost-tocharge ratios for 64 hospital cost centers were used
to assess charges and costs for specific resources, as
well as for the hospitalization overall.
Results Of 1049 patients studied, 26 percent un­
derwent laparoscopically assisted vaginal hysterec­
tomy, 54 percent underwent abdominal hysterecto­
my, and 20 percent underwent vaginal hysterectomy.
The average hospital stays were 2.6, 3.9, and 2.9
days, respectively, and the mean total charges (facil­
ity charges plus professional fees) for the hospital­
izations were $6,116, $5,084, and $4,221 (PcO.OOl
for the comparison of the laparoscopic technique
with both other techniques). The mean facility costs
were $4,914, $3,954, and $3,116, respectively
“(P^cOtOOTfor the's^me comparison), with similar
findings in all subgroups. The higher charges and
costs for laparoscopically assisted vaginal hysterec­
tomy were due to higher supply costs, particularly
when disposable supplies were used, and to longer
operating-room time.
Conclusions Despite shorter hospital stays,
in-hospital charges and costs for laparoscopically
assisted vaginal hysterectomy are higher than for
either alternative procedure, because of the dis­
posable supplies that are typically used and the
longer operating-room time. (N Engl J Med 1996;
335:476-82.)

APAROSCOPICALLY assisted vaginal hys­
terectomy has come into widespread use,
primarily because morbidity is presumed
to be less when the large abdominal inci­
sion and invasive intraabdominal manipulations as­
sociated with total abdominal hysterectomy are elim­
inated.13 The laparoscopically assisted procedure has
been controversial, however, largely because of con­
cern that it is sometimes used instead of total vaginal
hysterectomy, generally considered the simples? and
least morbid method of removing the uterus, and
because the costs of the laparoscopic procedure may
be substantially higher than those of either alterna­
tive procedure.4’8
According to several studies,9 n the average hos­
pital stay of patients undergoing laparoscopically as­
sisted vaginal hysterectomy is shorter than that of
patients undergoing the other procedures. However,
these findings have often been confounded by dif­
ferences in the surgical procedures performed in
conjunction with hysterectomy, such as salpingooophorectomy, adhesiolysis, and repair of pelvicsupport defects. Previous assessments of variation in
operating-room time for hysterectomy, as well as in
the costs of different procedures, have also not
determined whether the variation was due to dif­
ferences inherent in the three techniques or to dif-ferences in the types of patients undergoing each ■
procedure.
To assess costs, hospital charges, and use of re­
sources associated with alternative techniques of
hysterectomy, we grouped hospitalizations for hys­
terectomy on the basis of the surgical procedures
performed in addition to the removal of the uterus.
We then compared overall and specific costs and
charges associated with the various techniques^

©1996, Massachusetts Medical Society.

From the Department of Gynecology, Greater Baltimore Medical Cen­
ter. Baltimore (J.H.D., P.M.H.); and Coming HTA, Washington. D.C.
(R.I.G., M.M.M., E.P.S.). Address reprint requests to Dr. Doner at
the Depanmcnt of Gynecologv, Greater Baltimore Medical Center. 6569
N. Charles St., Suite 307, Baltimore, MD 21204.

476

I



August 15, 1996

COSTS AND CHARGES ASSOCIATED WITH THREE ALTERNATIVE TECHNIQUES OF HYSTERECTOMY

METHODS

*

Study Site

>

Pie Greater Baltimore Medical Center, a 372-bed community
teaching hospital, has the largest gynecologic-surgen' senice in
Maryland. Laparoscopically assisted vaginal hvsterectomv has been
performed since February 1, 1990. In'1993 and 1994, the period
ot this study, 21,610 gynecologic-surgery procedures were per­
formed at the center.
Sources of Data

Three sources ot data were used in the study: a computerized
tile containing hospital-discharge abstracts, with diagnoses, pro­
cedures, and charges for all hospitalizations in Maryland, as re­
ported to the state Health Services Cost RcA'iew Commission; a
computerized data base maintained by the medical center, con­
taining information on diagnoses and procedures, use of resourc­
es, and charges submitted tor each senice provided by the hos­
pital; and patients’ hospital records, which we reviewed in a
structured fashion.
Selection of Patients

Patients who underwent hysterectomy in 1993 and 1994 were
identified in the data bases of the Health Sen ices Cost Review
Commission and the medical center when one of the following
procedure codes established in the International Classification
of Diseases, 9th Rxvision, Clinical Modification (ICD-9-CM) was
assigned: 68.4 for abdominal hysterectomy, 68.5 for vaginal
hysterectomy, or 68.5 plus 54.21 (indicating laparoscopy) for
laparoscopically assisted vaginal hysterectomy. We reviewed all
secondary procedures listed in the data base of the medical center
for the 1420 patients identified, and we excluded 182 patients
who had undergone one or more major secondary procedures
unrelated to hysterectomy (such as partial colectomy or appen­
dectomy). We also excluded 45 patients who had undergone ci­
ther radical or subtotal hysterectomy (ICD-9-CM codes 68.6 and
68.o), since our focus was on hysterectomy performed to treat
benign conditions. Ot thc'rehiaining 1193 patients, hospital rec­
ords were not available for 39 (3 percent). Alter a detailed review
ot the charts, we excluded an additional 105 patients because
their hysterectomies had been performed to treat cancer. Thus,
1049 patients (74 percent ot those who underwent hysterectomy
at the center in 1993 and 1994) were included in the analysis.
The 1049 procedures were performed by a total of 96 surgeons.
Demographic and Clinical Data

I1

Data were abstracted from the hospital charts bv one of five
registered nurses. The information collected included each pa­
tient s age, height, weight, medical history, indications for surgery,
operative procedures (including information on whether a vaginal
hysterectomy had been converted to an abdominal procedure),
intraoperative findings and complications, postoperative compli­
cations, and results of pathological studies, including the uterine
weight. Each indication for hysterectomy was classified in one of
four categories: (1) uterine abnormalities, defined as anv clinically
diagnosed abnormalities that involved the uterus (such as myo­
mas) or resulted in uterine dysfunction (such as a bleeding disor­
der); (2) adnexal abnormalities, defined as any clinically diag­
nosed abnormalities involving an ovary, a fallopian tube, or both
(such as an ovarian cyst or a tubo-ovarian mass); (3) abdominal
or pelvic abnormalities, defined as any other abnormalities of the
abdominal or pelvic region, such as endometriosis, pain, pelvic in­
flammatory disease, or a mass that was not described as adnexal
or a uterine fibroid; and (4) abnormalities of pelvic relaxation, de­
fined as any abnormalities resulting from a support defect of the
pelvic fascia. Tims, a patient could have had more than one indi­
cation for surgery. An intraoperative finding of extensive adhe­
sions was recorded if the operative note described adhesions as
numerous, massive, thick, or requiring extensive adhesiolvsis for

the surgeon to “tree up" or gain access to a peine structure. From
the records of anesthesia and the nursing records, data were col­
lected on the rime spent in the operating room, under anesthesia,
and in the rccovcn* room.
Economic Data

Charges made by the facility for ail resources used during a
hospitalization, such as operating-room time and supplies, and
for the hospitalization as a whole, were obtained from the medi­
cal-center data base. Facility charges for anesthesia were based on
the time under anesthesia and on per-minute charges as provided
by the finance department of the medical center.
To estimate the professional fees for each hospitalization, we
assigned relative-value units based on Medicare’s resource-based
relative-value scale to all ICD-9-CM procedure codes that had a
professional component, as well as to surgical-pathology services
with professional-senice components but no ICD-9-CM proce­
dure code. The estimated professional fees for anesthesia were
based on the time under anesthesia. The anesthesia time, ex­
pressed in minutes, was divided by 15 and rounded to the nearest
whole number to obtain relarive-value units for anesthesia time
for each patient. A base number of relative-value units for each
procedure (8 for abdominal hysterectomy and 6 for vaginal hys­
terectomy and laparoscopically assisted vaginal hysterectomy) was
added to the relative-value units for time to obtain the total num­
ber of relative-value units for anesthesia. The rclathe-value units
were converted to charges with ^Medicare’s 1993 and 1994 con­
version factors for Baltimore. The estimated professional fees for
each hospitalization were then totaled.
The medical centers costs, as opposed to its charges, wrre also
estimated for each hospitalization. We computed the proportion
of the total hospital charges that was attributable to each of 64
cost centers (such as the operating room, medical and surgical
supplies, the pharmacy, and the pathology department). We then
selected the cost centers that accounted for the 10 largest shares
of the total charges (for example, the operating room accounted
for 30 percent of charges). For each of these 10 cost centers, we
evaluated all resources that were used in the care of the patients in
our sample and ranked the resources in terms of the proportion of
the cost center’s total charges that was attributable to each re­
source (for example, within medical and surgical supplies, sutures
accounted for 14 percent of charges, endoscopic staplers for 12
percent, and so on). We then identified the 10 resources with the
highest total charges in each cost center or, if 90 percent of the
charges in a cost center were attributable to fewer than 10 resourc­
es, the resources that accounted for 90 percent of the charges.
----- We estimated the direct and indirect costs of these resources
with data from the finance department. For example, we calculat­ --- 4
ed the direct cost for the operating rooms by dividing the totah
operating-room expenses by the total number of minutes patients*
spent in the operating room. The costs of supplies and pharmacy
senices were estimated by reducing the average charges for each
supplied item and pharmaceutical by the amount of the hospital’s
markup. Costs for the use of the blood bank and histologic tests
were derived from the cost-accounting system of the hospital lab-^
oratory.
For each of the top 10 cost centers, we computed a ratio of
cosjs to charges by dividing the total costs of the top 10 resources
in that cost center by the total charges for the same resources.
The cost-to-chargc ratio for each cost center was applied to all
resources in that cost center. We then calculated a weighted a\xrage cost-to-chargc ratio for the top,IQ cost centers by taking the
total costs for each center and dividing them by the sum of the
chargesjpr all 10 cost Tenters. Finally, we applied this weighted
ax erage cost-co-charge ratio to each of the 54 cost centers that
were not included among the top 10 cost centers. The cost-cen­
ter—specific estimates of cost were summed to obtain an estimate
of the total facility costs for each hospitalization. We did not as­
sign costs (as opposed to charges) to the professional-service
components of the procedures.

Volume 335

Number 7

477

-•M •

——-s-i,



-

;



|

The New England Journal of Medicine

I

-Statistical Analysis

To account for differences in operative complexity and the
postoperative care of patients who underwent different surgical
procedures, each patient was assigned to one of the following sev­
en mutually exclusive subgroups on the basis of the surgical pro­
cedures performed in conjunction with the hvsterectomv: (1) no
related secondary procedure; (2) repair of vaginal prolapse, surgi­
cal treatment for urinary incontinence, or both, but no other pro­
cedure (“surgical repair”); (3) adhesiolvsis, but no other proce­
dure; (4) salpingectomy, oophorectomy, or both, but no other
procedure; (5) surgical repair and salpingectomy, oophorectomy,
or both; (6) salpingectomy, oophorectomy, or both, and adhesi­
olysis; and (7) other procedures.
The characteristics and outcomes of the patients treated bv
each technique of hysterectomy were assessed separately for each
of the first six categories, and for all the patients in the’study, on
an mrention-to-treat basis. For example, a patient whose proccdure began as laparoscopically assisted vaginal hysterectomy but
v^-^Q.nvcried _t° abdominal hysterectomy because _of technical
difficulty or a complication was considered to have undergone
laparoscopically assisted vaginal hysterectomy. Pairwise compari­
sons were performed by Student’s t-test (for continuous variables)
or the chi-square test (for categorical variables).^ Separate multi-

variate linear regression analyses were performed for three catego­
ries — all patients, those with no related secondary procedures,
and those who undenvent salpingectomy, oophorectomy, or both
— in which there were enough patients for the independent as­
sociation between the technique of hysterectomy and various
economic outcomes (such as operating-room tune and facility
charges) to be assessed, with control for the patient’s age, the
number of coexisting medical conditions, and uterine weight.13
Finally, to assess the association between the use of disposable
instruments and the cost of laparoscopically assisted vaginal h\-sterectomy, the patients who undenvent that procedure were di­
vided into three mutually exclusive subcategories on the basis of
the type of supplies used in the operation. Disposable instru­
ments can be used in each step of laparoscopically assisted vaginal
hysterectomy. If such instruments (that is, an endoscopic stapling­
device, an endoscopic hemoclip, staple-reloading cartridges, dis­
posable hand instruments, and disposable trocars) were used dur­
ing even’ step of the procedure, we categorized tire surgery as
performed with disposable instruments. If the surgeon used sev­
eral types of these disposable instruments, bur nor all (for ex­
ample, if he or she used disposable trocars but relied on electro­
cautery or sutures for hemostasis), the procedure was classified as
one that used a combination of disposable and nondisposable in­
struments. The third category’ was one in which no disposable

Table 1. Characteristics of the Patients According to
Technique of Hysterectomy.*

Characteristic

Age (}T)
Body.-mass indexj
Coexisting conditions
No. per patient
Percent of patients
Preoperative indications (% of patients)
Uterine abnormality
Adnexal abnormality
Abdominal or pelvic abnormality
Pelvic relaxation
Secondary procedures (% of pnrirnr*)
None
Surgical repair
Adhesiolysis
Salpingectomy, oophorectomy,
or both
.4s only secondary procedure
With repair of vaginal prolapse
With adhesiolysis
Intraoperative findings (% of patients)
Extensive adhesions
Endometriosis
Uterine weight (g)

Laparoscopically
Assisted Vaginal
Hysterectomy
(N = 273)

Total Vaginal
Hysterectomy
(N = 210)

Total Abdominal
Hysterectomy
(N = 566)

44±8f
26.0±5.4§

50±13
26.2±5.0

45±9|
28.4±6.9f

0.6±0.9
58

0.4±0.7t
57

0.4*0.81
461||

f
V

78f

56

84f

7t tt

0.48

52|
10f

17

60

45f
6t

32tt
1
6

41
38
0

21
0.4
5

43
3
0.73

8
11
0.48

45
3
0

21t
26U
171.1±159.2tt

2
1
113.3*84.1

27|
15

1st

335.5*343.9

•Plus-minus values arc means ±SD.
fP<0.001 for the comparison with total vaginal hysterectomy.
tBody-mass index was calculated by dividing the weieht in kilograms
by the square of the height
in meters.

e

§P<0.001 for rhe comparison with abdominal hysterectomy.
1P=0.01 for the comparison with total vaginal hysterectomy.
||P=0.002 for the comparison with abdominal hysterectomy.
••P=0.05 for the comparison with abdominal hysterectomy.
ftP=0.007 for the comparison with abdominal hysterectomy.
ttP<O.OOl for each paiyivisc comparison.

I
478



August 15, 1996



:



I ' ■
COSTS AND CHARGES ASSOCIATED WITH THREE ALTERNATIVE TECHNIQUES OF HYSTERECTOMY

f

laparoscopic instruments were used.
i
Costs of medical and surgical
supplies, operating-room time, and total charges for these subcat­
egories were compared by Student's t-test. All statistical analyses
were performed with SAS version 6.10.

dergoing vaginal hysterectomy, and 4 percent of
those undergoing abdominal hysterectomy (P = 0.02
for the comparison of the first and second groups.
J and P = 0.16 for the comparison of the first and
RESULTS
i third groups). Twelve percent of laparoscopicallv as­
sisted vaginal hysterectomies were converted to open
Characteristics of the Patients
laparotomy, as compared with 2 percent of total vag­
Of the 1049 patients, 273 (26 percent) underwent
inal hysterectomies (PcO.OOl).
laparoscopically assisted vaginal hysterectomy, 566
Use of Resources and Costs of Care
(54 percent) underwent abdominal hysterectomy,
and 210 (20 percent) underwent vaginal hysterecto­
The mean hospital stay of the patients who under­
my (Table 1). The group undergoing vaginal hyster­
went laparoscopically assisted vaginal hysterectomy
ectomy was slightly older, on average, than the other
(2.6 days) was significantly shorter than that of those
two groups, which were similar in age. The patients
who underwent vaginal hysterectomy (2.9 days) or
who underwent abdominal hysterectomy tended to
abdominal hysterectomy (3.9 days) (P<0.02 for all
have higher body-mass indexes and heavier uteri than
comparisons). The mean stay for patients undergo­
the patients in either of the other groups. In 89 per­
ing laparoscopically assisted vaginal hysterectomy
cent of patients with a uterine weight of at least
was more than one day shorter than that for patients
400 g, abdominal hysterectomy was performed.
undergoing abdominal hysterectomy both when no
Secondary procedures related to hysterectomy were
secondary procedure was performed and when sal­
performed in 79 percent of the patients undergoing
pingectomy, oophorectomy, or both were performed,
abdominal hysterectomy, 68 percent of the patients
regardless of whether adhesiolysis was performed.
undergoing laparoscopically assisted vaginal hyster­
For the patients undergoing laparoscopically assisted
ectomy, and 59 percent of the patients undergoing
vaginal hysterectomy and those undergoing vaginal
vaginal hysterectomy (Table 1). The most common
hysterectomy, the mean stay was similar when no
secondary procedure performed in conjunction with
secondary procedure was performed. Salpingectomy,
laparoscopically assisted vaginal hysterectomy and
oophorectomy, and the two together did not influ­
abdominal hysterectomy was salpingectomy, oopho­
ence the mean stay, but adhesiolysis and repair of
rectomy, or both, with or without adhesiolysis. The
prolapse both increased it.
most common secondary procedure performed in
Despite the shorter mean stay with laparoscopical­
conjunction with vaginal hysterectomy was repair of
ly assisted vaginal hysterectomy, the mean total charg­
vaginal prolapse of surgical treatment for inconti­
es (including both facility'charges and professional
nence.
fees) were highest for the patients undergoing that
procedure
(Table 2). The average total charges for
Surgical Outcomes
laparoscopically' assisted vaginal hysterectomy' were
Intraoperative complications occurred in 6 per­
higher than those for abdominal hysterectomy* by
cent of the patients undergoing laparoscopically as­
$1,032 (P<0.001) and higher than those for vaginal
sisted vaginal hysterectomy, 2 percent of those unhysterectomy by $1,895 (PcO.OOl). The total charg-

Table 2. Total Charges for Various Subgroups of Patients, According to Technique of Hysterectomy.*

Subgroup

All subgroups
No secondary procedure
Surgical repair only

Adhesiolysis only
Salpingectomy, oophorectomy, or both
As only secondary procedure
With repair of vaginal prolapse
With adhesiolysis

Total No. Laparoscopically Assisted
Studied
Vaginal Hysterectomy

Total Vaginal
Hysterectomy

Total Abdominal
Hysterectomy

NO.

TOTAL CHARGES (S)

NO.

TOTAL CHARGES (S)

NO.

TOTAL CHARGES (S)

1049
293
85
43

273
87

6116*1816f
5804±1581t
78 56 ±3642$
6674±2389§

210
87

4221 ±1174
3522±737
4673*920

566'

26

5084±1768
4548±763
5808*1939
5078±1429

389
49
181

117

6030 ±1681
7694 ±1486l|
6236±2044

17
23
I

255
19
143

4890± 12521
6004±1210’*
5689±2724

4
17

7
37

•Plus-minus values arc means ±SD.

79
0

3976 ±702
5077±1344
6448

119
2

fPCO.OOl for each pairwise comparison^
= 0.009 for the comparison with abdominal hystcrcctomv.
||P“ 0.007 for the comparison with abdominal hysterectomy.

tP<O.OOl for the comparison with total vaginal hysterectomy.
TP = 0.003 for the comparison with total vaginal hysterectomy.
•*P = 0.03 for the comparison with total vaginal hysterectomy.
I

I
Volume 335

Number 7

-

479



lv*T

49

The New England Journal o/Medicine

i

. es for laparoscopically’ assisted vaginal hy’sterectomv
comparisons). These differences were tL.
“___regardsimilar
w’ere $1,140 higher than those for abdominal hvsterless of which secondary procedures were performed,
.
I,
ectomy’ when oophorectomy, salpingectomy, or both
or whether any were performed.
w’ere performed and $1,256 higher w'hen no related
We compared the total charges, facility charges,
secondary’ procedure was performed.
and costs for medical and surgical supplies that were
The differences in facility’ costs associated with
associated with laparoscopically assisted vaginal hys­
hospitalizations for the three ty'pes of hvsterectomy
terectomy according to whether the procedure was
paralleled the differences in charges (Table 3). The
performed with disposable supplies, nondisposable
mean overall facility’ costs for laparoscopicallv assist­
supplies, or a combination of the two (Table 4). Mean
ed vaginal hysterectomy were $1,167 higher than
costs for supplies were higher by $1,496 when pro­
those for abdominal hysterectomy w’hen no related
cedures were performed with disposable supplies
procedure was performed and S 1,060 higher when
than when they were performed with nondisposable
salpingectomy, oophorectomy, or both w’ere per­
supplies. Despite the potential savings of time asso­
formed.
ciated with the use of an endoscopic stapler, the av­
In part, the higher costs and charges for laparo­
erage operating-room time with disposable supplies
scopically’ assisted vaginal hy’sterectomv w’ere due to
wras greater, not less, than for operations performed
longer operating-room times. When there w’as no re­
with nondisposable supplies. Among the patients
lated procedure, laparoscopically' assisted vaginal hvs­
w’ho underwent laparoscopically assisted vaginal hys­
terectomy’ required 35 minutes more operating-room
terectomy' with no related secondary’ procedure, the
time than abdominal Hysterectomy’ (158 vs. T23~minmean operating-room time wras 165 minutes w’hen
utes) and 70 minutes more than vaginal hysterectodisposable supplies were used (67 patients), 143
my (
minutes) (P<0.00T for both comparminutes w’hen nondisposable supplies w’ere used
jsons). When salpingectomy, oophorectomy, or both
(7 patients), and 122 minutes when a combination
were performed, but not adhesiolysis, the mean op- . of the tw'o w'as used (13 patients). xnm
Thus,uverau
overallrafaerating-room time needed^jJaparoscopLcalRassisL- I cility charges and total charges^with proi^
ed vaginal hysterectomy w’as 46 jninutes more than ; radded)
J J “ w’ere substantially
................... higher
" •
- laparoscopically
for
for abdominal hysterectomy and 72 minutes more
assisted vaginal hysterectomy only when disposable
than for vaginal hy’sterectomv. A similar amount of
supplies w’ere used.
recovery-room time w’as needed for all three tech­
Multivariate regression analyses
w’ere
performed
9
4
iWA 1 * W4 to
Lvz
niques both when there w’as no secondarv procedure I compare the three techniques of hysterectomy with
--- 1--------- ---------UtOLUlUULUlin *>1L1I
and w’hen salpingectomy, oophorectomy, or both j respect to costs and the use of resources, with adjustwere performed.
, x
1■ rnent
ment for atre.
age, rhe*
the number nFmevicrinrr
of coexisting medical con­
_____ i___
The average charge for medical and surgical sup­ | rlirionc
ditions, uterine tt'/»irrkr
weight, and the secondarv
proce­
plies was $1,190 higher for laparoscopically assisted
dures performed. These adjustments had little effect
vaginal hysterectomy’ ($1,485) than for abdominal
on the results. *Regression analyses also demonstrathysterectomy ($295), and $1,251 higher than for ! ed that patients> at least 60 years old had mean total
vaginal hysterectomy ($234) (PcO.OOl for both ' charges that
1-------w’ere $938 higher than those of patients
^7

L

1

Table 3. Facility Costs for Various Subgroups of Patients, According to Technique of Hysterectomy.*

Subgroup

Total No.
Studied

All subgroups

1049

No secondary procedure
Surgical repair only

293
85

Adhesiolysis only

43

Salpingectomy, oophorectomy, or both
As only secondary procedure
W’ith repair of vaginal prolapse
W’ith adhesiolysis

389
49
181

Laparoscopically Assisted
Vaginal Hysterectomy
FACIUTV COSTS (S)

NO.

FACILm costs (S)

NO. F.MTI.m- COSTS (S)

273
87

210
87
79

3116=969
2626=659

4
17

4914± 17101
4642±1496t
6397±35I5t
5449±2207§

566 . 3954 ±1601
119
3475 ±676
2
4110±1485
26
3960±1312

117
7
37

4851± 1622
6177±1376t**
5010±1890

17
23
1

tl’<0.001 for the comparison with total vaginal hysterectomy.

V<0.001 for the comparison with abdominal hysterectomy.
1 -0.005 for the comparison with abdominal hysterectomy.

l



August 15, 1996

Total Abdominal
Hysterectomy

NO.

•Plus-minus values are means ±SD.

480

Total Vaginal
Hysterectomy

3400= 771

0
2978 = 621

255
19

3747=1242
5224

143

3791 ±I128||
4601 = 1082 ft
4510±2499

fP<0.00I for each pairwise comparison.
§P = 0.008 for the comparison with abdominal hysterectomy.
||P = 0.004 for the comparison with totjfl vaginal hysterectomy.

• ft I’83 0.024 for the comparison with toui vaginal hysterectomy.
I
I

1

1

■I- •

COSTS AND CHARGES ASSOCIATED WITH THREE ALTERNATIVE TECHNIQUES OF HYSTERECTOMY

II
!

Table 4. Charges and Costs for Lm’arosconcally .Assisted Vaginal
Hysterectomy According to Whether the Sltplies Used Were DisrosABLE.*

No. OF

Type of Suppues

Procedures

Total Charges (SI

Disposable
Nondisposable
Both combined

210
10

6419= 18 ISf
4563=7089$
5208=1401

53

i

Cost of Medical
ano Surgical
Facility Charges IS)
Supplies (S)
5514±1770f

3644=6115§
4295=1328

1782±I089f
286=515§
581=3441

•Plus-minus values are means =SD.
tPcO.OOl for rhe comparison with procedures using disposable and nondisposable supplies in
combination.

$P<0.002 tor the comparison with procedures using disposable supplies.
§P<0.001 for the comparison with pnicedures using disposable supplies.
l?=0.01 tor the comparison with procedures using nondisposable supplies.

i
IIt ’is not surprising
t ‘
1 costs associated
; ; with
that the
; laparoscopically assisted vaginal hysterectomy’ were
higher than those of the two alternative procedures,
Because laparoscopically assisted vaginal hvstcrectomy combines both abdominal and vaginal approachi es, two sets of instruments and drapes and two difter' en t operating, con figurations are"required, increasi ngi both time and labor. xMoreovcr, when uterine weight
is 400 g or more, the vaginal portion of the opera­
tion often becomes more difficult, increasing the
i operating-room time,
j
Few
Few
rew disposable
disposable inst
uisposaoic
instruments are routinely used in
:
either
abdominal
hysterectomy or vaginal hysterect
• tomy. With laparoscopically assisted vaginal hyster. ectomy, however, disposable instruments are availi able for every step of the laparoscopic portion of the
I procedure. The most expensive of these instruments
are the endoscopic stapling devices. Some surgeons
DISCUSSION
believe that stapling instruments substantially reduce
Despite the reduced invasiveness and shorter hos-the operating time required for this portion of the
pital stay associated with laparoscopically-assisted vag­ —hysterectomy. In addition, disposable instruments,
inal hysterectomy, we found that the operating-room
are always clean and sharp and are designed to facil­
time, anesthesia time, cost of supplies, facility costs
itate specific steps in the operation. Nonetheless,
and charges, and total charges (facility charges plus
there was no overall reduction in operating time
professional fees) for that procedure were substantial­
when disposable instruments were used for all steps
ly higher than those for either vaginal hysterectomy
in the laparoscopic portion of the procedure.
or abdominal hysterectomy. The cost of a hospitaliza­
We compared costs and use of resources in the
tion for laparoscopically assisted vaginal hysterectomy
care of patients defined as clinically similar on the
was higher regardless of which related surgical proce­
basis of the secondary procedures (if any) performed
dures were also performed, or whether any such pro­
in conjunction with hysterectomy. We adjusted the
cedures were performed. In addition, the procedure
analysis for age, the number of coexisting condi­
was substantially more expensive when disposable, as
tions, and uterine weight. As a result, we believe the
compared with nondisposable, supplies were used in
differences' in cost were attributable to differences
every step of the laparoscopic portion of the pro­
between the techniques of hysterectomy used, rather
cedure. When laparoscopically assisted vaginal hys­
than differences in the type of patients treated. To
terectomy was performed with either nondisposable
make our comparisons accurate and exclude patients
supplies or a combination of disposable and nondis­
who underwent more substantial surgery for cancer,
posable supplies, the facility charges were not sub­
we reviewed more than 1000 patients’ hospital charts
stantially higher than for abdominal hysterectomy.
in detail, instead of relying solely on computerized
under the age of 40, and facility charges that were
Sb 17 higher, after adjustment for the number of coexisting conditions and for uterine weight. After we
controlled for uterine weight and the number of coexisting conditions, an increase of 10 years in age
(for example, when a patient 65 years old was compared with a patient 55 years old) was associated
with an increase of S246 in facility costs (P<0.001).
In addition, patients with a uterine weight of at least
400 g had mean total charges and mean facilitv
charges that were S280 higher than those of patients
with a uterine weight below 400 g, after we controlled for age and the number of coexisting condi-•
j.:_____
” . . ti.. .
x
it
i
T
tions. Finally, after we controlled for age and uterine
weight, patients with one or more coexisting condi
j.- ­
tions had mean facility costs S239 higher than those
of patients with no coexisting conditions (P = 0.01).

Volume 335

r.

Number 7

481

__ f

The New England Journal o/Medicine

administrative data. These reviews enabled us to ob­
tain data on age, weight, and indications for surgery
for each patient and to identify procedures that be­
gan as laparoscopically assisted vaginal hysterectomy
but were converted to abdominal hysterectomy and
thus were coded and billed as the latter on the dis­
charge abstracts.
We estimated facility costs, as well as facility charg­
es, using cost-to-charge ratios specific to each cost
center. We could thus draw conclusions about the
costs associated with each technique from the per­
spective of the hospital as well as the insurer.
The most noteworthy limitation of our analysis is
that we studied only a single institution. The costs
and charges for the three techniques of hysterectomy
may differ at other institutions, but it is likely that our
major conclusions are generalizable. We examined the
experience of nearly 100 surgeons but did not control
for ainerences
differences among them in technical expertise.
ror
i
Judgments about the preferred approach for hvs- j
terectomy should be based on several factors, includ- i
ing the indications for the procedure, the clinical
characteristics of
characteristics
of the
the patient
patient (such
(such as
as estimated
estimated uteruter-

'

ine size and weight), and the morbidity and cost associatcd with the technique. When vaginal hvsterec- .
tomy can be performed, our analysis confirms that it '
is the least costly approach and that intraoperative
complications are less frequent with it than with lap- I
aroscopically assisted vaginal hysterectomy. Laparoscopically uddidivu
assisted vaginal
vagmai hvsterectomv
nysrcrecromy may offer
otter |
patients several advantages over abdominal hystcrectomy, such as reduced postoperative discomfort, a
shorter period of recuperation after discharge, and
the chance to avoid a major abdominal incision, but
our analysis suggests that when disposable instru- ;
ments are used for all steps in the laparoscopic por- ;
tion of the procedure, the in-hospital costs to both i
the hospital and the insurer are substantiallv higher '
for the laparoscdpic“pfocedure?
: ~
"
It is not likely that using disposable instruments
throughout laparoscopically assisted vaginal hvsterec- !
tomy confers enough advantages over abdominal ;
hysterectomy to justify the added cost. Whether the ‘
benefits to the patient of the laparoscopicallv assisted |
technique without the use of disposable instruments I
are worth the extra cost is a question requiring a val- j
ue judgment. A prospective study now under way ,
w ill evaluate patients’, quality of life, ability to rrrnrn
^Q^pfk, and prodiiciiYjty after hvsterectomv, to clnrifCrHe relative benefits and cost'effectiveness of the
two procedures.

f
I

482



August 1 5, 1996

Supported by Greater Baltimore HcalthCarc. Baltimore.
Dr. Dorsey has been a consultant to the United States Surgical Corpo­
ration, Norwalk, Conn., which makes disposable laparoscopic instruments,
and owns stock in C.R. Bard, which has a subsidiary that makes laparo­
scopic instruments.
Wc arc indebted to Lauric Beyer, C.P.A., and Brian Jacquc ofrhe
finance department of the Greater Baltimore Medical Centerfor as­
sistance in compittiiia cost-to-charpc ratios, and to Susanne Bendlcr,
Carolyn Pomponio, and Vicki Watkins for technical assistance.

REFERENCES
1. Reich H, DcCaprio J, McGIvnn F. Laparoscopic hvstcrccTomv. J Gvnccol Surg 1989;5:213-6.
2. Langcbrckkc A, Skar OJ, Umcs A. Laparoscopic hvstcrecromv: initial
experience. Acta Obstet Gynecol Scand 1992;71:226-9.
3. Boike GM, Elfstrand EP, DcIPriorc G, Schumock D, Holley S, Lurain
JR. Laparoscopically assisted vaginal hysterectomy in a university hospital:
report of 82 cases and comparison with abdominal and vaeinal hysterecto­
my. Am J Obstet G\-neco1 1993;168:1690-701.
4. Thompson JD. Hysterectomy. In: Thompson JD, Rock JA, cds. Te
Linde’s operative gynecology. 7th cd. Philadelphia: J.B. Lippincott, 1992:
663-738.
5. Pitkin RM. Operative laparoscopy: surgical advance or technical gim­
mick? Obstet Gynecol 1992;79:441-2.
6. Grimes DA. Technology follies: the uncritical acceptance of medical in- .
novation. JAMA 1993;269:3030-3.
7- Dorsey JH. ‘Technology follies’: curtain call. JAALA 1993:270:2298-9.
8.’ Harris
Hams *MB,
Olive DL
DL.’ Chan
Changing
introduc8
MB’olivc
Pins hysterectomy patterns atter imroduc1994:171:340-4;
MsiSKd va!?inji
'

.......J
9. Dorsey JH. Steinberg EP, Holtz. PM. Clinical indications for hystcrcctomy route: patient characteristics or physician preference? Am J Obstet
Gynecol 1995;173:1452-60.
10. Ncz.hat F, Nezhat C, Gordon S. Wilkins E. Laparoscopic versus ab­
dominal hysterectomy. J Rcprod Med 1992:37^247.50.
11. Daniell JF, Kurtz BR, McTavish G. ct al. Laparoscopically assisted vac-

-•'Xled
kd 1993;38:537-42.

Hall. 1984.

analysis. 2nd cd. Engicuood cliffs. n.i.: Prentice

13. Rawlings JO. Applied regression analysis: a research kx>I. Pacific
Grove, Calif; Wadsworth & Bn>oks/Cole Advanced Books & Software,
1988.

II

mi inn

*600

■I. mi i<i

__ I____ |!|||||

T

"

,

I.

I■

1

Hill I III

< ^THEI^CET,MARCHK»m

RECOMMENDATIONS FOR FUTURE STUDIES

The case for low-protein diets in CRF is not established in man
For further study we make the following recommendations. (1).
Patients should be proven to have progressive renal failure with no
obvious reversible factor before administration of a low-protein
diet. (2). The rate of decline of renal function should be assessed
over several months. This will allow for the placebo effect and
ensure treatment of conditions such as hypertension. (3). Renal
function should be monitored by isotopic riraranrrs (4).
Assessment of nutrition - should include anthropometric and
biochemical measurements. (5). Parient compliance should be
assessed by an experienced dietitian/and also by the measurement of
urea nitrogen appearance. (6). If a randomised trial is undertaken,
groups ofpatients should be matched for age, sex, diagnosis, rate of
progression, degree ofrenal failure, hypertension, and proteinuria.
Control and experimental diet groups should be treated and
followed up in the same way, and should be kept apart at clinics to
prevent inadvertent “crossover” of diets. (7). In future trials we
would favour the less restricted diets (standard 0*6 g/kg protein
• intake) since they are a more realistic option for large-scale use. (8).
Follow-up should be for at least two years.

We thank A. M. Davison for allowing us to quote data from the MRC


REFERENCES

1. Giordaoo C. Prorein resniedon in chronic renal failure. Kidney ha 1982; 22:401-08.
2. Klahr S, Beurken J, Purkcnon MP. Role of dietary faoon on the progreaaion of
. chrome renal ditcaic. Kidney ha 1983; 24: 579-87.
3. Giovanneni S. Dietary treaanem of chronic renal failure. Why is it ooc used more
frequently? Neyirrva 1985; 48: 1-12.
-4.-Maachx> G, Oldrizzi L, Tcadcore N, et at Effects of dimry prorem and phosphate
restnenou on the progroaiaa of early renal failure. Kidney int 1982; 22: 371-76.
*5? Danton AM, Cameron JS, Kerr DNS, er aL The natural historyofrenal funoion in
untreated idiopathic menbranous (fomreuionephnna in adnha. CEn Neyitni 1984;
.22:61-67.
6. Ane RS, Ramm JM, Owen JL, a at The namrai history of dtromepydonephnna in
the adult Qnaet ] Med 1982; 284: 396-410.
7. Oldrixri L, Rncm C, Vahro E, a aL Progrcsocn ofrenal foDwe a prexnts with renal
disease of dhrene etiology on protein restricted dire. Kidney ha 1985; 27:553-57.
8. Ideison A, Smithime N, Smith G, Harringron JT* Prognosia in steroid treated
idiopathic nephrotic syndrome iaadnha. re* Intern Med 1977; U7:891-96.
9. Kincaid-Smith P, Bcdor G. Reflux nephropathy and chronic atrophic pydonephritis.
■ Areriew. J InfeaDit 1978; 121:774-80.
- .
• J 0. Rosman J B, Mdjcr S, Sharer WJ, ex aL Prospccrivc randosniaed trial of early dietary
■ protein restnoion in chnaiic rrmlfwiMTr la~nt 1984; ik 1291-95.
11. El Nahas AM, Masters-Thomas A, Brady SA, er aL Sdcctnre e£ta offow protein diets
in chronic renal djaema.£r Med J1984; 288t 1337-41.
IXMhch WE, Walter M, Buffington GA,'er.aL:A aimpie mated for reriniarint
progression of chrome renal fidhire. Lancet 1976; ii: 1326-28.
13. Rutherford WE, Blondin J, Miller JP, a aL Chronic progressiva renal disenae: rare of
-change of scrum creatmine coocennanom Kidney Im 1977; 11: 62^-70.
14. Lcdingtam JGG, Han G. The optimum thne to sun regular haemodialysi*. .In:
Davnon AM, cd.Diatyan review. Phihdrlphir JB Lippmcoo. 1978:22-31.
15. Oksa H, Pasternack A, LuiouialaM, ex aL Progrewwn ofchromcrenai fidhire. Neydren
1983;35:31-34.

16. Mixch WE. The inflnrace ofthe din on the progrereion of renal msnffirirncy. Anau
RenMed 1984; 35:24W4.
17. Bldier RE, Schedl HP. CreaiiniDe excrexion: variabtlixy and relarionfoips to dire and
body sue. J Lab CEn Med 1962; 5»t 945-55.
18. Mitch WE, Collier W, WaberM. Crtatmmemetabolismwchromctvnalfaihxre. CEn
Sd 1980;58:327-35.

19. Lucre PA, Meadows JH, Roberts DE, Coles GA. The risks and benefits of a low
proremwsaential amino acid-ktto add diet. Kidney int (in pressL
•» 20. Guoerrez A, Qureshi A, Bcrptrttn J. Influence of meat-free dte on the urinary
- excretion of 3-Methyi Histidine <3MH) and creatinine(Cr) in chrome renal fiuhire.
■ Kidney ha 1985;2t:J89(abrer).
...
21. Sheraesh O, Gdlbctz H, Krte JP, Myers ED. Linmationa of creanmne re a fihranon
marker in giomenxioputhic presents. Katoy far 1985; 28:830-38.
'
*22. Bareotn G, Morelli E, Gianaom A, et aL Ffltm of low prorein rti—^ errammw
dearance of normals and chronic uraemMS, Kstery int 1984; 28t 498 (abotr).
*23. Fonseca V, Weerakoon J, Mikhailidis DP, et aL Plasma errsraww and creatmine
clearance of normals and chronic uraemia. Xabary far 1984; 28r 498 (atetr).
24. El Naha* AM, Gornacz G, Hutchings V, a aL Effects of dituey prorem on renal
haemodynamics. CEn Sd 1984; Mt P36 (abstr).
25. Lubowuz H, Slatopoiaky E, Shankd S, ex aL Glomerate filtration tare. JAMA 1967;
189: 100-04.

. *26. Barsotti G, Giannoni E, Morelli E, a *L The dedine of renal funcoou stowed by very

ChikLHealth
COST OF NEONATAL INTENSIVE CARfil
FOR VERY-LOW-BIRTHWEIGHTINFANTO
B. Sandhu1
R. W. I. Cooke2

R. C. Stevenson*!
P. 0. D. Pharoahs

D^anmaus ofEantomic and Businas SouSa,1 J
Child Health,2 and Community Health,3 UnwersityofLm
Summary

A derailed costing of the Mersey reg
neonatal intensive care unit was marl
1983 (at 1984 prices) for three levels of care; costs
inpatient day were £297, £138, and £71 for intensive, spe
and nursery care, respectively. Regression of ungrot
patient-specific costs against birthweight showed]
explanatory power of birthweight to be negligible.1!
average cost per very-iow-birthweight (<1500 g) infant
study elsewhere showed an almost*six-fold difference in:
between survivors and non-survivors. It is postulatedi!
medical management policy largely determines M
difference and is crucial to any investigation
efficiency.

- A. EL NAHAS AND G. COLES: REFERENCES—comimud
29. Ba±HA,Brann»FP.Diaxryprottmre«riaioninchromcren«ifiam2
e465.
30. GrraN,Kx>rbE,Stnod>M.L(roproamdteiapf>tementzdbykn»«cidi
renal failure a proapccrive controlled stndy. Kidney Im 19C3; 2«X
S263-67.
31. Giordano C. ProtongaDon of anrvival for a decade or more by tow prero
Giordano C, cd. Urenure pattebtoiogy of pattens treated for ren yvre
Proceedings ofthe third mama cotterenm. Mtem Widnif Ediwre,41
32. Mitch WE, Water M, Sreinman TL, re aL The efleas of a keto odbi
supptenent to a raoteed dte and progresrioa of chrome rate fifitam
Med 1984;311:623-29.
33. WaberM, Coulter AW, DigteS, cat The dfitaofteownatogms ofesea
arida maevtre chrome uraema. J Gnt/area 1973;52:678-901 3^
34. Laoaari.D, Klrinknecta Q Coanot-Wkmcr G, a.aL Bcntfictecia
pboepboraus dte in uraereic ren. CSaAa 1982:83:539-48. v'JJl
35. Kikuchi K.MatraahteT.HfaataK. Improveddtearytreatmeatwiffito^
pteepterouaratriatoo amtax: tea. Kidney im 1983;24(Mppil6pC
36. Kempf JD, Fncter HC, Kernel M. Efficacy ofan wmdeaed pram dtej
minor oral supply of eareotial ammo arida and keto anatogsKS compa
•decrive protein dte <40 g) in chrome renal foOma. Am J Obe Ntej
1673-77.
'■‘*1
37. Baraora G, Morelli E,MaaeaS,aaL Hypcrpteephreaamtea poeeihiecaaM
renal failure, he Giordano C, ed. Urazua: pattebiotogy of peocna tren
yean or more. Proccotepofthe third aremre crafervnre. Milan: Wkhd
1981:21-24.
38. Giordano GAaunfrarida and imn triifc nliaa^ie and poftte Amjg
loan- TV 1649-53
-dl
39. Broyer M, Guillot M, Nteate P, « iL Comparimn of three towJ
caxtemmg oeential amino arida and titer alpha anatogms for
duldren.Xiitery/iwl983;24(rappil6):S290-94.
40. Koppto JD, Roberta CE, Grotterm GP, a aL Lowproten dteaandthanre
uraemic patent. Lx: Avmtd MM. cd. Prevention ofkidMydteaoaand|
smvivaL New York, London: Pfenam Mated Book Co, lM2:3r2ti|
41. Koppto JD, Cotew JW. Evahmtioo of chrome urtm». hnp9Vtaace«^
ditrogen. seram creariniae and their rate JAMA 1974; Wt dWCifl
42. GrodetemG.KoppieJD.Ureamtrogenappeennca.ateBptoandpraaic^
total nnrogtn output. ZSteryte 1979; 18r 953 (aberr).
43. DegouteP, Reach LAimeF, cat Rhk&axCTmduumcterenodteyvteA
1980;17: 149-54.

i

r

|r

*5

44. Kleintaiechx G Sahtety I, Broytr M, a aL Effia ofvretom prorem diriMi
renal fimetem and rarvtvri of rnmaac no. Kidney im 1979; 1*»
45. ElNahaaAM,ParaakavatonH.ZoobS,aaLEfioaofJ m ij ; i mi HHq
development of renal faiiare after rabtote nephrectomy in fare. Cfcj
399-406.

Ntfkwi 1984; 21:54-59.
27. Cohen ML. RMiooudide dome* tedmiqtm.
NmMtJ 1974; 4: 23-38.
28. Berportek J, Ahrestrand A,BachtH, GnCKTrexA. ProptMMM ofchrome renal finhm
(CRF) is retarded by more frequem clinical follow up*. JCidmy ha 1985; 28: 283

Rtjmnta amrirtutd at foot of nect column

46. Btamenkrantx MJ, Kappie JD, Gutman RA, etaL Moteteofarei
natua of poteata with rate Eteure.Xw J CEn Nmr 1980; 33r
47. Cote GA. Body compociDon in chronic renal tetore. QamtJMedl9mm
48. LongDUHaverterg LM, Young VR,cr8LMaateitenof3Jtethy»H*S
Motetem 1975;24:929-34.
49. Whitehouve S, Kes N, Scteflcr G, Khnte R. Hbridwre red 1-^ •J
Neydni 1975:3: 24-27.
.

'■

601

^LANCET.MARGn !5,1986

I

INTRODUCTION

I, THE reducrion in mortality associated with improvements
k.0eonatai care is well known,1 but costs may still outweigh
gnomic benefits for infants weighing less than 1500 g at
Compared with larger infants, the costs for very-lowjirthweight (VLBW) infants are likely to be high and benefits
^uced by greater mortality and morbidity in the postnatal
rffiod. North American evidence supports this view. Walker
^4 colleagues demonstrated the low viability and high cost
^tensive care for infants ofbirthweight below 900 g,23 and
joyle et al reported that neonatal intensive care had more
tyourable effects among infants weighing 1000- 1499 g than
efrnings and costs were taken into account, the smaller
j^nts showed a net economic loss for any positive discount
It does not follow that it is not worth treating the smallest
j^fants, but Boyle et al concluded that if resources are scarce,
&^y might be better concentrated on infants in the higher
fcrthweight ranges.
*'In the United Kingdom, economic evaluation of neonatal
jrc is hampered by a shortage of costing studies. We report
jjere a costing of the regional neonatal intensive care unit
(NICU) at Liverpool Maternity Hospital and investigate the
gelation between birthweight and the cost of care.
METHODS

£ The total cost ofthe NICU in 1983 was estimated in terms of 1984
|ty and prices. Most costs were incurred at the NICU but services
iere provided by other centres including Liverpool Maternity
HospitaL other hospitals, and specialised units. Costings were as
‘gxcific as possible to neonatal care and to each cost centre. In
addition to neonatal care, the NICU carried out research, teaching,
nd outpatient clinics. Estimates ofthe time and equipment used for
these activities were subtracted from total costs to give the cost of
0re for all 542 infants treated. The costs were allocated between
three care levels defined as:
^Intensive care—infants. given respiratory support by either
mechanical ventilation or constant positive airway pressure.
I Special, care—infants given both electronic monitoring and
intravenous infusions for any reason; respiratory support not
required.
eNursery care—infants given special observation or care but fed
tally; respiratory support or intravenous infusions not needed.
The study included all infants treated, some of whom received
■either intensive nor special care.
i The NICU provided 7193 days of inpatient care. These were
^located to care levels’ from case notes. The cost of care for each
kvei was divided by inpatient days to give an average day cost.
Alultiplying the day costs by the number ofdays each of 182 VLBW
®hnts spent at each care level gave a patient-specific cost estimate
each infant.

Costing Methods
Medical and nursing staff.—Medical staif could be readily
•^ted to care levels. Nurses had principal responsibilities in
*P«ifjc rooms corresponding to the three care levels, but they
between rooms as required. To take account of this
®Ovement, the working of the unit was observed and several case
•sdies were made to estimate the nursing time spent at each level.
. fyutpment.—The estimate of the cost of capital equipment
^ded items donated by charities and took account of capital
a^niption and maintenance. Estimates were available of the
^cement cost in 1983 ofthe227 items ofequipment identified in
This estimate was converted into an annual charge for capital
gumption by straight-line depreciation on the assumption that
’verage lifetime was 10 years for mechanical and 7 years for
Tronic equipment. These were towards the upper end of the

range of plausible assumptions; another calculation was done for
lifetimes of 5 and 3 years.
Consumables.—Annual consumption was estimated from
quantities of 55 items used in one week in mid-1984. It was pnced
from delivery notes, invoices, and information provided by
manufacturers. On the advice ofthe nursing officer, it was allocated
in the proponions 2/1/1 between intensive, special, and nursery
care.
Drugs and pharmaceuticals.—The estimate ofannual consumption
was based on a 6-week sample period in mid-1984. 92 items were
costed and allocated to miscellaneous (allocated equally between the
three levels), parenteral nutrition, and antibiotics (both allocated
equally between intensive and spedai care).
Diagnostic tests.—Costing tests was more difficult than any other
but a request usually required more than one test. Numbers oftests
were estimated by multipiying requests by an average weighting for
each type. Tests were carried out in departments serving other
hospitals as well as the NICU. An existing study of the costs of
radiology to the NICU was updated. The average cost of an X-ray
(£18.04) included taxi fares and other extra costs for out-of-hours
tests. The cytogenetics department estimated the cost ofa test at £40
in 1983. The average costs ofhaematology and biochemistry (£1.16)
and bacteriology (£1.37) tests were calculated from an estimate of
the NIClTs share of each department’s workload. Milk tests were
not recorded in case notes and were priced separately. Tests were
allocated between intensive, spedai, and nursery care in the
proportions 6/4/1. This assumption, based on medical advice and
observation, was not used in the analysis ofcosts for VLBW infants
because patient-specific estimates were used.
Overheads.—Most overhead expenditure and a small amount of
income was shared between the NICU and Liverpool Maternity
HospitaL Seven item* (administration, records, training and
education, laundry, transport, crtche, and income) were allocated
by inpatient days. Only staffcatering costs (43% of the total) were
apportioned to the NICU, since food for newborn infants was
prepared in the unit. Four items (cleaning, engineering
maintenance, building maintenance, and estate management) were
allocated according to the-, cubic capacity of the NICU as a
proportion of the Liverpool Maternity HospitaL Specific estimates
were made for ponenng, heat-steam, water, electricity, and oxygen.
No allowance was made for the depreciation of buildings,
Overheads were allocated between care levels by inpatient days.

RESULTS

VLBW infants accounted for 34% of admissions but 65%
of inpatient days. Ofthe 182 VLBW infants, 131 (72%) were
bom at the Liverpool Maternity HospitaL The mothers of35
-of. rhrse had hooked
rhe hnsniral, .96 transferred rhrrr
during pregnancy, and 51 were postnatal transfers.
VLBW infant mortality was 25%. It ranged from 100% for
birthweight <700 g to 10% in the 1400-1499 g range. The
mortality rate was 22% for inborn infants and. 33% for
outborn. Of the 182 VLBW infants,’ disabilities were
identified in 16 (8*8%).
Table I shows the cost ofthe whole unit for 1983 in terms of
1984 pay and prices by prindpai cost categories. The most
expensive items were nursing staffs tests, and overheads. If
average lifetimes of equipment had been taken as 5 and 3
years for mechanical and electronic items (instead of 10 and 7
years) equipment costs would have been £91 069, total costs
£1 057 903, and the proportion of equipment costs in the
total 8 *6%.
Table I also shows total costs allocated to the three r^rr
levels and divided by inpatient days to give average day costs.
Day costs are sensitive to the capacity level achieved in the
unit. In some months during 1983 the NICU treated more
than twice its notional capacity. If it had cared for fewer

.a.5-_

-J
602

THE LANCET, MARCH 15,19^



TABLE 1—TOTAL COST* OF CARE FOR 542 INFANTS ALLOCATED
BETWEEN CARE LEVELS

(£) ,

Special
care
(£)

41 060
8- 6
201 239
35-6
4-7 ; 24 482
48 980
9- 7
8-3 : 33 797
73 446
17-2
34 398
15-9
457 402
1541
£296.82

41 060
82 616
13 375
24 489
33 797
74 447
53 625
323 409
2343
£138.03

■ Intensive
care

Total
(£)
Medical staff
Nursing staff
Equipment
Consumables
Drugs
Tests
Overheads
Total
Inpatient days
Cost per day

87 691
361 129
47 873
97 958
84 689
174 179
161 188
1 014 707
7193
£141.07

%

Nursery
care
(£)


:
j
i

5571
77 274
10 016
24 489
17 095
26 286
73 165
233 896
3309
£70.68

transfers, the number ofweeks of gestation, and the square of
birthweight (a test for non-linearity). Over many equation^
most of the signs of the independent variables were
predicted and their coefficients were usually significam
(p<0’05) but their explanatory power was negiig^
(R2<0-1). No improvement was obtained by restricting the
sample to infants who had at least 1 day in intensive care ncr
was there evidence of a structural break in the data at either '
900 g or 1000 g.
DISCUSSION

Clinicians and economists concerned with neonatal cart
have analysed outcomes in birthweight ranges. Boyie et al*
implied that birthweight ranges might guide the planning of
’In terms of 1984 pay and prices.
medical resource allocation. When full data are available, this
procedure can be shown to be statistically inefficient and
possibly misleading. The division of the VLBW range at
infants, day costs would have been higher, since fixed costs
1000 g suggests that an infant weighing 999 g is different
would have been spread over fewer days.
from one weighing 1000 g. Of course, this would not be true
Intensity of effort by staff was an important missing
everrifbnrhweighr were measured'as accurately as this. The.
vatiabk- Its mfiueme mighrbe rbtmd m stafFrumover and
selection of birthweight ranges is arbitrary’, and grouping
quit rates and in other less readily quantifiable forms. A true
involves the use ofaverages taken from cost data which havei
cost would include an adjustment for achieved capacity
high variance and are bimodal for the smallest imams.
levels.
Our data, grouped in 100 g ranges, showed a correlation
The mean (±SD) cost of care for all surviving VLBW
coefficient of0»6 between birthweight and cost, which leavei .
infants was £3615±3014 and that for non-survivors was.
40% of the variance unexplained. Whether this correlation
£3446±6143. Outbom infants (survivors £4664±3685, non­
would be an adequate guide to resource allocation in marten ;
survivors £4656±7667) were more expensive than inborn
of life and death is doubtful. However, the result was shown :
(survivors £3265±2685, non-survivors £2736±5064). These
to depend on the choice of birthweight ranges and was not J
data had very high variance,, especially in the lowest
confirmed by the ungrouped data. The severity’of infantf1 j
birthweight ranges. Some ofthe smallest infants survived for
symptoms was probably the principal determinant ofcost but ;
only a short time and were among the cheapest to treat.
it was not captured by restricting the sample to those who |
Others survived for long periods in intensive care and were
received mechanical ventilation.
.J
the most expensive.
These findings fall short ofa full evaluation to be compared. |
Table II shows the cost of producing a survivor in 100 g
with that of Boyle et at4 In particular, the costs
birthweight ranges. If the objective of neonatal care is to save
restricted to those incurred in the NICU. No attempt ww^
lives, without reference to their expected length or quality,
made to quantify either the benefits or the future
this figure is a rough measure of the cost-effectiveness of
attributable to intensive care, though work is proceeding in M
intensive care for different birthweight ranges.
this direction. It may be that as future costs become known,
The best statistical relation between birthweight and cost
the cost-benefit calculation will swing against the smaUe«rj
for VLB5^ infants was obtained from ordinary least squares
infants, but this is by no means certain. For all but the motf >
regression of birthweight grouped in 100 g class intervals
severely incapacitated, neonatal costs are likely to be the mojH
against the cost of producing a survivor (R2=0’6) but it was
expensive episode in the medical history, and these costs (Mh ■
shown to depend on the selection ofclass intervals, which was
not seem to be related systematically to birthweight.
'-*5
entirely arbitrary. This variation in the explanatory power of
The problem of selection bias in a regional referral
,
binhweight was due partly to the tendency to round weights
well known. It has been claimed that regional units atn»
to the nearest 10 g.
from district hospitals the infants with the best prog11056^ * j
When ungrouped, patient-specific cost estimates for all
that improvement in outcomes in referral units may be
VLBW infants were used, the relation between birthweight
by deterioration elsewhere.5 It is equally likely
*
and cost disappeared (/?2=0‘04). Improvements in the
intensive-care rrrhmque* become more widely dissetnin**’-*’
statistical explanation of cost were sought by including
local hospitals will retain infants with good Prognosct^ •
dummy variables for survivors, inborn infants, and in-utero
refer the difficult cases. Referral rates from district
. differ widely livMertey Region. They are not
.
TABLE II-TOTAL COST PF;PRODUCINGA SURVIVOR'
distances from the NICU and seem to be rc^atCost of a
No of
Cost of care
unexplained variations in medical practice. In this
survivor’ (£)
survivors
Birthweight (g) |I_ n_
(£)
outbom cases were more expensive than inborn and
0
i 4
27 906
500-599
mortality rates were higher. The selection bias seem
0
22 878
600-699
work against the NICU.
.
13413
5
67 066
700-799
II
The only similar UK study was of infants treated m
6324
11
69 561
20
800-899
5065
regional NICU of Birmingham Maternity Hospit*1^^
65 847
13
18
900-999
6702
16
107 226
1000-1099
21
1980-81.5 Comparisons must be very tentative, s*ncCg^<
4356
22
1100-1199
i| 26
95 836
studies are separated by time and regional
2513
25
62 831
1200-1299
I 28
differ somewhat in their methods and costing ProccaiDL^
2826
25
1300-1399
I! 28
70 646
3176
19
The day costs for care levels (defined similarly but •n*
60 336
1400-1499
; 21
identically) in 1984 pay and prices are given in table I
•Totai cost of care divided by number of survivors.

...t. yv<rt

CLINICAL ECONOMICS
MODULE 3
ECONOMIC COSTS AND DISCOUNTING

module you should:
At the completion of this
understand the distinction between economic and
a.
financial costs;

b.
c.

understand the reasons for discounting, costs which are

expected to accrue in the future,
calculate the present value cost of an
be able to
illness for a given discount rate.

ECONOMIC COSTS AND DISCOUNTING

Introduction

i

concept of discounting was
In Module 2 it was shown that the an illness.
Discounting is
central
to estimates of the cost offirst the meaning of the term
but
this module,
the focus of
be
clarified.
must
’’economic cost”
individual, a hospital, or a government is
to an
The cost
for example, the fee an
to pay­ for something
the money used
the
money the government
visit a doctor, or
to
pays
individual
These costs are
for
the
poor.
health care
pays to subsidise
actual
money
transactions.
They measure
called financial costs .

i sj the cost to society as a whole rathejr
economic cost
cost to
to one
one section
of_societyn This corresponds to
s
than the
t^Toncepror opportunity cost that was described in Module
.

An

suppose a
a decision is made to build
your memory , suppose
To refresh
a hospital.
The resources consumed (building materials, labour,
to build rural health clinics or
linen etc.)
cannot be used
etc. )
ThA_op
portunity cost is the value
sports centres,
for example,
centres,
best
alternative
use for resources that
the
society places
on
decision
to
build the hospital is
The
decision
to
must be foregone,
value society places on the services produced
efficient if
the
by the hospital exceeds the opportunity cost.
Transfer Payments
often differ from financial costs.
Consider
Economic costs
purchased by a local doctor at a cost of
an X-ray machine
Assume that this price included a government^tax of $450
$4950.
e of $4500.
The private' financial cost of'
on the wholesale pr.
to
the doctor was $49 5.0, 'but this was not the cost
the machine
The tax payment did not use up any resources which
to society.
used elsewhere - it involved no opportunity
then cou Id not be
It simply transferred
the command
over
cost to soci ety.
person
in
society,
ihe
doctor,
to
the
from one
resources
represents
society
as
a
whole.
Accordingly,
which
government
taxes are examples of transfer payments which are financial but
The true cost of the machine is the value
not economic costs.
used in its construction which cannot be used
of the resources
security
payments are also
transfer
Soci al
elsewhere.
they
transfer
the
command
over
resources
In this case
payments.
from society as a whole (the government) to the recipient.

U

(st

2

Economic Costs and Cost of Illness Studies
Cost of
i1Iness
studies seek to measure the costs which a
di sease or condi tion
impose on s o c i e t y-- i^a.t jier__ th a n—on a n
individual.
As__ such ,
ecjonojnijc __costs shou 1 d be estimated jn
preference
to
f inane ial
costs.
Consider a direct cost such as
the cost
of hospitalisation.
The resources used to provide
hospi tai
be used elsewhere, so it is the value
services cannot
of the
foregone opportun i t y which should be
included
in a
costing
study.
The charge to a patient for hospitalisation wj^ll
rarely be a good
indicator of
the economic cost, especiaXJLy
where governments subsidise hospital care .
The
indirect costs of an illness are similar in that they
represent
the value of potential production lost because people
cannot work,
work , or cannot work efficiently.
They are opportunities
lost because of the illness.

It.
is
sometimes difficult
to estimate economic costs
accurately.
Th i s
issue is considered again in a later module,
but at
this
s tage
is
it
important for you to recognize that
economic costs are not necessaril.y the same as financial costs.
DISCOUNTING

Reasons for Discounting
Most
people would prefer to receive $1000 today than $1000
in a year’s time.
time .
This is largely because the person loses the
opportunity to earn
income
from
the money,
for example by
i n t e r e s t,
earning
interest,
iiff payment is delayed for a year.
year,
Likewise,
pref er
people prefer
to pay a
bill of $1000 in a year’s time rather
today.
than today.
Again,
Aga i n, the reason
is that the person loses the
opportunity
to earn
income from the $1000 over the forthcoming
year by paying the bill today.

In both cases,
cases,
th .
implication is that people value $1000
today more highly than
$1000 a year
later.
The general
principle
is
is
that money now is valued more..highly than an equal
sum that will
be paid
or
received
in 'the- future.
This
preference
for money now over money in the future is called the
. rate of time preference.
Assume
that the best interest rate you can obtain is 10% per
year.
If
I
offered you the choice between $1050 payable in a
year and
$1000 now, you probably would take the money now.
If
you
invested
the money
now it would be worth $1100 in a year.
However,
if
I offered you the choice between $1200 payable in a
year and
$1000
now,
most
of you would choose to wait for the
most.
year as
Long as you were reasonably confident of my honesty,
There fore,
there must be a sum of money payable in a year,
somewhere between
$1050 and $1200, which you consider to be of
equal
value
to having $1000 today.
(For some of you it may be
greater than $1200.)

3

$1150
($100
( $100
to
lost
compensate
for the
is
For me it
to compensate
for the uncertainty).
I am
interest,
and $50
a payment of $1150
in a year and $1000
indi f f erent between
long as
the interest rate is 10%.
In other
payable
today as
I have d i scounted
woxds,
$1150
in a year is worth $1000 today.
the Xu-ture—sum of money to find its value to me today, which is
I discounted $1150 by 15% to obtain
called its present value.
the present value of $1000, so 15% is called the discount rate.

As long as the discount rate is known, it is possible to
find the present value of any sum of money that is expected to
If you think
be paid or received at anjr time in the future,
you
will
discover
through the
logic of the earlier examples,
sum
that the further into the future a
of money is expected, the
Similarly,
lower will
be
its present value.
the higher is the
discount rate, the lower will be the present value.
Discounting and the Cost of Illness
You will remember from Module 2 that in calculating the
cost of an illness it is necessary to estimate the cost it will
Consider a disease which is
impose on society
in
the future,
estimated to cost $1 million a year for 5 years.
Viewed from
perspective, the total cost of the disease to society is
today’s
not $5 million because $lm this year is worth more than $lm next
which
in turn is worth more than $lm in 2 years time etc.
year ,
year’s cost to its present
It
is necessary to convert each
values
can then be summed to
The
stream
of
present
value.
p
resent
of
the
illness.
The following
the
present
value
cost
obtain
cost
of
our
example
is $3.79m
shows
that
the
present
value
table
using a 10% discount rate.

YEAR

COST ($)

1
2
3
4
5
TOTAL

1m
1m
1m
1m
1m
5m

PRESENT VALUE

1 ,,
’I ■>

'■ '

909,000

826,000
751,000
683,000
621,000
3 , 790,000

($)

<
.

'7
J 4(2 ' A -

------- 7.

74 ■ '■ ■

’j3 JI JL 4 *
22
3

r 1^’9

i 0 0 o. pe

(I


i!- :

» i

4^< ^ ’

this process are described in_Evans et al
The mechanics of
(1980),
so will not be repeated here.
It
( 1984 )
and Drummond
read one of these references in order to ’Xl—
will
be necessary to
complete question 2.

The Discount Rate
illness estimates are calculated as present values.
Cos t of
value
costs are
sensitive to
the choice of discount
Present
and the appropriate rate depends on a number of factors?
rate ,
There
is disagreement in the literature about the correct rate,
and
appendix 4
in Drummond
(1980)
discusses
some of the

4

problems.
The arguments cannot be summarised easily, and they
to
your under-standing of why
are not particularly important
At this point you should be able to
discounting
is necessary.
present value cost of an illness if you are given
calculate
the
Question 2 asks you to consider some of
rate.
the appropriate
the factors influencing the choice of the rate.

REFERENCES

Major References:

Barnum,
H.
"Evaluating healthy days of life gained from health
projects”, Social Science and Medicine, 1987, 24 ( 10):833-841 .
Evans et al,
A guide
South Pacific, SPC, Noumea,

to project planning and appraisal in the
1984, pp72-74.

Other References:

Drummond M.F.
Pr i noi pies of economic appraisal in Health Care,
Oxford Uni Press, Oxford, 1980, pp 47-51 & appendix 4
Hodgson T.A.
"The state of the art of cost-of-i1Iness estimates”,
Advances in Health Economics and Health Services Research 1983,
4 : 129-164 .

QUESTIONS

1.
Read
the article by Barnum.
Are you convinced that it is
necessary
to adjust healthy days of life gained by an intervention
for time preference and for productivity?
Discuss.

2.
THIS QUESTION IS BASED ON QUESTION 3, MODULE 2.
a.
ITnn Q3,
parts b and d you calculated the present and future
costs of
treating
the
di seases.
four diseases.
Now calculate the present
value cost of
treating each disease using a 5% discount rate.
(Define the year of onset as year 1 ) .

b.
Repeat
part. a using
differences you observe.

a

10%

di scount

rate .

Explain any

part f, you calculated lost earnings due to premature
c.
In QB,
mortali ty.
Explain
how you would calculate the present value of
these
lost e a r n i n g s .
(Do not do the calculations unless you have a
lot of spare time. )
d.
Using
your results in QB, part h, calculate the present value
of
losses
from d i s a b i 1 i t y days
for each disease,
using a 5%
discount rate.

5

e.
Had you completed the calculations for part c above, you would
have found
that the present values of lost production due to
premature mortality,
with a 5% discount rate, were A = $789_,_92.8 ; B
= 627,285;
C - 2,202,307; and D = 1 ,265,912. - Calculate the total
present value cost of each disease at a 5% discount rate.

\

3.
has

A company borrows money to build a private nursing home.
It
to repay the loan and interest over 10 years.
Is the repayment
of a .
the loan;
—•
f'
;
b.
the interest,'
an economic cost?* ’
4.
Everyone has the same rate of time preference, so it is easy to
determine society’s rate .
Discuss.

5.
a. What factors would influence the choice of discount rate in
a cost of illness study?
b.
;•••’■
A high discount rate discriminates against preventive
medicine in favour
medicine.
— of
„curative
------ - ------Discuss .

Copieri for use by students enrolled
in Clinical Economics,
Lecturer: Dr. D.B. Evans.
University of Newcastle, 1987.

A GUIDE TO PROJECT PLANNING AND APPRAISAL
IN THE SOUTH PACIFIC

(Revised Edition)

Volume I

D.B. Evans, F.V. Sevele, and A. McDonald

South Pacific Commission
Noumea, New Caledonia
May 1984
1 7/»-

6

David Evans et al (1984)
Time Value of Money

Most people if offered the choice between a certain amount of
money today and the same amount in several years‘ time would choose
to have the money today. This preference for money now as distinct
from money later has several explanations, but most of these are
connected with a loss of opportunity, commonly called opportunity cost.
In other words money obtained now can be put to a productive purpose
which will generate income. If the receipt of money is delayed then
’ so 'is the opportunity to generate income. Similarly, money obtained
now may also be used for ’non-productive' (in terms of income)
activities such as leisure pursuits and current consumption. Although
these may also be enjoyed in the future, the uncertainty of the future
may mean that people prefer to enjoy them now rather than at a more
uncertain later date.
Because money received in the future as distinct from the
present represents a loss of opportunity, anyone lending money, and
thereby foregoing opportunities, will need to be compensated when
the money is repaid in the future. Thus, if a farmer lends money to
his neighbour he is foregoing the opportunity to use that money to
generate income by increasing, say, his fertilizer applications. On
the other hand, his neighbour is gaining the use of that money to
put to a productive purpose, say to irrigate his land. Obviously the
lender would expect to be compensated for the income he is foregoing.
This compensation generally takes the form of interest where the
borrower is in fact paying the lender for the use of his money, The
interest rate reflects peoples' preference for money now as distinct
from money in the future; that is, it is the 'time rate of preference'.
The higher the farmer values the opportunity of income foregone the
higher the rate of interest that he will charge the borrower.

Compounding
an amount A is loaned to a person at an interest rate of
If an
r per annum,, then after 1 year the amount to be repaid is:

A! + A^r

A1 (1 + r)

7

In this as well as in the remainder of the chapter, r is
expressed in decimal form, e.g. for an interest rate of 10%, r = 0.10.
After two years, the amount to be repaid would be:

A1 (1 + r) + A 1 (1 + r)r = A,1 (1 + r) (1 + r)
2
= A1 (1 + r)

Therefore, if the loan is for n years, the amount to be repaid
in year n is A^ fl
(1 + r)

The amount (1 + r) n is the factor by which a sum of money will
increase in value when earning compound interest at a rate of r%
per annum for n years; this is referred to as the compounding factor.
For example, an amount of $100 invested for 3 years at 10% per annum
will be worth after 3 years:
Value

A1 (1 + r)n
100 (1 + 0.10)3
100 (1.1)(1.1)(1.1)
110 (1.1)(1.1)
121 (1.1)
133.10

Through the process of compounding, the $100 grows to $110 at the
end of year 1, $121 at the end of year 2, and to $133.10 at the end
of year 3.

Discounting and Present Value
The time rate of preference can be considered from the opposite
point of view. Rather than ask how much a particular amount of money
would be worth at sometime in the future, the present value of an
amount of money expected in the future can be calculated. A ’penalty'
must be imposed on this money to compensate for the fact that it is
to be received in the future rather than now.
The process of finding the present value of a future amount is
called discounting and the discount rate at this stage will be assumed
to be the interest rate since they are both concerned with the same
time rate of preference, but applied from different ends of the time
scale. This assumption is discussed further in Chapter 5. The
interest rate involves looking from the present to the future, while
the discount rate looks backward from the future to the present.

It has' been shown that for an interest rate of r%:

A

n

A1 (1 + r)n ; therefore A.
1

A

n

( 1 + r)n
and

1

is the factor by which a sum to be received n years in
(1 + r) n
the future must be discounted to determine its present
value; it is referred to as the discount factor. Thus, if a person
is promised $133.10 in 3 years’ time, and the rate of interest is 10%,
the present value of this amount is:

8

An

A1

(1 + r)n

133.10

$100.

(1 + 0.1)

3

As expected, this is consistent with the earlier compounding example.

The concept of discounting may be applied to each year of a
v.
-intst as well as to an amount applying for only one
yelr ind the present value of a future income stream can be determined.
The World Bank has produced a set of Compounding and Discounting
Sbles for project evaluation which gives the different discount
factors for different combinations of n and r, and these or similar
associated with
tables can be used for the various calculations
<
,
III)
contains
tables giving
project appraisal. Volume II (Part
selected
combinations of
compound, discount and annuity factors .for
n and r.
the tables
For example for a discount rate of 15% per annum,
give the following discount factors for a five-year period:
Discount Factor

Year 1

Year 2

Year 3

Year 4

Year 5

.870

.756

.658

.572

.497

Note that these discount factors can be calculated using the formula
■--- --outlined earlier. Note also that costs and returns
can1 accrue at any
different
discount rate
time during a given year which implies that a
This
obviously
is
could apply for each day of a project s 11 eis
assumed
that
all
impractical and for convenience it generally
This is the
transactions are made oni 31 December of each year, • of the project
reason that the costs and
*— benefits of the first year
are also discounted.
The following shows how to calculate the present value of a
five-year income stream discounted at 15%.

Discount
Income to be
factor
received
(2)
(1)_____

Year 1
Year 2
Year 3
Year 4
Year 5

$400
500
400
600
500

Total

$2,400

.870
.756
.658
.572
.497

Present
Value
(3) = (1) x (2)

$348
378
263
343
249

$1,581

I


I

VO:

-1

- X
vo3Vt2-

■— z

’5

! ..^4 0
Z;l.

Yen

1%

i?r

57,

69f

87,

l(K

123

143

15%

16%

18%



X i r..'

I -1

x.i )

t

I

1
■’

20%

22%

24%

25%

26%

28%

30%

35%

40%

45%

5O',i

■H

I
2"
3
4
5

0 990 __0,971
0.980
0.943
0.971
0.915
0.961
0 888
0.951
0.863

6
7
8
9
10

U.942
0.933
0.923
0.914
0.905

11
12
13
14
15

IL952
0.907
O.8b4
(•.823
0.784'

U.943
0.890
0.84(1
0.792
0.747

0.926
0.857
0.794
0.735
0.681

0.909
11.826
U.751
ll.oKS
0.6 21

U.S93
(•.797
0 712
U.636
0.56 7

0.877
0.7o>
0.675
0.592
0.519

0.870
0.756
0.658
0.572
0.497

(1.862
11.743
IM.41
0.552
0.4 z6

0.847
0.718
0.609
0.516
0.437

0.833
0.694
0.579
; 0.482
• 0.402

0.820
0.672
0.551
0.451
0.370

0.806
0.650
0.524
0.423
0.341

0.800
0.640
0.512
0.410
0.328

0.794
0.630 *
0.500
0.397
0.315

0 781
0.610
0.477
0.373
0.291

0.769
0.592
0 455
0.350
0.269

0.741
0.549
0.406
0.301 1
0.223

0.714
0.510
0.364
0.260
0.186

0.690
0.476
0.328
0.226
0.156

0 667
0.444
0.296
0.198
0.132

0.789
0.766
0.744

0.746!
(I 711:
0.677
0.6'.5i
0.614

0.705
0.665
0.627
(1.592
0.558

0.6 30
0.583
0.540
0.500
0.463

0.564
U.513
0.467
0.4 24
0.386

U.5U‘
0.452
0.404
U.36I
0.322

0.4 f 6
0.400
11.351
0.308
0.270

0.432
0.376
0.327
U.2K4
0.247

0.410
0.354
U.305
0.263
0.227

U.37O
0.314
0.266
0.225
0.191


;
i


0.335
0.279
0.233
0.194
0.162

0.303
0.249
0.204
0.167
0.137

U.275
0.222
0.179
0.144
0.116

0.262
0.210
(1.168
0.134
0.107

0.250
0.198
0.157
0.125
0.099

0.227
0.17b
0.139
0.108
0.085

0.207
0.159
0.123
0.094
0.073

0.165
0.122
U.091
0.067
0.050

0.133
0.095
0.068
0.048
0.035

0.108
0.074
0.051
0.035
0.024

0.088
0.059
0.(139
0.026
0.017

0.896
0.887
0.879
0.870
0.861

0.722
0.701
0.681
0.661
0.642

O.5L5<
0.557
0.530/
0.5(15
0.4 81

0.527
11.497
0.469
0.44 2
0.417

0.429
0 397
0.368
0.340
0.315

0.350
0.319
0.290
0.263
0.239

0.287
0.257
0 229
0.205
0.183

0.237
0.208
0.182
t0.160
*0.140

0.215
0.187
0.163
0.141
0.123

0.195
0.168
0.145
0.125
0.1 U8

0.162
0.137
0.116
0.099
0.084

0.135
0.112
0.093
0.078
0.065

0.112
0.092
0.075
0.062
0.051

0.094
0.076
0.061
0.049
0.040

0.086
0.069
0.055
0.044
0.035

0.079
0.062
0.050
0.039
0.031

0.066
0.052
0.040
0.032
0.(125

0 056
0.043
0.033
0.025
0.020

0.037
0.027
0.020
0.015
0.011

0.025
0.018
0.013
0.009
0.006

0.017
0.012
0.008
0.006
0.004

0.012
0.008
0.005
0.003
0.002

16
17
18
19
20

0.853
0.844
0.836
0.828
0.820

0.623
0.605
0.58?
0.570
0.554

0.458
U.4 36
0.416
0.396
0.377

0.394
0.371
0.350
0.331
0.312

0.292
0.270
0.250
0.232
0.215

0.218
0.198
0.180
0.164
0.149

0.163
0 146
0.130
0.116
0.104

0.123
0.108
0.095
0.083
0.073

0.107
0.093
0.081
0.070
0.061

0.093
0.080
0.069
0.060
0.051

0.071
0.060
0.051
0.043
0.037

0.054
0.045
0.038
0.031
0.026

0.04 2
0.034
0.028
0.023
0.019

0.032
0.026
0.021
0.017
0.014

0.028
0.023
0.018
0.014
0.012

0.025
0.020
0.016
0.012
0.010

0.019
0.015
0.012
0.009
0.007

0.015
0.012
0.009
0.007
0.005

0.008
0.006
0.005
0.003
0.002

0.005
0.003
0.002
0.002
0.001

0.003
0.002
0.001
0.001
0.001

0.002
0.001
0.001
0.000
0.000

21
22
23
24
25

0 811
0.803
0.795
0.788
0.780

0.538
0.522
0.507
0.492
0.478

0.359
0.342
0.326
0.310
(1.295

0.294
0.278
0.262
0.247
0.233

0.199
0.184
0.170
0.158
0.146

0.135
0.123
0.112
0.102
0.092

0 093
0.083
0.074
0066
0.059

0.064
0.056
0.049
0.043
0.038

0.053
0.046
U.040
0.035
0.030

0.044
0.038
0.033
0.028
0.024

0.031
0.026
0.019
0.016

0.022
0.018
0.015
0.013
0.010

0.015
0.013
0.010
0.008
0.007

0.011
0.009
0.007
0.006
0.005

0.009
0.007
0.006
0.005
0.004

0.008
0.006
0.005
0.004
0.003

0.006
0.004
0.003
0.003
0.002

0.004
0.003
0.002
0.002
0.001

0.002
0.001
0.001
0.001
0.001

0.001
0.001
0.000
0.000
0.000

0.000
0.000
0.000
0.000
0.000

0.000
0.000
0.000
0.000
0.000

26
27
28
29
30

0 772
0.764
0.757
0.749
0.742

0.464
0.450
0.437
0.424

0.412

U.28I
0.268
0.255
0.243
0.231

0.220
0.207
0.196
0.185
0.174

0.135
0.125
0.116
0.107
0.099

0.084
0.076
0.069
0.063
0.057

0.053
0.047
0 04 2
0.037
0.033

0.033
0 029
0.026
0.022
0.020

0.026
0.023
0.020
(MJ 17
0.015

0.021
0.018
0.016
0.014
0.012

0.014
0.011
0.010
0.008
0.007

0.009
0.007
0.006
0.005
0.004

0.006
0.005
0.004
0.003
0.003

0.004
0.003
0.002
0.002
0.002

0.003
0.002
0.002
0.002
0.001

0.002
0.002
0.002
0.001
0.001

0.002
0.001
0.001
0.001
0.001

0.001
0.001
0.001
0.000

o.uoo

0.000
0.000
0.000
0.000
0.000

0.000
0.000
0.000
0.000
0.000

0.000
0.000
0.000
0.000
0.000

0.000
0.000
0.000
0.000
0.000

35

0.706

0.355

0.181

II. 130

0.068

0.036

0.019

0.010

0.008

0.006

0.003

0.002

0.001

0.001

0.000

0.000

0.000

0.000

0.000

0.000

0.000

0.000

40

0.672

0.307

0 142

0.097

0 046

0.022

0.011

0.005

0.004

0.003

0.001

0.001

0.000

0.000

0.000.

0.000

0.000

0.0(10

0.000

0.000

0.000

0.000

45

0.639

0.264

0.111

0.073

0 031

0.014

0.006

0.003

0.002

0.001

0.001

0.000

0.000

0.000

0.000

0.000

o.uoo

0.000

0.000

0.000

0.000

o.uoo

50

U 608

U.228

II.US’’

l».05l

U 1121

0 009

ll.llii?.

n.uoi

(1.001

0.001

0.000

0 000

0.000

0.000

0.000

0 000

o.uoo

0.000

0.000

0.000

0.000

0. ill II)

I
0.837

6.813

TABLE 3.2

0.022

DISCOUNT FACTORS - VALUE of (1+r) n

I

I

<£>

CLIMICAL ECONOMICS

MODULE 4

COST-MINIMIZATION ANALVSIS

A f ter completing
thi s
shou 1 d un der st and s

module and the recommended - read.i nqs, you

a..

the approach used in a cost-mini mi cation analysis;

b.

the strengths and limitations of the approach.

1
COST-MIMIMIZATION

hNhL¥SIS

NOTEs
In modules 1-3 it was necessary to introduce some
basic economic concepts in addition to Cost of Illness
and Cost Analysis.
Now that you understand some of the
principles underpinning Clinical Economics, the
remaining modules will be substantially shorter.
you
should devote more attention to the recommended r eadings
and the associated questions.

INTRODUCTION
economics is concerned with the economic efficiency
Cli ni cal
where efficiency depends on both the
i nter vent i on s,
of
health
of
a
programmeThe relationship between
outcomes
inputs and
inputs and outcomes i s represented in the following diagrams

RESOURCES
CONSUMED

■>

HEALTH
CARE
PROGRAMME

HEALTH IMPROVEMENT

■>

Cost (C)
Cl = Direct costs
C2=Indirect costs
(Production losses)
C3=Intangible costs

(adapted •from Drummond et al ,

P-

2)

you should
be aware how the uost of a
shOLll d
previ ous modules,
be
estimated.
The
benefits
are more difficult to
programme can
,
the
technique
introduced
in Nodule 3,
Cost-analysi s,
measure,
the
benefits
generated
by
alternative
programmes
i s used wh er e t.. _ __
It does not require the
can be assumedI to be identical.
to
be measured.
The techniques which
will
be
outcomes
in this and subsequent modules do require outcomes to
i ntroduced
be measured.

From

COST-MINI lvl I Z AT I ON ANflLYS IS

Cost-minimization
analysis is
i s used to compare alternative
the same disease or condition where it can be
interventions for
outcomes.
The
proved t hat
t h e alternatives have identical
technique requires a test of equivalence of outcomes rather than
that outcomes are equivalent which is central to
the assumpti on
cost analysis..

To test for equivalence it is necessary to define an outcome
i nd i cat or ..
1 s not possible uD recommend a general outcome
It
is
i nd i cater
because
outcome
of “ an
depends on the type of
i ntervention
and medical conditi on.
Ac cordingly„ a large number
o f i n d i c a t o r s are found in t h e 1 i t a r a t u r e.

For

ex amp1e,
P i a c h a u d a n d W e d d e 11
tested whether surgery
1)
and
i n j ec t i on -c omp r es s i on
sc 1 er o t h er ap y (method 2)
were equal 1y e f f e c t i v e t r e a
t m en
e n t s for
f or
at
v a r ic a s e veins.
They
r an d cm1y al 1ocated
p at i en t s t c -i h e t w o t e c hi n i q u e s and followed
t h e s Lt b.j e c t s f o r three years.
(method

Th e i r
outcome indicator was the proportion of patients who
had required n o f u r t h e r t r e a t m e n t
in
t h e 3 years.
These
pati enta were regarded as having been treated
successful1y
because unsuccessful treatment s wQuid have r equi r ed some form of
r cd low up action.
Of the patients treated with method 1, 86Z
required no further action compared to 787 of those treated with
method 2u

However,
a larger number of patients failed to attend for
surgery
than for
sclerotherapy.
The authors argued that
pati ents who avoided treatment should be added to the group that
were treated
unsuccessfully.
With this adjustment, 757. and 737.
of
patients allocated initially to methods 1 and 2 respectively,
did
not require further treatment.
This difference was not
stat i st i cal 1y si gni fleant.
□nee the outcomes of alternati ves are shown to be identical,
cost-mi nimi zation
emp1oys the same technique as cost analysis.
It
c h o o s e s t h e cheapest alternative.
Remember that this may
depend on whose v i ewp oi n t is considered.
Piachaud and Weddell
showed that
sclerotherapy was cheaper
than
surgery from the
v igwp o i nt of the pat i en t ■ the Health Service and the community.
j

REFERENCES
Drummond M.F.,
G.L. Stoddart and 8.W.Torrance,
Methods for the
ec on cm i c
evaluati on p f
he a 1 tn care proqrammes 3
□UP, Oxford,
1987, chapter 4.

Pi achand
vei ns” ,
287-294.

Weddel 1,
Ij. & J.M.. Weddell,
"The economics of treating varicose
Internati onal
Journal
of
Epidemiology.
1972,
1 (3) 2

Pi neault R.
et
al ,
su rg er y", Medical Care,

"Randomized
clinical
trial
1985, 23 s 171 -182. (attached )

0+ one day

3

QUESTIONS

QUESTION 1 NILL BE ASSESSED
1u

hMD

MUST E<E SUBMITTED

Read the article by PineaLilt et al.
What outcome indicator is used?
Do you think the authors have proved that the outcomes
b.
of the 2 alternatives are identical?
If not, why not?
Describe how your response to this paper might vary
au
t he
depending on whether you took the viewpoint of <i)
patient.
(ii)
the surgeon,
(ill)
the hospital,
( i v)
the cornmun i ty of pati ents requi r i ng surgery.
•3 u

Cl u

b..

3.

Do you agree with F’iachaud & Weddell that the outcomes
of surgery and injection-compression sclerotherapy are
identical?
If not, why not?
Define a single outcome indicator which could have been
used in both studies - F'ineault et al and Flachand and
Weddel 1.

What are some of the strengths and limitations of cost­
minimization analysis?

4

Copied on behaiS^f'

Medical Care
February 1985, Vol. 23, No. 2

j

The University of Nev/castle

/□r. Medicine Ref. No.

DateC^-^ST- j

Randomized Clinical Trial of One-day Surgery
Patient Satisfaction, Clinical Outcomes, and Costs
Raynald Pineault, MD, PhD* Andre-Pierre Contandriopoulos, PhDI
Marie Valois, iMSct Marie-Lynn Bastian, BAJ
and Jean-Marie Lance, MSc§

One hundred and eighty-two patients undergoing tubal ligation, hernia re­
pair, or meniscectomy were randomly assigned to either one-day or inpatient
surgery. The study’s objective is to compare these two modes of care with
regard to patient satisfaction, clinical outcomes, and costs of the episode of care.
A significantly higher proportion of one-day patients than their hospitalized
counterparts found their stay to be too short and would prefer hospitalization as
an alternative. Clinical outcomes were comparable in both groups. One-day
tubal ligation and hernia repair were found to be cost-efficient and averaged
hospital savings of $86.00 and $115.00 more than inpatient care. Meniscectomy
deviated from this trend in that treatment costs were significantly higher for
one-day surgery patients. Analysis of personal and physician costs did not show
any significant difference between the two modes of care. Key words: satisfac­
tion with surgery; outcome of surgery; costs of surgery. (Med Care 1985, 23:
171-182)

One-day surgery has been advocated as
an efficient means for increasing hospital
productivity in the context of cost con­
tainment.1,2 During the last decade, the
development of 1-day surgery has been
well documented.3-5 In response to vari­
ous measures that impose severe limita*From the Departement de medecine sociale et
preventive, Croupe de recherche interdisciplinaire
en sante, Universite de Montreal.
f From the Departement d’Administration de la
sante. Chairman, Croupe de recherche interdiscip­
linaire en sante. Universite de Montreal.
t From the Croupe de recherche interdiscip­
linaire en sante. Universite de Montreal.
§ From the Commission de la Sante et Securite du
Travail.
Supported in part by a grant from le Ministere des
Affaires sociales du Quebec. Editorial assistance was
provided by Louise Valois.
Address correspondence to: Raynald Pineault,
MD, Croupe de recherche interdisciplinaire en
sante, Universite de Montreal, 2375 Chemin CbteSte-Catherine, Montreal, Quebec, H3T 1A8.

tions on the supply of hospital beds, there
has been a tendency to admit increasingly
complex cases to 1-day surgery units.8
Much of the literature concerned with
1-day surgery has dealt solely with the or­
ganization and the functioning of these
units as well as with utilization variables
such as number and type of surgical proce­
dures performed, differential length of
stay, and use of resources7-9 Among
research-oriented studies, few have in­
cluded comparison groups in their re­
search protocol.10-13 Special attention has
been given to hernia repair.12-16 With few
exceptions, most studies have limited their
investigation to one or two of the three
aspects that our research embraces, i.e., pa­
tient satisfaction, clinical outcomes, and
costs.18-18
The development of this new surgical
mode raises an important question:(How
efficacious and efficient is 1-day surgery

171

5

Medical Care

PINEAULT ET AL.

compared with traditional inpatient care?
This question constitutes the central con­
cern of this article.’More specifically, the
objective is to compare 1-day and inpatient
surgery in terms of patient satisfaction,
clinical outcomes, and the cost of the
episode of care for three selected surgical
procedures: tubal ligation, hernia repair,
and meniscectomy.

Methods

Study Setting
The study was carried out in a Montreal
acute-care hospital with an inpatient
capacity of 350 beds and a 20-bed 1-day
surgery unit. This unit uses all hospital
facilities including operating and recovery
rooms. Nine surgeons participated in the
study: two gynecologists, three or- •
thopedists, and four general surgeons. For
each operation, the different surgeons
agreed to use the same techniques and
protocol.
For a better understanding of this article,
it is necessary to outline some of the main
characteristics of the Quebec medical care
system. Physician and hospital services are
totally covered under a national health in­
surance program providing free access to
these services. Hospitals are financed
through a global operating budget estab­
lished on an annual basis by the Depart­
ment of Social Affairs, whereas physicians
are directly paid on a fee-for-service basis
by the Quebec Health Insurance Board.
The fee schedule for surgical procedures is
established through an agreement be­
tween physicians’ professional associa­
tions and the government. Overbilling is
prohibited.

Selected Procedures and Surgical Modes
The criteria for selecting tubal ligation,
hernia repair, and meniscectomy were
1. The procedures had to be frequently
performed in the last current year before
the study began;
172

2. They had to be relatively complex in
order to include “border line’’ cases as well
as more simple procedures generally per­
formed on an ambulatory basis;
3. General anesthesia had to be used in
all cases;
4. The surgical procedures had to repre­
sent different major surgical specialties;
5. The performance of these procedures
in an ambulatory setting had to conform to
ethical requirements.
Two surgical modes were studied:
“One-day surgery” refers to the process by
which the patient is admitted the morning
of the operation and discharged the same
afternoon. “Inpatient surgery” refers to the
more traditional process, whereby the pa­
tient is admitted for a hospital stay of at
least one night. In both surgical modes, the
operation was performed under general
anesthesia.
The decision as to whether a patient was
eligible for inpatient or 1-day surgery was
made by the surgeons, on the basis of
explicit criteria e.g., the severity of the
condition, the existence of previous or
chronic health problems, and the patient’s
age. Our research protocol also specified
that no other surgical procedure be per­
formed concurrently to avoida|he com­
bined effects of multiple procedures on
outcome measures. Furthermore, all sub­
jects had to be 18 years old or older to
circumvent the requirement of obtaining
parental consent.
Once the patients were found to be eli­
gible for 1-day surgery, the surgeon in­
formed them that the procedure could be
carried out in either setting. If a patient
expressed a preference for one mode of
care, he was then booked according to his
personal preference, but not included in
the study. Those stating no preference
were invited to participate in the experi­
ment. They were told that the treatment
choice would be left to chance. Those who
accepted were asked to sign an informed
consent. In all cases, the operation was
performed by their own physician.

6

Voi. 23, No. 2

Patient Satisfaction

0/

Since consumer acceptability can be a
contri buting factor in the development and
widespread use of a new program such as
1-day surgery, patient satisfaction becomes
an important element in this study. Patient
satisfaction has generally been measured
.^ither directly or indirectly. Direct meas; ureFbFsatisfaction are' obtained by asking
the patient to what extent he (she) is satis­
fied with various elements of the medical
care process. There is a great deal of litera­
ture on this subject.19,20 The problem with
direct measures of satisfaction is that they
are not specific enough and thus fail to
discriminate between different modes of
care.
For these reasons, our study retained in­
direct measures of satisfaction with regard
to patients’ perception and their assess­
ment of the process of care. Specifically,
the following indicators were selected:
1. accessibility, as measured by percep­
tion of distance between home and hospi­
tal, controlling for real distance;
2. physician availability, as measured by
at least one postoperative visit (excluding
follow-up visits);
3. patients’ opinion concerning the ap­
propriateness of the length of stay, and
overall preference for the alternative mode
of care.

Clinical Outcomes

The clinical outcomes evaluated in this
study come from two sources: the patient
and the medical chart. The patient’s view
was obtained on the following variables:
the seriousness of discomfort felt in the
first 24 postoperative hours and the self­
reporting of postsurgical problems. The
medical chart provided objective data re­
garding complications, general health
status, symptoms, and complaints.
Costs of the Episode of Care

The episode of care is defined in this
study as the period of time from the sur-

ONE-DAY SURGERY

geon’s request for the patient’s admission
up to the 3rd postoperative month. Total
costs for an episode of care have three
components, each of them financed by a
different party (Table 1).
Hospital Costs. Previous attempts to es­
tablish the potential savings associated
with 1-day surgery were made by referring
either to average daily costs or direct pa­
tient charges. Both costing methods are
considered inaccurate since they do not
take into account all the types of services
received by the patient.21 A more appropri­
ate technique consists in identifying and
costing all services received during an
episode of care. The financial comparison
between a 1-day and an inpatient episode
can then be performed more accurately.
For this purpose, assessment of hospital
costs per episode of care was carried out in
two steps: 1) determination of all services
used by the patient during the episode of
care. The .utilization data obtained from
medical records included items such as
number of inpatient days, number and type
of diagnostic tests (laboratory, radiology,
electrocardiogram), number and type of
medication, number of units of physi­
otherapy, operating room and recov­
ery room time, ward nursing care time, and
home nursing care visits; 2) calculation of a
unit cost for each service. Since the 1-day
surgery program was an established unit,
already in operation within the hospital
confines, no capital expenditures are con­
sidered in the cost accounting process.
Table 1.

Financing the Costs of the
Episode of Care

Components

Financed by

Hospital costs

Physician costs

Department of Social Affairs
through the hospital’s
operating budget
Quebec Health Insurance

Personal costs

basis
Patient

Board on a fee-for-service

173

-II J .1.IU111 .L. .

7

PINEAULT ET AL.

Medical Care

Table 2. Patient Distribution by
Surgical Procedure and Mode
of Care (n = 182)
Mode of Care
Surgical
Procedure

One-day

Inpatient

Total

Tubal ligation
Hernia repair
Meniscectomy

31
32
30

Total

93

30
31
28
89

61
63
58
182

Rather, it is based on the hospital’s 19791980 annual operating expenses. Because
it is a global budget that does not provide
actual costs for individual surgical proce­
dures or episodes, a step-down costing
technique was applied.22 This is done by
allocating support costs (e.g., plant over­
head, housekeeping, laundry and linen,
dietary, central supplies, medical records,
and admission costs) to patient treatment
costs, which finally produces a unit cost for
each service. After these two steps had
been taken, a disease costing analysis was
performed in order to compute total costs
for each episode of care.23,2411
Physician Costs. Since under the
Quebec National Health Insurance
scheme, uniform fees are paid to physi­
cians and overbilling is prohibited, it is
possible to determine the cost of physician
services throughout the episode of care by
multiplying the tariff contained in the fee
schedules by the number of services
rendered.
Personal Costs. There are few personal
expenses in the national health insurance
system. But the financial burden imposed
on the patient and his/her family during
the episode of care may be different for the
two modes of care. Information pertaining
to this type of cost was gathered through
the questionnaires, in order to substantiate
this presumption. Included are transporta­

tion, domestic help, baby sitting, medica­
tion, supplies, special equipment, room
charges, # and private physiotherapy costs.
Cost estimations for unpaid help and salary
loss were also calculated. The main pur­
pose here was to determine if personal
costs increase when surgery is performed,
in the 1-day mode, since the hospital does ■
provide medication, support, and hostelry
services during immediate postoperative
recovery for inpatients.
Data Sources

h

Study Population

During the study period from October
1979 to March 1981, a total of 672 patients
were operated for the three selected pro­
cedures: 249 for tubal ligation, 296 for her­
nia repair, and 127 for meniscectomy. Of
this total, 295 patients (44%) were effec­
tively considered eligible for 1-day surgery’ ’
by their surgeons. Of this number, 182
(62%) were randomly assigned to either
mode of care. Of the remaining subjects, 54.
(18%) chose their mode of care, 13 (4%)‘
served as pretests, three subjects refused to*
-----------

1 See reference 24; this report can be provided by
authors on request.

174

J

Financial data were obtained from the
Finance Department’s record files. The’
other data source is three questionnaires >
The first was a home interview conducted.,
on the 7th postoperative day. Subsequent!
telephone interviews were done 1 and 3 j
months after the operation for follow-up;
purposes. The first questionnaire gathered ’
information on patient satisfaction, im-1
mediate clinical outcomes, and costs of the I
episode of care. The secondandjhird were]
aimed at collecting further information on J
these parameters. In addition, the medical ;
record of each patient was reviewed to
identify the use of specific services and
evaluate clinical outcomes.
;

# Hospital can charge for a semiprivate or private '
room requested by the patient without medical ‘
prescription.
4

8

ONE-DAY SURGERY

Vol. 23. No. 2

participate (1%), and 43 (15%) were lost to
the study because of communication prob­
lems with the surgeons. The overall par­
ticipation rate can thus be established at
84%. This article, however, is concerned
only with the 182 patients that have been
randomly allocated to the two alternative
forms of surgical care. Distribution of the
study population by mode of care and for
selected procedures is shown in Table 2.
Since the sample size is small, the
groups were compared for several factors,
e.g., age, income, education, and previous
hospitalizations. No significant difference
was found between the two groups.

Results
Accessibility. This variable was
expressed by the patient’s perception of
the distance between his home and the
hospital. As shown in Table 3, the total
study population of the 1-day surgery
group found the distance between the hos­
pital and their home significantly longer
than their hospitalized counterparts, al­
though an objective measure of actual disTable 3. Patient Perception of the
Distance from Home to Hospital by
Surgical Procedure and Mode of Care
(n = 182)
Mode of Care
Surgical
Procedure
Total for all
procedures
Too far
Not too far
Tubal ligation
Too far
Not too far
Hemia repair
Too far
Not too far
Meniscectomy
Too far
Not too far

°P « 0.05.

One-day

Inpatient

(%)

(%)

100.0
15.0
85.0
100.0
19.4
80.6
100.0
6.3
93.7
100.0
20.0
80.0

100.0
6.0
94.0
100.0
6.7
93.3
100.0
3.2
96.8
100.0
7.1
92.9

X2

4.33 n

2.15

0.32

2.01

Table 4. Patient Postoperative Contacts
with Surgeon by Surgical Procedure
and Mode of Care (n = 182)
Mode of Care
Surgical
Procedure
Total for all
procedures
At least one
visit
None
Tubal ligation
At least one
visit
None
Hemia repair
At least one
visit
None
Meniscectomy
At least one
visit
None

One-day

Inpatient

(%)

(%)

100.0

100.0

33.3
66.7
100.0

79.8
20.2
100.0

39.8“

19.4
80.6
100.0

60.0
40.0
100.0

10.5°

50.0
50.0
100.0

80.6
19.4
100.0

6.5 a

30.0
70.0

100.0

30.7 ”

X2

aP < 0.01.

tance between the hospital and their home
failed to show any significant difference
between the two groups.
Physician Availability. This variable re­
fers to the situatiomwhere at least one visit
has been made by the surgeon before the
patient leaves the hospital. As shown in
Table 4, only 33.3% of the 1-day surgery
patients had a visit from their surgeon
compared with 79.8% of inpatients. This
significant difference, seen for all three
surgical procedures, could be anticipated
since the reduced length of stay associated
with 1-day surgery makes it increasingly
difficult for surgeons to visit their patients.
Appropriateness of Length of Stay. The
patient was asked whether he/she found
the length of stay too short or appropriate.
As revealed by the data in Table 5,53.9% of
the 1-day surgery group think their hospital
stay was too short as opposed to 21.3% for
the inpatient group. Again, this significant
difference holds for all three surgical
procedures.

175

9

'I

PIN EAULT ET AL.

Medicae Caae^

Table 5. Patient Perception of the
Appropriateness of Length of Stay by
Surgical Procedure and Mode of Care
(n = 182)
Mode of Care
Surgical
Procedure

Total for all
procedures
Too short
Appropriate
Tubal ligation
Too short
Appropriate
Hernia repair
Too short
Appropriate
Meniscectomy
Too short
Appropriate

One-day

Inpatient

(%)

(%)

100.0
55.9
44.1
100.0
51.6
48.4
100.0
59.4
40.6
100.0
56.7
43.3

100.0
21.3
78.7
100.0
20.0
80.0
100.0
25.8
74.2
100.0
17.9
82.1

X2

22.8“

6.6“

7.3“

9.3“

aP < 0.01.

Preference for Alternative Mode of
Care. Patients were asked if, given their
actual experience, they would choose the
same setting again or the alternative mode
of care. The data in Table 6 are unequivo­
cal. For all three categories, a significantly
greater proportion of 1-day surgery pa­
tients expressed their preference for hos­
pitalization than did inpatients for 1-day
surgery.
Cx
sumniary, hospitalized patients seem
to express a greater degree of satisfaction
than short-stay patients. Differences be­
tween the two groups are both important
and statistically significant. Furthermore,
dissatisfaction with 1-day surgery is much
greater among meniscectomy patients.
This seems to indicate that, for meniscec­
tomy, 1-day surgery is a less acceptable
form of treatment than for the two other
conditions. )
Clinical Outcomes
The study found that clinical outcomes
as a whole, differed very little between
1-day surgery and inpatient care.

Postoperative Complication Rate. Post-’^
operative complications, i.e., complica.^
tions occurring before the patient’s discharge from either surgical facility, were S
established at 5.3% for 1-day surgery and -i.8% for inpatients. These differences are %
not statistically significant. The slightly1®
higher rate for inpatients could well be ex-*^
plained by the longer observation period £
that averages 2.7 days for inpatients com3
pared with 8.7 hours for 1-day surgery pa-^
tients. The nature of the reported compli-’^
cations were relatively minor, none of J
them constituted an emergency or lifethreatening situation.
Severity of Postoperative Discomfort.^
Patients’ assessments of the severity of ?
postoperative discomfort were obtained J
during the home interview on the 7th post- ?
operative day. Data show no significant dif-W
ference between groups, and this for alLS
surgical procedures (Table 7). Approxi-®
mately 88%' of 1-day patients and 91% of

Table 6. Patient Preference for Alternative^
Mode of Care by Surgical Procedure
and Mode of Care
(n = 182)
Mode of Care

Surgical
Procedure
TotaJ for all
procedures
Same mode
Alternative mode
Undecided0
Tubal ligation
Same mode
Alternative mode
Undecided0
Hernia repair
Same mode
Alternative mode
Meniscectomy
Same mode
Alternative mode

One-day

Inpatient

(%)

(%)

100.0
50.5
48.4
1.1
100.0
56.7
46.3
1.0
100.0
53.1
46.9
100.0
43.3
56.7

100.0
86.5
13.5
100.0
93.8
6.7
100.0
77.5
22.5
100.0
89.3
10.7

iW
x

s.

26.3;j

“I
CTHf

4.09*®
23.54c.

“ Not included in analysis.

“P
0.05.
eP =s 0.01.

176

i

10

Vol. 23. No. 2

ONE-DAY SURGERY

inpatients rated their discomfort in the first
24 hours following surgery as being either
“not very serious” or causing “no discom­
fort.”
Postoperative Symptoms Rate. Patients’
assessment regarding the absence or pres­
ence of symptoms during the 3 months fol­
lowing their initial surgery is shown in
Table 8. Data analysis reveals that there is
no significant difference in the postopera­
tive symptoms rate between 1-day surgery
and inpatients. It should also be noted that
meniscectomy patients show a much
higher rate than hernia repair or tubal liga­
tion, since 53.3% of 1-day and 60.8% of
inpatientjneniscectomies still indicate the
presence of symptoms at the 3rd postopera­
tive month. Patients were also asked the
Table 7. Patient Perception of the
Seriousness of Postoperative Discomfort
by Surgical Procedure and Mode of Care
(n = 182)
Mode of Care

Surgical
Procedure

One-day

Inpatient

(%)

(%)

Total for all
procedures
100.0
Very serious or
serious
11.9
Not very serious or
not serious
82.8
No discomfort
5.4
Tubai ligation
100.0
Very serious or
serious
9.7
Not very serious or
not serious
90.3
No discomfort
Hernia repair
100.0
Very serious or
serious
9.4
Not very serious or
not serious
87.6
No discomfort
3.1
Meniscectomy
100.0
Very serious or
serious
16.7

X2

100.0

8.9
86.5
4.5
100.0

0.5

3.3

1.0

96.6
100.0

9.7
83.9
6.5
100.0

0.4

14.3

70.0
13.3

78.6
7.1

(n = 180)
Mode of Care

Surgical
Procedure
Total for all
procedures
No symptoms
Symptoms
Missing data“
Tubal ligation
iNo symptoms
Symptoms
Hernia repair
No symptoms
Symptoms
Missing data0
Meniscectomy
No symptoms
Symptoms

One-day

Inpatient

(%)

(%)

X2

100.0
72.0
26.0
2.0
100.0
90.3
9.7
100.0
78.1
15.7
6.2
100.0
46.7
53.3

100.0
72.0
28.0

0.07

100.0
90.0
10.0
100.0
87.1
12.9
100.0
39.2
60.8

0.00

0.00

0.32

“Not included in analysis.

general question, “How do you feel,” 1
month and 3 months after their operation.
Here again, the data (Tables 9,10) indicate
that the recovery period is much longer for
meniscectomy, but no difference cpuld be
found between hospitalized and short-stay
patients.
In sum, the different parameters used to
evaluate clinical outcomes reveal a great
similarity between the two surgical groups.
In this regard, 1-day surgery can be con­
sidered as efficacious as inpatient care.
The case of meniscectomy, however, de­
serves special attention. It is clear that the
recovery period is much longer for menis­
cectomy than for other surgical procedures.
Since our observation period covered the
first 3 postoperative months, this study
cannot assess long-term recovery for these
patients.
Costs

Not very serious or

not serious
No discomfort

Table 8. Patient Reporting of the Presence
of Symptoms Three Months after
Surgery by Procedure and Mode of Care

0.7

Hospital Costs. Average hospital costs
for each surgical procedure are presented

177

11

PINEAULT ET AL.

Medical Cake

Table 9. Self Rating of Health Status by
Patient One Month After Operation
by Surgical Procedure and Mode of Care
(n = 180)
Mode of Care

Surgical
Procedure

Total for all
procedures
Good
Fair
Bad
Missing
Tubal ligation
Good
Fair
Bad
Missing
Hernia repair
Good
Fair •
Bad
Meniscectomy
Good
Fair
Bad

One-day

Inpatient

(%)

(%)

100.0
74.0
16.0
9.0
1.0
100.0
81.0
6.0
10.0
3.0
100.0
81.0
19.0
0.0
100.0
60.0
23.0
17.0

100.0
76.0
14.0
10.0
0.0
100.0
86.0
7.0
7.0
0.0
100.0
84.0
13.0
3.0
100.0
58.0
21.0
21.0

X2

1.3

1.2

1.4

in Table 11. One-day surgery costs range
from a low of $278.62 for tubal ligation to a
high of $816.66 for meniscectomy. Hernia
repair occupies a middle ground at
$367.58.11 Inpatient procedures follow the
same progressive increase in costs, reflect­
ing the relative importance of resources
used in the treatment of each condition.
Analysis shows thatinpatienthospital costs
are significantly higher for tubal ligation
and hernia repair and are mainly imputa­
ble to nursing care costs. The itemized cost
breakdown also indicates that 1-day tubal
ligation incurs significantly higher outpa­
tientvisits costs. Because a high proportion
of 1-day surgery patients do not see the
physician before their discharge, it is pos­
sible that they prefer consulting the sur­
geon rather than their family physicians for
subsequent follow-up visits. The higher
cost of diagnostic services is related to an
U All money figures presented are in Canadian dol­
lars.

178

increased use of test procedures for tubal
ligation inpatients.
Meniscectomy presents a totally differ­
ent pattern, since average hospital costs are
significantly higher in the 1-day mode. Al­
though 1-day meniscectomy continues to
show lower overall nursing care costs, al­
beit the added expenses of home nursing
care, a condition set by the surgeons in this
study, any potential saving is offset by the
substantial increased use of physiotherapy
services. This rise in costs for treatment
services would tend to support the clinical
evidence that has found more severe post­
operative symptoms associated with this
mode of care.
Physician Costs. In a national fee-forservice system in which overbilling is pro­
hibited, the costs of medical services
should not vary considerably. As expected,
no significant difference was found in total
medical costs between 1-day and inpatient
surgery (Table 12).
4

Table 10. Self Rating of Health Status by
Patient Three Months After Operation by :.J
Surgical Procedure and Mode of Care
(n = 182)
'■

Surgical
Procedure

Total for all
procedures
Good
Fair
Bad
Missing
Tubal ligation
Good
Fair
Bad
Hernia repair
Good
Fair
Bad
Missing
Meniscectomy
Good
Fair
Bad

Mode of Care
—-------One-day
Inpatient

(%)

(%)

100.0
72.0
24.0
2.0
2.0
10.0
90.0
10.0
0.0
100.0
78.0
16.0
0.0
6.0
100.0
47.0
47.0
6.0

100.0
72.0
21.0
7.0
0.0
100.0
90.0
7.0
3.0
100.0
84.0
16.0
0.0
0.0
100.0
39.0
43.0
18.0

X2

4.2

1.2

2.0

1.7

.=

12

Vol. 23, No. 2

ONE-DAY SURGERY

Table 11.
Mode of Care
One-day

Surgical Procedure

($)

Tubal Ligation
Outpatient visits
Diagnostic services
Operating room
Nursing care“
Medication
Total
Hemia Repair
Outpatient visits
Diagnostic services
Operating room
Nursing care*
Medication
Total
Meniscectomy
Outpatient visits
Diagnostic services
Operating room
Nursing care®
Home nursing care
Physiotherapy
Medication
Total

Inpatient

($)

Value
(t)

48.35
46.43
113.05
62.75
8.04
278.62

38.46
54.74
116.25
147.58
7.84
364.87

2.44 6
- 3.65*
- 0.66
-12.59*
0.27
- 8.55*

47.54
33.83
215.79
62.07
8.34
367.58

41.47
34.27
212.12
185.97
8.85
482.68

1.14
- 0.14
0.28
-13.70*
- 1.24
- 5.65*

85.00
70.18
200.35
64.20
94.38
274.33
8.22
816.66

87.71
74.96
210.68
200.10

- 0.40
- 1.52
1.02
-13.48 e

41.49
9.18
644.12

3.94*
1.42
2.76

“Nursing care costs include hostelry costs.
bP
0.05.
CP =« 0.01.

Table 12.

Average Physician Costs by Surgical Procedure and Mode of Care

Mode of Care
Surgical Procedure
Tubal Ligation
Visits
Surgery
Diagnostic services
Total
Hemia Repair
Visits
Surgery
Diagnostic services
Total
Meniscectomy
Visits
Surgery
Diagnostic services
Physiotherapy
Total

One-day

Inpatient

($)

Value
(r)

23.84
164.42
35.60
223.86

20.28
163.78
37.79
221.85

0.90
1.74
-1.93
0.47

24.42
204.31
15.46
244.19

20.78
199.45
17.98
238.21

0.60
1.19
-1.46
0.75

48.93
191.48
49.46

50.41
191.38
49.87
27.36
319.02

-0.19
0.10
-0.29
-1.41
1.32

($)

289.87

179

*

13

PINEAULT ET AL.

Table 13.

Medical Cake'/
Average Personal Costs by Surgical Procedure and Mode of Care

J

Mode of Care
Surgical Procedure

Tubal Ligation
Transportation and domestic help
Medication
Room
Total out-of-pocket
Unpaid help
Salary loss
Total
Hemia Repair
Transportation and domestic help
Medication
Room and other
Total out-of-pocket
Unpaid help
Salary loss
Total
Meniscectomy
Transportation and domestic help
Medication
Room and other
Total out-of-pocket
Unpaid help
Salary loss
Total
aP

One-day

Inpatient

($)

($)

18.11
3.49
21.60
75.35
63.73
160.68

19.40
1.87
19.27
40.54
113.76
73.10
227.40

17.09
5.65
0.75
23.50
45.51
210.33
279.33

31.79
3.15
32.06
67.00
41.09
368.53
476.62

121.18
22.28
7.30
150.76
69.23
146,95
366.94

74.22
21.38
49.45
145.05
48.49
266.54
460.08

Value
(t)

■'

-0.26

1.93

-3.10*
-1.28
-0.15 •
-1.08 ,

-1.11
1.85
-2.92\
0.24 -i.3o:
-1.60 j
7

1.59 '
0.18 7
’’'J

0.18 i
1.13 "■

-1.02
-0.68

0.01.

Personal Costs. Data presented in Table
13 show that there is no statistically signifi­
cant difference for total personal costs be­
tween the two surgical modes. However,
out-of-pocket expenses, i.e., actual dollars
paid by the patients, were significantly
higher in the case of inpatient tubal liga­
tion and hemia repair. This is mainly due
to supplementary semiprivate and private
room charges. Although inpatient menis­
cectomy also registers room supplement
charges, its significance is offset by higher
transportation costs for 1-day surgery pa­
tients who required an ambulance for their
home return. The high variability of esti­
mated personal costs for unpaid help and
salary loss accounts for the statistically
nonsignificant results found in total per­
sonal costs.

Discussion and Conclusion

In putting together the findings on pa-,
tient satisfaction, clinical outcomes, and,
costs, the following pattern emerges. Tubal
ligation and hemia repair seem to be surgi­
cal procedures for which 1-day surgery is
an appropriate form of care. More than 50%.
of 1-day surgery patients would accept toj
repeat their experience of 1-day surgery in*both conditions. In addition, this form om
care is cost efficient, since it represents anl
average saving of $86.00 and $115.00 perj
patient for tubal ligation and hemia repair,.?
respectively. Meniscectomy, on the other^
hand, represents the extreme case in which i
1-day surgery is associated with a lower;
patient acceptance rate of 43% and is actu-j
ally cost inefficient, since it increases total
•ji

180

14

Vol. 23, No. 2

hospital costs by an average of $173.00, in
comparison with hospitalization. Clinical
outcomes are similar in both modes of care.
■ The findings tend to support the general
consensus that 1-day surgery can be an ac­
ceptable, efiicacious, and cost-efficient or­
ganizational arrangement for performing
certain types of surgery.25 They also reveal
the limits of this form of care, which must
be implemented with concern for the po­
tential disadvantages associated with dif­
ferent types of surgical procedures that
may be performed on a short-stay basis.26
At another level, one must ask the dif­
ficult “cui boni” question, who benefits
from 1-day surgery, the patient, the doctor,
the hospital, or society? This study clearly
indicates that, for the patient, there is a
limit beyond which 1-day surgery becomes
a less acceptable form of treatment. As far
as cost is concerned, although one-day
surgery can reduce out-of-pocket expenses
by avoiding supplementary room charges,
it may also increase personal costs by im­
posing ambulance expenses, as seen with
meniscectomy patients.
In a fee-for-service mode of remunera­
tion for doctors, it is evident that shorten­
ing the length of stay of surgical patients
may increase the volume of surgical proce­
dures over a period of time, thus yielding
financial benefits to doctors, in a context
where the supply of acute care beds is
limited.**
The hospital may take advantage of
1-day surgery to maximize the use of
operating room facilities when occupancy
rates are high. After reaching this point, the
relative advantages depend on the prevail­
ing financing mechanism. A shorter
length of stay means increased average
daily costs. For example, in the Canadian
system, there is no real incentive for hospi­
tals to increase this type of productivity by
** In Quebec, the number of acute care beds per
1000 dropped from 5.32 in 1974 to 4.30 in 1978.

ONE-DAY SURGERY

shortening length of stay, since they are
financed through a global operating
budget established annually by the
government.
At the societal level, the question of the
utility of these measures must also be
raised. A program that has been proven to
be cost efficient at the organizational level
can be cost inefficient at the societal level.
For example, in a situation where rates for
surgical operations have been reported as
unusually high, possibly reflecting un­
necessary surgery, one can question the
rationale for extending eligibility of 1-day
surgery to cases normally requiring inpa­
tient care.27 Thus for society, 1-day surgery
must be assessed in the light of the alterna­
tive use of the beds that are freed by 1-day
surgery' as well as in terms of the quality of
care as reflected by the appropriateness of
these procedures.28
References
1. Grossman RM. Is ambulatory surgery less ex­
pensive? Hosp 1979;53:112.
2. Lairson DR. Baron EA, Swint MJ, et al. Donot-admit versus inpatient surgery in an HMO: de­
terminants of choice and implications for medical
care costs. Health Serv Res 1980;vol 16:378.
3. Ford JL. Outpatient surgery: present status and
future. South Med J 1978;71:311.
4. Davis JE. Day surgery: a viable altemajtive.
Assoc of Operating Room Nurses J 1974; 19:641*.
5. Mamak MB. The case for Ambulatory surgery.
Dimens Health Serv 1974;51:42.
6. Detmer DE, Buchanan-Davidson DJ. Am­
bulatory surgery’. Surg Clin North Am 1982;62:685.
7. Epstein B, Coakley OS, Levy ML. Outpatient
surgery. Hosp 1973;47;80.
8. Hawthorne DD. Hospital-based units im­
proves utilization. Hosp 1975;49:62.
9. Reed WA, Ford JL. The surgicenter: an am­
bulatory surgical facility. Clin Obstet Gynecol
1974;14:217.
10. Komhall S, Olsson AM. Ambulatory inguinal
hernia repair compared with short-stay surgery. Am J
Surg 1976; 132:32.
11. Shaw CP, Robinson GC. Kinnis C, et al. Day­
care surgery for children. Med Care 1972;5:437.
12. Adler MW, Waller JJ, Day I, et al. A ran­
domized controlled trial of early discharge for ingui-

181

15

PINEAULT ET AL.

nai hemia and varicose veins: some problems and
methodology. Med Care 1974; 12:541.
13. Adler MW, Waller JJ, Creese A, et al. Ran­
domized controlled trial of early discharge for ingui­
nal hernia and varicose veins. J Epidemiol Commun­
ity Health 1978;32:136.
14. Prescott RJ, Cuthberson C, Fenwick N, et al.
Economic aspects of day care after operation for her­
nia or varicose veins. J Epidemiol Community
Health 1978;32:222.
15. Rockwell E. Outpatient repair of inguinal
hemia. Am J Surg 1982; 143:559.
16. Carraway WM, Cuthbertson C, Fenwick N, et
al. Consumer acceptability of day-care after opera­
tions on hernia and varicose veins. J Epidemiol
Community Health 1978;32:219.
17. Marks SD, Greenlick MR, Hurtado AV, et al.
Ambulatory surgery in a HMO. Med Care 1980,
18:127.
18. Ruckley CV, Ludgate CM, Maclean M, et al.
Major outpatient surgery. Lancet 1973;24:1196.
19. Ware JE Jr, Davies-Avery A, Stewart AL. The
measurement and meaning of patient satisfaction.
Health Med Care Serv Rev 1978; 1:1.
20. Zastowny TR, Roghmann KJ, Hengst A. Satis­
faction with medical care: replications and theoretic
reevaluation. Med Care 1983;21:294.

Medical
21. Evans RC, Robinson GC. Surgical day
measurements of economic payoff. Can Med Assocp^
1980; 123:873.

22. Evans RG, Kinnis C, Robinson GC. A cmSV
analysis of alternatives to traditional inpatient care ihS?;
a children’s hospital. Unpublished final report oTtS?
project funded by grant 610-1103-43. Dept, of NiSgi
tional Health & Welfare, 1978.
23. Babson JH. Disease costing. Studies in Social^
Administration. Manchester University Press, Thexj
University Press, 1973.
24. Valois M, Bastian ML, Contandriopoulos, AP^»
et al. Evaluation d’un programme de chirurgie d’uifjy
jour: methodologie etresultats. Equipe de recherch^w
interdisciplinaire en sante. Montreal, 1982.
25. Velez Gil A, Galarza MT, Guerrero R, etd^jS
Surgeons and operating rooms: underutilized re~5p
sources. AJPH 1983;73:1361.
26. Miller JC. Ambulatory surgical centersSg
Forum 1980;3:19.
27. Vayda E. A comparison of surgical rates inlgg
Canada and in England and Wales. N Engl J Met
1973;289:1224.

28. Egdahl RH. Ways for surgeons to increase therfg
efficiency of their use of hospitals. N Engl J MedSffi
1983;309:1184.

CLINICAL ECONOMICS
MODULE 5

COST-EFFECTIVENESS ANALYSIS
After completing this module you should understand:

a.

how a cost-effectiveness analysis is undertaken;

b.

the strengths and weaknesses of the approach;

c.

what is meant by incremental cost-effectiveness
analysis.

1

COST-EFFECTIVENESS ANALYSIS

INTRODUCTION.
analys i s are used to choose
Cost and cost-minimisation analysis
when the
between
alternative
ways
of
using
resources
give
The
alternatives have
identical
benefits.
The
techniques
to
the
lowest
cost
alternative.
In
many
cases,
preference
It
both the benefits and costs of alternatives differ.
however,
rational
either
to
choose
the
least
cost
alternative
is then not
to benefits,
or the most effective alternative
regard
wi thout
Costs
must be compared to benefits in
to costs.
without regard
some way.
COST EFFECTIVENESS

(CE) ANALYSIS.

CE analysis measures the benefits of a health intervention
This can be in terms of either the
in terms
of physical units.
intermediate or the final health output.

compared three anti-smoking
Altman et al
example,
For
intermediate outcome indicator, the
and used an
interventions
or
to measure
the success
who quit,
smokers
number
of
In contrast, Cummings et al
effectiveness
of
the programmes.
measured the success of their anti-smoking intervention in terms
of a final outcome indicator, the number of years of life saved.
,y /1

a

*

ndicators <6f
<of effectiveness must be
Intermediate output- iindicators
to~ be
relevant
to the particular programme.
re 1evant
chosen carefully
indicators are found in the
Accordingly,
a wide variety of
luding
the
number
of
correct
diagnoses, the number
literature
inc
immunisations,
and
the
number
of new acceptors in
of successful
planning
programme.
However,
only
two final output
a family
indicators are used frequently:the numberf lives, and
(i)
(ii) the number of vears of Life (often called
”11fe-years”)
saved by an intervention.
economi s ts must rely on the epidemiological data
Clearly,
In fact, Drummond et al
i nf ormation about effectiveness.
for
stud i es are more commonly criticised for the
argue that CE
evidence on which effectiveness
med i cal
the
of
quality
the subsequent economics.
You
are based
than on
ind icatorjs
techniques and problems of estimating
should be aware of* the
the epidemiology modules.
In any case, some
f rom
ef f ect iveness
in this module illustrate the difficulties
of the ref erences
involved.

|

Ll

» V4 fX

U

£ to

0 0^

■,,'CCi >
r ^4 c
c I<

2
THE^SoST-EFFECTIVENESS RATIO.
is obtained by dividing the total cost of a
The CE ratio
It represents the
programme by the indicator of effectiveness.
of effectiveness for example, cost per life
cost per unit
or oas±-- per-- suoc,ftR.g.._fpr
saved,
cost per 1 £f^-year gained ,
In general, dthe—lQw.ex^ tbis
intermediate outcom.fi j ndi-QaXojis—
r a t. io_Atie7*mo?^ efji ci e n tl st h e-p r o g rammed

CE analysis can be used to compare alternative designs of
the
same
project,,
such as different methods_ of treating
project
dehydration
in children caused by diarrhoea.
It can also be
used to compare unrelated programmes
if
they have the same
objective.
Projects aimed at saving lives can—be compaxfid, but
it does not make sense,
for example, to compare the cost per
life saved of a medical
programme with
the cost per bTrth
prevented,
the latter being the CE index commonly used in family
planning projects.

L

is used as a means of choosing
Note
that CE analysis
It is meaningless to compute a
programmes.
between competing
CE index for a single project in isolation.
NET COSTS.

^7

>TY->

—\j r

*7

(J

(Vi

j

Xc/

v

The numerator in a CE ratio is usually the net direel- cost?
For example,
a hypertension treatment
example,
of the intervention,
drugs,
direct
costs
(doctor
visits,
programme
incurs
yet also saves direct costs by preventing
investigations etc),
is the direct cost of the
net_cost
future strokes.
The
in
direct costs resulting from
saving
intervention
minus
the

T h er e _i s_ci i s put e
(with appropriate discounting),
fewer strokes
about
whether
the__ economic
literature,
,
however,
in
the
life
s
ho
uld^
__
b
e__
±n
elude
d in the
of_ extending
consequences
For.
exampLe
,

the
hyperLenslon^contro
1
calculation of net cot. ‘c s .
people
to
work
longer
(an
indirect
programme may allow soire
but on
the other hand some of the people whose lives
benef it),
extended may develop cancer andxincur additional direct
w p^costs.
A fairly persuasive reason fox ignoring these_ef-f-ects is
into
the
are___ already -—implicitly —i-ncorporated
that _L1—
of benefit,
assuming that society
indicator
numerator,
the
values extensions of life regardless of the net productivity of
the life that is extended.



Ml?

r'

YJc
.



r

INCREMENTAL ANALYSIS.

et al
compared three ways of diagnosing symptomatic
Hull
(DVT) and measured effectiveness in terms
deep vein thrombosis
of the number of sufferers correctly diagnosed by a technique,
they
found that impedance
For a given group of sufferers,
(IPG)
cost a total of $395,359 (Canadian) and
plethysmography
142~~cases of DVT.
The CE ratio was $2784
correctly identified
IPG with Fibrinogen J.eg scanning cost
per correct diagnosis.

r

4 t/xZttx

I

r

I•-

i Jj

__________________ __



ft T

vyjjLc]

f' 4^1^ /kif

identi fled
$550,046 and correctly
indicator in this case was $2989 .

184

cases

L DVT.
of

Tty

<lr

The CE
<4/

the costs of making a correct
that
imply
These resu1ts
for IPG with leg scanning,
than
IPG
1 ower
f or
diagnosis are
that
the
first method
is
not necessarily imply
They
do
The second method costs more but it identifies more
preferable.
it is necessary to decide if it is worth spending the
cases,
so
extra money to identify the extra cases.
$154,687 but
cost an extra
leg >scanning
w i th
IPG
The cost per extra correct
42 cases.
identif ied an extra
called an incremental
This
is
$3683.
was
diagnos i s
The decision maker must decide if the
ratio.
cost-ef f ect iveness
extra cases could be identified in a more efficient manner using
technique.
However, if there is no alternative the
a di f ferent
it is necessary to decide
is more difficult because
question
$154,687 spent to identify an extra 42 cases
whether
the extra
of DVT could have been spent more effectively in an alternative
It
programme - for example, in a measles immunisation campaign.
this
type of decision using the
impossible
to make__ this
would be
used
in
the article
intermediate
indicator of effectiveness
programmes
like
measles
control
strategies
do not
because other }
_
identify cases of DVT.
Final
Final output
output indicators
indicators allow
allow such
such
comparisons, but at the cost of increasing complexity.
-- -D-rujnmond
e t al make a distinction between incremental and y iv&l
Incremental
arFaTysis’ •
cost-effectiveness analysis.
,margi rial
one programme
the additional costs and benefits
to
/ret ers
analysis describes the
imposes over another while max-g-inal
impact of an expansion or contraction in a given programmed

d DISCOUNTING REVISITED
Module 3
l! e
concept of discounting was explained in
to
costs.
The
r
uture benefits of an intervention which
II ' ivy**«r elation
CJv> ’c a^ be measured in money terms should also be discounted on the
ca^
(/c^g rounds
that society prefers
pref ers
income sooner rather than later.
there
is
controversy
about whether
future benefits
j•X ,J f,Ulowever
,
v
'
5*^ expressed in physical units, such as life-years gained, should
be discounted.

The consensus at present
is that
thej^slu^uld^Fe.
The
is technical.
Discounting costs but not effects
major reason
lead to
logical
inconsistencies in reasoning.
The issues
can
are discussed more fully in Drummond et al.


^7^4

rif

'/' ••

/ /» ■ / ■ •'s/'t

i

^5-3
:

y

At <

,

4
REFERENCES
Major References:

Drummond et al,

(see Module 4), chapter 5.

Creese A.L.
"Cost effectiveness of potential
immunization
interventions against diarrhoeal disease”, Social Science &
Medicine 1986,23(3):231-240.
Other References:

’’The costAltman D.G., J.A.Flora, S.P.Fortmann & J.W.Farquhar,
effectiveness
of three smoking cessation programs”, American Jnl
of Public Health, 1987, 17:162-165.

<
Cummings S.R.
et
al ,
’’The cost-effectiveness of counselinf
smokers to quit.”, JAMA 1989, 261(1):75-79.
Hull R,
J.Hirsh, D.L.Sackett &. G.Stoddart,
’’Cost effectiveness
of clinical diagnosis,
venography and noninvasive testing in
patients with symptomatic deep-vein
thrombosis”,
NEJM, 1981,
304:1561-1567.

QUESTIONS
ANSWERS TO ALL QUESTIONS SHOULD BE PREPARED BEFORE THE CLASS

1.
a.

b.
c.

d.

9
e.

Read the article by Creese.
Are final or intermediate outcome indicators used? What
are the advantages and disadvantages of each type (in
general)?
f..
How do you think the results would have differed if
effectiveness had oeen measured in terms of the number of
years of life rather ‘'han the number ofZ/1 iv.es saved?'"
Discuss whether it is preferable to use years of life saved
or the number of lives saved as an Tndfcator of
---ef f ectiveness?
Discuss the way that costs were measured, inoXudlng^wh.os.e
viewpoint was taken, whether any important costs were
omitted, and whether marginal or average costs were used?
What policy conclusions can be drawn from this article?

2.
What are the advantages of CE analysis over cost and cost­
minimization analysis?
What are some of the weaknesses of CE
an«dys is?
3u
Would you expect the net costs of any programmes to be
negat i ve?
If so, what type of programmes?



6

d.
e.

What difference does it make to your CE ratios if you find
that only 100 patients comply with 120 drug courses?
You can now compare the CE ratio of the scheme with the
education to the CE ratio without education under the two
assumptions about compliance.
If the CE ratio with
education is the higher, does this imply that the education
programme is worth undertaking? If not, how should the cost
effectiveness of the education programme be assessed?

(Wtj <'
t

J r-1

l^y
* 'z- •'

! 5^

\6

5I>

2. lr X .

fn

lu
pu/t K*-C^

Jc\

U r^tj-Dk-

•r-

co

QrtcC-

t /tx-o^

5

4 . a.
Calculate the CE ratio for the following programmes.
Rank
them in order of cost effectiveness.
(Taken
from
Shepard
&
Cash,
Manual
for assessing
the
cost-effectiveness of oral rehydration therapy in the treatment
of diarrhoeal disease, 1983J
Programme

Annua] cost per
100,000 pop.

Annual deaths
averted/100000.

$400,000

303

oral rehydration

$18,000

75

immunization

$5,000

35

measles vaccination

$30,000

63

DDT spraying for malaria

$200,000

800

$7,000

69

pilot projects in
primary health care

home distributed oral
rehydration

b.

^7-

rv > t

7^' \ 7

/S’/'-f ■

Do these data imply that pilot projects in primary health
care should not be undertaken?

4.
You are responsible for a dispensary which provides drugs
worth
$400 per month,
You are worried that many of the patients
who receive
these drugs do not maintain the prescribed regimen
of drug
tak i ng.
(You accept tbat correctly taking the drugs in
the doses prescribed i
good for your patients.)
You feel that
much of your drug
d rug budget
is wasted by patients who are not
compliant.
In
an .'.th,
. th, the dispensary served
the
last m<
patients and
dispensed
250 differen'.
d i f f e ren
drug courses.
You have followed up
these patients and
found
that ;80>_were compliant, and that 88
regimens were
foilowed correct1yY" You wish
followed
to
provide an
educati on programme that will enhance patient compliance and you
wish
to discover
if the additional
cost of education
is
worthwhile.
You expect that the cost of Jthe_.e.<iucAtiDn programme
will be $10.0 — per month
and
that
it
would serve all of the
patients attending the dispensary.

a.
b.

c.

v

What is the cost nf drugper
J-„pat±ent_..uncler _ the exi std-ng 4^;^
d rUX d i s pe n sj ngs c hem(no
education/)?
u
-on// ? What is the cost per'
drug
- course?^
:---- f ■ 4^7^
What is the cost per patient under 'the scheme with
|
education?
With education, you expect that 120 ±patients
--------zz---wi11 complvjV'
with 144 drug courses,
Calculate a CE
(
rat±p for,the\scheme
without education, and for the scheme with eduoationA

\ S
\ "I s p
. I'f

"

.

/

4-4

' j

Cosi-EfFectiYeness of h Worksite 'Hypertension
Treatment Program
Alexander G. Lccan. M.D.. F.R.C.P.(C), Sahara J. Milne. B.Sc.N..
Chr—;n= Achber. R.N.,
Wendy ?. Camprell. M.A.. and R. Brian Haynes, M.D., Ph.D.,
LC.P.(C)

SUMMARY ..-.e cost-.fTeetirenttS of treating hypertension at the patient's place cfw-rc -as co —ared in
a ranoomued controlled trial with care delirered in the community. The arera- -al —- l! n"''ier' r‘r

I"

i

n n,7
< 0-001) but the Pinene cost
signincantlv less (S45.40 ,s $3- qh
" :ne me:,a Auction in diastolic blood pressure (SP) at the year-e.-.d aacess.ment w*as signinc.n.lY greater in the worksite group (12.1 = 0.6 »s 63 = 0.6 mm Hg.p < 0.001;. ~- mareme-tal cos^
eoectneness ratio_o. S5.o3 per mm Hg for worksite care was less than the base cost-5—beness” ratio of
S.-_l per mm ng for regular care, indicating that the worksite program was suos^criajjy more coste eettre. Our ..ndt.-.gs support health policies that faror allocating resources to work-based r.-errension treat­
ment programs for the target group identified in this study, t Hypertension J: 21I-2IX. l-sl j

Key-Words • cost-benefit analysis • 'allied health personnel •
industrial medicine • deiirery of health care • hypertension •
-occupational health senices • patient acceptance of health care

ambulatory care

?

T the present time, physicians in private
practice are largely responsible for the de­
livery of health care in North America.
Most surveys in the past nave shown that these pfactidoners have been successtul in getting no more than
30% of the total hypertensive population under good
blood pressure (BP) centre!?-3 The main reasons doc­
umented for tnis poor rate of success are low detec­
tion rates, high treatment dropout, and low compli­
ance with medication? An additional problem found
in the studies is the lack of therapeutic vigor in the
application of antihypertensive therapy by manv
physicians?

As a solution to these przx.c.-Tjs, nkerr.auve medicul care approaches to cor:-ci hypertension have
been developed?*” Akhouzh c most instances these
approaches have been judged :a de effective, safe, and
acceptable, detailed economi:z C"alyses of these pregrams have not been pubiii:
The importance of
such analyses is paramaun: caw (hat the benefit o_f__
treating mild hypertension 203 been convincinziy
demonstrated?1- 13 makinz it oeairubie to treat ev^n
larger numbers of peooie -::h available health
resources. The randomized a_cntroHed
trial• Ireported

~
W
I tw
here was undertaken to osseas tz= cosi-e'fectiveness of
a work-based hypertensicn program in which all cure
for hypertension was provideo onsite.

rrum (he Department of Medicine. University of Toronto,
i uruntu. Canada, and Departments of Clinical Epidemiology and
Sicsiaiistita and of Mee:c:.-.c. McMaster University. Hamilton.
Canada.
Supported by the Ontario Ministry of Health. Demonstration.
Medel Grant DM 23J.
Address lor reprints: Dr. A. G. Lozan. Suite 435. Meant Sinai
Hospital. Toronto. Ontario M5G 1X5. Canada.
Received April II. 1950; .*c-.-;s;cr, acceotcd August 22. 1930.

Merhods

211

Par.icipancs in th: tris!
s=!e=;-d frQm 2I.9C6
volunteers, aged 18 to 69 yearn, m 41 business localions in .Metropolitan Tcranic --20 were screened for
hypertension in 1976-77. There: •*i:h an average fifth
phase aiastoiic 3? greater thar r• mm Hg on the sec-

4-5
212

HYPERTENSION

ond and third readings were scheduled for-a second
screen I week later. At that time a nurse administered
a brief cligibility^uestiarmaire and took three more
3? readinzs.^Eligibility criteriaj^ere as follows: 1)
mean diastolic BP at or move 95 mm Hg, or a mean
diastolic a? between 91-94 mm Hg and mean systolic
3? greater than 140 mm Hg; 2) intention to remain in
employment tor the year following entry into the
stuay; 3) not on any treatment for hypertension for at
least 3 months before screening; 4) not on other daily
meatcaticns. oral contraceptives, or estrogen replace­
ment therapy; 5) not pregnant nor planning to become
so during jhe year of the study; aid. 6) no objections
from the ramily physician.'^nose eligible and willing^
to participate were scheduled 'Tor' a "thir'd’scHen"and>
received two attitudinal Questionnaires to be answered
ind returned at the third session. At Session 3, additional 3? readings were taken as well as baseline
laboratory tests (hemoglobin, white blood cell count,
urinalysis, serum electrolytes, serum creatinine, serum
cholesterol, serum uric acid, blood glucose, and an
electrocardiogram). Individuals who were still eligible
by 3? criteria and who had no evidence of remediable
secondary forms of hypertension were given an ex­
planation ci the study and invited to sign a consent
form, indicating their willingness to participate.
Participants were stratided for sex, median age, and
meaian diastolic BP, and thenCrandomly ahocateB^
within strata to treatment at the worksit£/(“worksite
care”, W5) or in the community from physicians in
private practice (“regular care”, RC). All WS pa­
tients were evaluated at entry by a physician to ex­
clude complicating or concurrent problems, to estab­
lish a goal BP, and to initiate antihypertensive
therapy. Long-term follow-up was provided at the
worksite on company time by two nurses who were
taught to manage hypertension according to a stan­
dard protocol13 and who reported to physicians of the
Hypertension Service of the Mount Sinai Hospital in
Toronto. An appointment with their own doctor was
made for ail individuals in RC groups. All screening
data including the results of the baseline laboratory
tests were enclosed in the referral letter to the family
doctor. At 6 and 12 months after entry, all par­
ticipants were assessed at work by a specially trained
B? technician who was unaware of group allocation.
A questionnaire was administered to determine
medication status, and three BP readings were taken.

Medical Care Costs



These were itemized for each patient under health
system and patient costs. Health system costs were
these attributed to cose-nnding and treatment. Patient
costs were those related to lost time from work or
leisure and travel. Casts incurred before randomiza­
tion (screening costs) w-re distributed equally across
betn grouos. Treatment costs redacted actual
di.Terence :n the costs generated by the two types of
hea.tn care delivery. Costs are expressed in 1977
/—
12.-

Vol 3, No 2. .March-April 1981

Screening Costs

Costs for all who underwent ar.y

part of screening
were included in the analysis. The costs
involved
------------- ini
screening included personnel, equip .ent and supplies,
travel, participants’ time, and ad. inistrativc costs,
Personnel costs for screening com ■ised the salaries
and fringe benehts of the five 3? technicians and two
nurses. For the latter, time wax ~orded prospec­
tively on a weekly log, including :.~vel, service, and
omce time, and was converted into monetary terms
using their salaries and fringe benenos. The cos: of the
laboratory evaluation, which included the interpre­
tation fee of the electrocardiogram by a cardiologist,
was determined using the Ontario Medical Associa­
tion fee schedule and was considered a personnel cost.
Equipment used in the study included seven sphyg­
momanometers and stethoscopes and one electrocar­
diograph. Equipment costs w-re calculated using an
annual depreciaiipjourate of 15.75“; on the purchase
price (The Canadian Hospital .Association’s average
annual depreciation rate for clinical equipment was
used, as described in the Ccrcd'c.e rzaspital Accounting iMar.uol, 1963.) Cost of the pardcioants’ time was
prorated according to hourly waj^
the year-end
assessment, and included waiting, service, and ques­
tionnaire completion time a: screening visits. Actual
wage values were obtained on
of the par­
ticipants. and an average annual income was substi­
tuted for missing data (source of data was Statistics
Canada). The administrative cos: was estimated as
30?o of the health system cos: of the screening
program and included arranging the screening opera­
tion, scheduling participants and technicians, print­
ing, telephone, physical facilities, postage, and other
such expenses.
Treatment Costs
Health system costs of trentmen included the
provision of care and laboratory examinations.

physician visits and laboratory exar •-nations was the
Ontario Government universal heal: : insurance plan.
Each participant had a contract number and, within a
contract, services were sorted by date for all medical
services received during the study year. For each
claim, ths service nlc ider.tined me physician who
provided the service, the specialty of the physician, the
diagnosis specined by the physician, and/or the type
of service provided and the amount paid. Diagnostic
and therapeutic procedures,,diagnostic radiology and
laboratory tests were designated by specihc codes.
Only services properly identidea as related to
hypertension were included in cost calculations. The
cost of the nurses’ service was cnicu.cied from weekly
legs by converting the total time spent in patient cure
activities (direct cure, travel, and pacer work related
:o patient care) into a dollar vaiue according to their
hourly wcze. The cos: of physician*-i lime :o supervise
the nurses was added to tne cos: aftr.e nurses’ service
and was cnicuiated cy converting
sBcr-:

WORKSITE TREATMENT PROGRAM/Logcn et al.

r

{.

Ir

a dojlar value using the average net income and annual
working hours of physicians in Ontario (source of data
w'aa~'the—'Ontario'Medical Association). ’ The total
number of days of hospitalization per participant was
obtained from the service file, and only those days
used for diagnostic evaluation and management of
hypertension, as determined by direct questioning at
the year-end assessment, were included in the cost
calculation. The average per diem cost of hospitaliza­
tion in Ontario in 1977 was used to determine hospital
cost rather than more elaborate methods because o(
the small numbers involved. Drug costs were deter­
mined from many of the insurance companies who
sponsored drug insurance programs in industries in­
volved in the study. Complete drug cost data were
available for 36.6% of those in the WS group and
44.1% in the RC group on medication. Missing data
were due to varying insurance company accounting
- -. practices at the different worksites rather than lack oi
patient cooperation. For those with no drug data ■
(15.1% of those on medication), the average cost for
those with complete data in each group was used. For
the remainder with incomplete data, an individual
average monthly drug cost was computed from
available data, and this value was used for missing
monthly data to calculate the total cost.
’ Patient cost for physician and laboratory services
was calculated in the following fashion. Log terms,
recording the distance travelled, and time spent in
travel, waiting, and service for a single visit to the doc­
tor’s office or laboratory, were obtained from 95 RC
and 82 WS patients. Time was converted into dollar
va 1 ue_using_the midp_qi_nt_pf the wage category_.to
which_the_jndividual belonged, and travel costs were
calculated by~converting distance into dollars at the
.rate of 17 cents per mile. .The total patient^cost for
single visits to the laboratory or doctor’s omc: was
then multiplied by the number of visits for these ser­
vices by the individual during the study year, as deter­
mined from the health insurance plan contract file.
For individuals who did net complete single visit
logs, patient cost was calculated as the product of their
actual wage and visit frequencies times the average
value for travel and time from the logs. In the WS
group, the waiting and service time for each participant to visit the nurse was obtained from encounter
forms completed at each visit. The patient cost of
hospitalization was taken as the monetary value ot
time lost from work.
Onlv complete data were used to compare results ior
individual cost items between groups. In the calcula­
tion of health system, patient and total costs, averaged
values for individual cost items were substituted for
missing results. In addition, a sensitivity analysis was
performed using extreme rather than average values,
us described below.
Effect of Program
The effect of each treatment program was
calculated as the average reduction in diastolic BP.
Entry BP was calculated by averaging all diastolic oP.

213

measurements at the first and second BP screens and
the endpoint BP by averaging all diastolic BP
measurements at the year-end assessment.
’ Cost-Effectiveness Analysis

The cost-effectiveness (C/E) of each program was
the ratio of the average cost per patient to average
reduction in diastolic BP over 1 year. The incremental
^C/E_ratio was obtained by dividing the petjnerease in
medical carj_cost7bf~the~'WS program by the net in­
crease in effectiveness7~No discounting .of.future.costs
and effects _was employed because of the short duration of the study.,
Sensitivity Analysis

v

Assessment of_the_effects_of variation in .key
estimated'parameters was carried out by substituting
for missing data the maximum cost in the WS group
and minimum cost in theKC group..In the WS group,
the highest individual drug cost from the group with
complete data were used in all patients on medications
for whom no data were available. To calculate the total
cost for individuals with incomplete drug cost data,
the highest monthly cost was substituted for missing
monthly data. In the RC group, the least drug cost
was substituted for missing data in an analogous
manner. The maximum patient cost_to_visit—the
physiegarfs qmce £nd laboratoryjn the WS.group .was
calculated by. using the longest_time (travel, waiting,
and service) and the farthest distance when no timedistanee _data_were availabIe_and the highest salary
category when no wagecategpry was designated. The
minimum value for each was used in the RC group ,
whenever data were missing. For patient cost of nurs­
ing visits, the highest participant’s salary was used
whenever salary data were missing. Patients com­
pletely lost to follow-up were not included in the sen­
sitivity analysis because they were small in number
and evenly distributed between two groups (8% in each
group).
Statistics

Means are presented with a standard error of the
mean as the index of dispersion. Statistical analysis
was carried out using the unpaired two-tail Student’s t ■
test, with p value < 0.05 indicating a statistically
signif.cnnt difference. The chi-scuarc statistic was
used to assess differences in proportions.
Results

The volunteers who were screened for hypertension
represented approximately 50% of the employees
olfered this service. As shown in figure I. 457
employees, or 2.1% of the initial population screened.,
were eligible and witling to participate in the study.
Selected characteristics of the two study groups are
outlined in table I. and no significant differences were
noted. Data on cost and effect were obtained on 214
(92.2%) employees in the WS group and 207 (92.0%)

• 2U'

HYPERTENSION
HRS’ SCREEN



Vol 3, No 2. March-April I93i

Treatment Costs

N-21.90d-

Health System Cost

03P

90 mm Hg

OB? > 90 mm Hg (10.4%|

89.6%

N=2,273

Daclin* Second Screen

SECOND SCREEN

23.9%

N=I,7aa

Y
S? Not E!»vat«d

ELEVATED BP’ (3.9%l

~~ 5127.

Non- participants
Xd.37.

• DB?

-Nsasi---

PARTICIPANTS (2.1%)
N = 4S7

95 mm Hg or 91-94 mm Hg plui lynolic 8? > HO mm Hg

Figure I. Flo^ diagram showing results of two-stage
blood pressure screening in industry and government, 1977.
The percentages in brackets represent the percent of initial
population screened.

in the RC group. The other 36 patients had either no
cost or effect data and were excluded from further
analysis.
Screening Casts

The cost of individual screening items is outlined in
table 2. The total screening cost was S 102,009; the cost
per study participant. S223; and the cost per employee
with elevated BP after two BP screenings. 5120. Per­
sonnel and participants’ costs were the two largest ex­
penses. accounting for 86.1% of the total expense.

Table I. Comparability of Worksite Care (WS) and Reg­
ular Care (RC) Participants at Entry by Selected
Characteristics
Patient data
ws
RC


Age (mean yr)
Systolic BP (mean ram Hg)
Diastolic BP (mean mm Hg)
Male («=>)
White (%)
Known hyperTension (%)
B? measured in past year (%)
B? indicates blood pressure.

232
46.3
152.9
100.3
80.6
88.0
37.9
52.5

225
46.3

153.9

100.4
76.9
88.0
38.7
54.2

Table 3 summarizes the health system cost of care
of the two programs. As expected, the mean govern­
ment insurance expense was higher in the RC group,
being 576.03 ± 3.19 compared to 553.17 ± 2.92 in the
WS group (p < 0.001). ine WS expense included both
the cost oi laboratory tests ordered by the WS care
team as well as the cos: of any care for hypertension
received from community physicians during the study
period. Although 37.5% of WS patients including
study dropouts made physician visits, the frequency of
their visits was substantially lower (2.9 per annum?
compared to 5.7 in the RC group; p < 0.001). The
mean cost of the nursing service alone was 567.38 ±
J.29. The frequency of visits to the nurse was 8.6 per
annum, being significantly higher than the number of
physician visits in the RC group (p < 0.001).
There were four admissions to hospital for
hypertensive evaluation and management in the RC
group, and the mean healm system cost was 51,080.71
± 280.62. No WS patients were admitted to hospital
for diagnostic assessment and treatment'.
Significantly more WS than RC participants were
on drug therapy at some point during the study (205 vs
145, p < 0.001). Moreover, 55.8% cf those on drugs in
the WS group, compared with 14.9% in the RC group,
were on more than one type of antihypertensive
medication (p < 0.001). Of those with complete data,
the mean drug cost in the WS group was significantly
higher (587.34 ± 7.16 compared to 551.01 ± 5.24 in
the RC group, p <_0.C0l), reheating the more fre­
quent initiation of efficacious therapy and the more
vigorous application of this treatment.
The total health system cost of hypertensive care
provided by the WS care team was 5197.36 = 4.99
compared to 5129.33 ± 13.34 by community physi­
cians, a difference that was significant (p < 0.001).

Patient Cost
As shown in table 3. the average patient cost for
. physician visits in the RC group was 565.57 ± 4.84.
which was significantly higher than the comparable
cost of 538.71 ± 10.54 in the WS group (p < 0.03).
The difference was due to more frequent physician
visits by RC participants. The average patient cost to
visit the laboratory was not significantly different in
the two groups, with the WS grouo cost being 537.40
- ± 18.24, and the RC group. 525.67 ± 4.38. The
monetary value of the loss of lime from work to visit
the nurse was 524.09 ± 0.92 per employee and. for the
four RC participants to be evaluated in hospital.
5372.25 ± 91.65 per patient. The total •putieni cost
was 545.50 = 3.23 in the WS group and 532.00 ± 6.20
in the RC group (p < 0.01).

Total Cost
Using treatment costs oniy, w* found that ths
average total cost of the WS program was S242.36 =
6.94 per participant, which was net signincnatiy

4-8

WORKSITE TREATMENT PROGRAM/Lo^ et al.
Table 2.

Cost of the Screening Program

Item variable

1

Hg respectively (table 4 and fig. 2). When screening
costs were also included, the C/E ratios were higher
(WS, 538.50 per mm Hg; RC, 566.82 per mm Hg).
The WS program, although more costly by 531-52 per
patient per year, was able to achieve an addftional
mean reduction in diastolic BP of 5.6 mm Hg. Thus,
the incremental cost of lowering BP in me WS
program, that is, Ihc'cbsfover and above R.C- was a
55.63 per mm Hg reduction. Because this is substan­
tially less than the cost of lowering BP in the RC
group (at 532.51 per mm Hg reduction), the WS
program was highly cost-effective by comparison. If
conventional treatment of hypertension (RC i con­
sidered worthwhile, it is clearly mere cost-effective to
replace RC with WS treatment for the target group
identified in this study.

Cost ($)

Personnel

41.139

Equipment/supplies

898

Travel

490

Participants' dme

46,724

Administration

12,758

Total screening cost

102,009

different from the cost of 5211.34 ± 18.66 for RC.
When screening costs were added to the treatment
costs, the difference in cost of medical care between
the groups was still not signincant.

Assessment of Incomplete Data

'' Effect

Data for the travel and time part of visi^ :o the
physician’s oface were complete for signiheandy more
patients in the RC group than in the WS group :,50.3%
vs 33.3'70 respectively, p < 0.025). No sigziacant.
differences in completeness existed for drug cost or the
patient cost to visit the laboratory. All other daiz were
available for more than 85% of patients ix both
groups.

The mean redaction in diastolic BP, the measure of
effectiveness, was 12.1 i 0.6 mm Hg in the WS group
and 6.5 i 0.6 mm Hg in the RC group (p < 0.001).
Cost-Effectiveness Analysis

The C/E ratios of the WS and RC programs using
treatment costs only were 520.07 and 532.51 per mm

Table 3.

Cost of Medical Care Per Patient for WS and RC, Total and Patients With Complete Data

Cost variable (S)
Total patients:
Health system cost
Total
Patient cost
Total

Patients vrith complete data
Health system cost: V*
Government insurance

Worksite

P.egular care

/

6

/'197.36 ± 4.99
< (214; 100%T>

(2^100%)^

/ 45.50 ± 3.23
. (214; 100%)*

/82.C0 = 5.20 \
( (207; 100%)’ )

129.33 ± 13.34

.

Signizcance (jj)

}

2 rr^—

C.GOL

0.01

" -

53.17 ±2.92
(214; lOOrn)"

76.03 ±3.19
(207; 100%)-

0.001

Dn:^

87.34 ±7.16
(75; 37%) •

51.01 ±5.24
(64; 44%)*

0.001

Nurse

67JJ3 ± 1.29
(211; 99%)-

Hospital

I

215

Patient cost:
Visiting doctor

1CS0.71 ±250.52
(4: 100%)-

38.71 ± 10.54
’ (26; 33%)'

65.57 ± 4.34
(95: 50%)*

0.00

Visiting laboratory

37.40 ± 18.24
(22; 25%)'

25.67 ±4.48
(29; 22%) •

0.53

Visiting nurse

24.09 ± 0.92
(184; 86%)-

Huspical

372.25 =91.65
(4: lC0%r

WS indicates work site core; and RC, regular care.
•Figures in parentheses are the number of participants for whom data are available used in czlculatias
the mean = SEJt and the percentage for whom the item is applicable, ror example, fur government
insurance, complete information wax available on all 214 WS patients, whereas for drujrs. complete
information on expenditures was available on 75 WS patients who represented 37% of those for whom
drugs were prescribed.

HYPERTENSION

Vol 3, No 2, Maich-Afsul 1981 }|

Table 4. Cost-Effectiveness Analysis of Worksite (WS) and Regular Care (RC) Programs
Cost-effectiveness
WS
'v RC
.<
WS-EC
Cost ($)
Treatment costs
242.86
211.34
- 31.52
r Treatment & screening costs
465.86n
/
^L52
434.34-1
Effect (mm Hg)
’ Reduction in diastolic BP
, 12.1-J
. 6.5J . 5.6


Cost-effectiveness ratio ($/mm Hg)
With treatment costs only
With treatment & screening costs
.. ,/ - .

J

... .-

20.07
/ 38.50



\

-

5.6"
5.63 .

<66.82)

f



5

?

- : ^4
A

<3^13

■HT

.

J -.•.Wf I

...

Patients with incomplete data did not differ from
patients with complete data in terms of entry BP,
year-end BP, change in BP over the study year, or
medication compliance.

■/
7
The health system, patient and total cchzs were
recalculated using rpaximum cost for
individual items for the WS group and minimsiEQ cost
for the RC group (table 5). The incremeatal C/E ratio
was still less than the C/E ratio for RC.
?
Sensitivity Analysis

I
|
1
4
j

300

Discuss™. <

:

-; i < i

We have shown that treatment of'employed |
hypertensives at their place of woiic is both more drec--l
tive and more cost-effective than usual care in‘‘The 1
RC
//
/i
community.
/
200
The C/E ratiosJbr the WS and RC groaps jvere i
/
calculated in each case under the assumption thaFthe T
effect was entirely caused by the treatment program.-f
T^his su pposcs tr if die tvvo
*** '* ^ r,*
been identified and treated, the group’s average~BP J
would be unchanged after 1 year. In the ahsrscr of a |
third “no treatment” control group, this assumption '
100
cannot be tested. In the report of the Medical
Research Council Working Party on miW-tomoderate hypertension, however, control subjects tak­
ing inert tablets or only under observation had ap
ap-­
. .. hi
- their
- - lEastcfe
- BP 1[
proximately a 5 mm• Hg fall
year after entry,.which was aftriimteri rp fF^F^nty
_______ _•____
the measurement
- __ 0
'-14
- toward the mean of the BP in tbc genera] popnSziSon.14 |

■ EFFECT
?
• To test the effect ofisuch
o&such a change, we reralculaicd
recalcuizleid the t
C/E
from the
C/E ratios
ratios after'subtracting
after'subtracting 5
5 mm'Hg
mm Tig &um
the effect,
effect, !
(reduction in diastolic blood pressure
and found that patients receiving RC had EnJe
finJe’.BP
BP ;
reduction (table 6 and fig. 2). The WS group,
group. oa the i
. in mmHg)
other hand, continued to experience 2 substzsitial ‘
K. Figure 2. Cost-effectiveness graph. The points represent
effect beyond the estimated natml
natural redaction inBP.
m"BP \
the treatment cost and effect of each program. If it is
The health system cost of the WS
WS program^as^*
program jvas**7
cionifirnntlv higher
hioh»»r> than
fh*in RCL<This
Df"*
'
assumed that, each group's average blood pressure (BP}
significantly
. was "• rribited J’
would be unchanged in the absence of identification, the^
primanly to the use of morejnedGcatiDn bjrtfee^
■ v -A'
slopes of the solid lines represent the cost-effectiveness’J-j
teaJP^ conlfpj hypertettsidfl
Ibe^ciSL 0^-'
_
( C/E) of each program. The interventionfor which the slope ■ ^parallel care from^communitir
T-"
J!5jE£ds ■
is steepest is the most rwt^f^ffffGi~£ffe£t. Under 'th^ 51 expense appeared
td reflect some mkial ambiratence' 7 ■■
assumption that diastpiic^^wwf^^^^rmm Hg ^^2 of WS •pati6nK->lo^idp«^i^^^g5aP>B
year without any exp^ndtld^^^i^^^lh^mall-dashetF / '- program without 'some SlaboraBih
p
_____________ .1- _ Z- t r~
.■
r
1 '
'■•:•'.•
'

vr
'
■ .•
lines represent the C/E ratio of each program. The incre- ' ■ ci an-However, the-in frequency of physirTan vr
mented C/E ratio is given by the slope of the large-dashed : /: the low dropout rate in the WS group snggeitoire
f
(W^j/tdicaies z patient acceptance of die medKal circ
~~:^,^r^^
. Imefconnecting
linefconnecting the RC and WS.
WS. points.
poims. (W^jjidicates
care provided
a ||^
* worksite^RC, regular care).
workplace.
a /fe

WS

i

i

»t

b

I

-j -

W







.

.

A W

;

...

JWfeS

; I.

4-10
WORKSITE TREATMENT PROGRAM/Logan er al.
217

Table 5.

Sensitiuiiy Analysis of Worksite (WS) and Regular Care (RC) Programs

Ccst

__________________________



______________WS

Health sysum cost ($)

RC

$249.44

Patient cost ($)

74 81

S13C.51

55.22

Total cost (S)

26.80

C/E treatment & screening costs ($/mm Hg)

35r

30 -

o

5c 25

cn
cn

LU

z

UJ
F“

o

UJ
U.

u.

RC

WS-RC

Treatment cost (S)

242-86

31.52

Effect (mm Hg)“

7.1

211.34
1.5

Cost-effectiveness (5/mm Hg)

34.21

140.89

O

o

5.63

10 -

5 -

0,

0.50
0.25
FRACTION CF PATIENT COST

l.o

0.75

J , *
0

Figure 3. Use offraction of patient cost tFPCl ue computing cost-ejfectiveness (C/E) ratios. The effect on in­
cremental C/E ratio far the worksite care ( WSl program is
shown under differing assumptions of FPC. Wkife health
system and patient costs may be of equal value I FPC - I j.
patient cost may also be valued at some fraction of the
health system cost. For all FPC values, the incremencad C/E
ratio (closed circles] is less than the C/E raiia for RC

5.6

•Subtraction of 5 ram Hg from the observed change in
diastolic blood pressure in the WS and RC programs.

\

05

.

WS

15 -

UJ
I

U
A Vi--

Co%t-<ifccnvcnei5

20 -

>

TABLE 6. Cost-egectiveness Analysts of Worksite (IVS)
and Regular Care IRC} Programs Assuming 5 mm Hg
Reduction in Diastolic Blood Pressure ujith no Treatment

i

24.71
24.71

equal value to patient cost in both groups is coz an un­
reasonable assumption.
Use of the work setting to manage hypene.ison has
many advantages. First it facilitates access to care for
a population for whom usual care in the comsnunity •

-both groups on C/E ratios is illustrated in figure 3.
Even if patient cost is completely ignored (fraction of
patient cost = 0) and as a consequence the health
system cost becomes the total cost, the WS program is
still more cost-enective since the C/E ratio for RC,
while falling to S19.90 per mm Hg, continues to be
higher than the incremental C/E ratio of S 12.15 per
A mm Hg.
. While all hypertensive care in the WS program was
given on company time, in the RC group visits to the
doctor may have been made either on company or
leisure time. Because time away from work is not
closely monitored in the white collar companies that
participated in this study, it was not possible to quan­
titate this cost. In our cost calculations, equal value
was assigned to the time lost from work in the WS
group and from work or leisure in the RC group.
Since loss of leisure time may not represent a cost to
society (no effect on worker productivity), it may be
argued that patient cost in the RC group should be
valued at some fraction of the patient cost in the WS
group. Using different fractions of patient cost for the
RC program only, we found that the incremental C/E
ratio for the WS program was less than the C/E ratio
for RC until the fraction of patient cost was 0.01. We
know, however, from the year-end questionnaire that
51.7% of RC patients (roughly equivalent to a frac­
tion of patient cost = 0.52) stated that they took lime
off work to visit their physician. Moreover, since this
lime often appeared to be in excess of the actual time
required to obtain medical care, the assignment of

W'-«

19.59

—_

WS pregram was the
r,£iiSi*on.in
cost’ wnHc the health system cost
Q\^^,.Progr^
more expensive.. Up to this point
we have assumed that patient and health system costs
are of equal value. It may be argued, however, that the
former should be valued at some fraction of the latter.

I

23.59

X'45?2iP)

C/E indicates cost-efectiveness.

I

$118.83

138.40

C/E treatment costs only ($/mm Eg)

I
I

WS-RC ‘

.

program (dosed squares/. While equal value wx*r a^ngnrd
for patient cost in the two groups (FPC
IL based on
different assumptions the patient cost in the RCgreusp mav be some fraction of that in the WS group. The incremental
G/c ratio for the WS program (open circles) is les: than the
<-/£ ratio for RC (closed squares) until the FPC ix <0.01.

I

n
ns

HYPERTENSION

may be inconvenient. Over a third of the participants
at entry were previously aware of having hypertension,
and almost one-half had not had a BP measurement
, for more than I year. Second, the population reached
at the worksite is primarily middle-aged men in whom
the risk’ of adverse consequences of hypertension is
large and the benefit of therapy is more likely to be
high?3 Third, as we have shown here, it is a costeffective alternative to primary-care practice for the
treatment of hypertension. Finally, health care facili­
ties are already available in most places of work with
200 or more employees (in Ontario this is mandatory),
which eliminates the need for high capital expense to
develop worksite hypertension care programs.
The economic impact of cardiovascular and
cerebrovascular disease is enormous, ranking fourth
as a cause of sick leave absenteeism among male
employees in a heavy industry plant in Ontario?4 In
the same study, the mean work days absent per sick
leave episode was much higher for this disease group
han for all other diagnostic categories, and vascular
deaths accounted for almost half of the total mortality
among male employees during the assessment period.
Since hypertension is a major independent risk factor
for cardiovascular and cerebrovascular disease in the
adult17*20 and antihypertensive drug treatment will
reduce morbidity and mortality from hyperten­
sion?1- 2X*U effective industrial hypertension control
programs have great potential for improving produc­
tivity and/or reducing costs associated with
absenteeism or premature death for those employees
with asymptomatic, uncontrolled hypertension.
From a policy perspective, decision makers may
lake into account any or ail of the following objec­
tives: I) to spend the limited resources available for
hypertension screening and treatment in a way that
•/ill maximize the average BP reduction: 2) to identify
and reduce the BP of hypertensive patients by a
specified amount as economically as possible; and 3)
to maximize the BP reduction per patient. In each in­
stance. the worksite program is more advantageous.
Thus.-for the target group identified in this study our
findings support health policies that favor allocating
resources to work-based . hypertension treatment
programs.
Acknowledgments
The following organizations cooperated with mis study: Abitibi
Paper Cu.. Ltd.: Bank of Montreal: BP Canada: Burns Fry Ltd.:
Canada Packers Ltd.: Canada Permanent Trust Ca- Canada Wire
Jc Cable Ltd.: Canadian Broadcasting Corp.: Canadian imperial
Sank of Commerce: Canadian National: Clarkson Gordon and Co.:
Dominion Dairies Ltd.: Federal Government of Canada: Gilbcy
Canada Ltd.: Maclean«Hunter Lld4 National Trust: Ontario
Government: Royal Bank of Canada: Royal Trust: Shell Canada
Ltd.: Texaco Canada fnc.: Toronto Board of Education: Toronto
Dominion Bank: Toronto Stan Woody Gundy Lid. We thank Dr.
George W. Torrance. Faculty of Busineis. McMaster University.
Hamilton, Ontario, for his valuable comments on the manuscript.

References
I. Schoenberger JA. Stamier J. Shekei’e R3. Shexeile S: Current
status of hypertension control in industrial pcruiauon. JAMA
121: 559. 1972

Vol 3. No 2, March-April 198 i ’

2. Heller RF: Detection and treatment of hypertension in an inner
London community. Br J Prev Soc Med 30: 253. 1976
3. Shapiro M. Bleho J. Curran M. Farrell K. Klein D. Weigensberg A. Weil K: Problems in the control of hypertension in the
community. Can Med Assoc J 118: 37. 1978
4. Haynes RB. Sackett DL. Taylor DW: Practical management of
low compliance with antihypertensive therapy: A guide for the
busy practitioner. Clin Invest Med 1: 175. 1978
5. Ontario Council of Health. A report of the Ontario Council of
Health: Hypertension. Toronto, Ontario: Ontario Council of
Health. 1977
6. Wilber JA. Barrow JG: Reducing elevated blood pressure. Ex­
perience found in a community. Minn Med 52: 1303. 1969
7. Finnerty FA Jr, Shaw LW. Himmelsbach CK: Hypertension in
the inner city. Pan 2: Detection aad follow-up. Circulation 47:
76. 1973
8. Runyan JW Jr. The Memphis Chronic Disease Program. Comparisons in outcome and the nurse's extended role. JAMA 231:
264. 1975
9. Alderman MH, Schoenbaum EE: Detection and treatment of
hypenension at the work site. N Engl J Med 293: 65. 1975
■10. Hypenension Detection and Follow-up Program Co-Operative
Group. Therapeutic control of blood pressure tn the Hypenen­
sion Detection and Follow-Up Program. Prev Med 8: 2. 1979
IL Hypenension Detection and Follow-up Program Co-Operative
Group. Five year findings of the Hypenension Detection and
Follow-up Program: I. Reduction in mortality of persons with
high blood pressure, including mild hypertension. JAMA 242:
2562. 1979
12. The Australian Therapeutic Trial in Mild Hypertension.
Report by the Management Committee. Lance: I:.!26l. 1980
13. Logan AG. Milne BJ. Achber C. Campbell W?. Haynes RB:
Work-site treatment of hypertension by specially trained
nurses. A controlled trial. Lancet 2: 1175^ 1979
14. Medical Research Council Working Party on Mild to
Moderate Hypenension: Randomized controlled trial of treat­
ment of mild hypertension: design and pilot trial. Br Med J 1:
1437. 1977
15. Weinstein MC. Stason W3: Hypenension. A Pelicy Perspeclive. Cambridge. Harvard University Press. 1976
16. Gibson ES. Haynes RB. Martin RH: Vascuiar disease in
employed males. J Occup Med 17: 424, 1975
17. Society of Actuaries: Build and Blood Pressure Study, 1959,
(vol. I). New York. P.F. Mallon. Inc.. 1959
18. McGee D. Gordon T: The results of the Framingham Study
applied to four other US-based epidemiologic studies of cardio­
vascular disease. In The Framingham Study: An
Epidemiological Investigation of Cardiovuscuiur Disease.
Publication No. (NIH) 76-1083. Washington. DC. US Dept of
Health. Education and Welfare. April 1976. section JI
19. Pooling Project Research Group: Relationship of blood
pressure, serum cholesterol, smoking habit, relative weight and
ECG abnormalities to inadencc of major coronary events.
Final report of the Pooling Project. J Chronic Dis 31: 201. 1978
20. ShurtleiT D: Some characteristics related to the incidence of
cardiovascular disease and death: Framingham Study, a 18yeur follow-up, M The Framingham Study: A Epidemiological
Investigation of Cardiovascular Disease. Puciicanon No.
(NIH) 74-599. Washington. DC. US Dept of Health. Education
and Welfare. February 1974. section 30
21. Veterans Administration Co-Operative Study Group on An­
tihypertensive Agents: Effects of treatment on morbidity in
. hypertension. I. Results in patients *ith diastolic blood
pressures averaging 115 through 129 mm He. JAMA 202:
1023. 1967
22. Veterans Administration Co-Operative Studv Group on Antihypertensive Agents: Effects of treatment on morbidity in
hypertension: U. Results in patients with diastolic blood
pressures averaging 90 through 114 mm Hg. JAMA 213- I 14J
1970
23. Veterans Administration Co-Operative Study Group on An­
tihypertensive Agents: Effects of treatment on morbidity in
hypertension. III. Influcncs of age. diastolic pressure, and prior,
cardiovascular disease; further analysis of sidc-etfccu. Circula­
tion 45: 991. 1972

*

Randomised Controlled Study of Chloramphenicol vs Ofloxacin
in the Treatment of Enteric Fever :
A Pharmacoeconomic Analysis
D Mathai*, Abi Abraham M*, KR Johni**, Mary V Jesudason***, MK Lalitha***,
Molly Thomas****,, OC
' Abraham*, TJ John*
INTRODUCTION
Globally, each year, 13.5 million persons are af­
fected by typhoid fever. It remains a serious health
issue in the developing world and is a major reason for
hospitalisation. The traditional treatment with
chloramphenicol is long, usually given for 2-j weeks,
is associated with relapses and exposes the patient to
needless toxicity. Besides these, the causative organ­
isms, Salmonella typhi and Salmonella paratyphi,
exhibit varying degrees of resistance 20%-74% in dif­
ferent areas.
Therefore, an effective alternative choice would be
beneficial. Ciprofloxacin, a fluoroquinolone, is given
orally,. 500 mg twice daily for 14 days4 or 750 mg
twice daily for 7 days.5 It is a safe, bactericidal drug
and has an efficacy of 90% to 100% in the treatment
of typhoid fever.4 In an open non-randomised trial we
have shown that pefloxacin is safe, curing effectively
25 of 25 culture positive cases of S. typhi at doses of
400 mg administered twice daily for periods of 5-7
days.^ ofloxacin, another quinolone could be an ideal
choice for the treatment of typhoid fever as it concen­
trates well in tissues, especially in the biliary tree, and
also within macrophages and needs to be given orally
only once a day as it has excellent bioavailability. ’

Moreover, the need to deliver health care efficiently
in the developing world has made economic evalu­
ation of care a major consideration. The costs of
health care are largely supported by the national health
plan and/or by the individual payer. Thus, cost-con­
tainment and effective' therapy with newer but more
expensive drugs need to be addressed. We therefore
conducted a prospective, randomised, controlled trial
to study the safety and efficacy of ofloxacin in com­
parison with chloramphenicol, in the treatment of
*Department of Medicine; **Community Health;
***Microbiology and ****CIinical Pharmacology, Christian
Medical College and Hospital, Vellore. India.
Received : 19.5.1998; Accepted : 28.8.1998

JAP! 1998, VOL -16, NO. 12

acute uncomplicated typhoid fever occurring among
adults and hospitalised at a large, referral, teaching
centre. A pharmacoeconomic analysis, from the pur­
chaser’s perspective, of the choice of the higher priced
drug ofloxacin, against the less expensive drug
chloramphenicol was done, and we developed a clini­
cal decision analysis model for physicians to
determine the choice of initial therapy using either
drug.

MATERIAL AND METHODS

Study patients
Adults patients (> 16 years) with fever of over 5 days, clini­
cally resembling enteric fever, (seen between March 1991 and
December 1992) were enrolled in the study after obtaining in­
formed consent They were admitted in the hospital and
randomised to receive either chloramphenicol (CHLORO) 500 mg
six hourly for 10 days (Group-l), or oxfloxacin (OFX) 400 mg
once daily for 10 days (Group-II). Blood culture, blood smear ex­
amination for malarial parasites and microfilaria, and urinalysis for
pus cells were done on all patients. Only those with S. typhi or S.
paratyphi A isolated in blood culture were continued in the study.
Pregnant or lactating women, those known to have allergy to
nalidixic acid, subjects with serum bilirubin > 3 mg% SCOT or
SGPT > 120 U/L, creatinine > 2mg%, concomitant infection re­
quiring other antibiotics, gastrointestinal bleeding, intestinal
perforation or encephalopathy and those requiring parenteral ther­
apy were excluded from the study. The study protocol was
approved by the research review board of the institution.
The sample size was calculated based on bacterial clearance
rates after 48 hours. In the first assumption, clearance with oflox­
acin was 80% and chloramphenicol 60%; and in the second
assumption, clearance with ofloxacin was 95% and with chloram­
phenicol remaining at 60%. Therefore, it was ’estimated that for
fulfilling the first assumption, 91 cases in each arm was needed
and for the second assumption, 27 in each arm, would be required
to prove a significant difference with a type I error (alpha) of 5%
and the power of the test being 80%.
The randomization was done by computer generated random
number. The blinding was not done as the outcome chosen was
clearance of organism noted by microbiologist at the end of 48
hours. The time of deferevescence was noted by the nursing staff
not involved in the study.
Microbiological techniques
Five ml of blood was inoculated into each of biphasic infusion
991

medium, MacConkpy biphasic medium and brain-heart infusion
broth, which were incubated at 35°C. Any organism grown was
subcultured on MacConkey and blood agar plates and identified
using standard techniques. Antibiotic susceptibility was tested by
the disc diffusion method of Kirby and Bauer using the newer in­
terpretation chart. The minimum inhibitory concentration (MIC)
was determined by agar dilution method using Mueller-Hinton
agar with 18-24 hours incubation in air and with an inoculum of
105 organisms/mi. Serum levels of ofloxacin were determined by a
bioassay using Pseudomonas aeruginosa ATCC 27853 as the test
organism.9
The stool sample was directly inoculated onto plates of xyloselysinc-desoxychoialc agar, dcsoxycholalc-citrale agar and
MacConkey agar. In addition, about 1-2 ml of liquid fecal suspen­
sion dr about I gm of solid fecal matter was inoculated into a tube
containing selenite F enrichment broth. The latter was subcuitured

on MacConkey agar and Salmonella-Shigella agar.
Patient Monitoring and Follow-up

Ofloxacin, (Tarivid 200 mg : Hoechst) (OFX) 400 mg, was
given orally once a day or chloramphenicol (Chloromycetin :
Parke Davis each cap. 500 mg every 6 hours) for 10 days. Oral
temperature was measured every 6 hours. Defervescence of fever
was defined as achieving temperature not > 99.5° over a 24 hours
period. The hematocrit leucocyte count, platelet count and meas­
urement of serum creatinine, bilirubin, SCOT, SGPT, albumin,
globulin and alkaline phosphatase were done on admission and
prior to discharge. Blood cultures were drawn on day 1,2 and 3.
Stool culture was done on day 7, and at least once during the fol­
low up period. The patients were followed up through hospital
visits, letters and house visits. Where stool samples were collected
from patients in their homes, phosphate buffered glycerol saline
was used to preserve viability of salmonellae.
After giving at least four doses of ofloxacin or chlorampheni­
col, scrum levels of the drugs were determined. Blood for drug
assay was drawn 30 minutes after an oral dose (for peak level) and
30 minutes before the next oral dose (for through level).

Tabic 1 : Baseline characteristics of the patients at randomisatioT,

Duration of illness prior to
hospitalisation (Mean ± SD)
Male
Age (Mean ± SD)
Body mass index (Mean ± SD)
Weight, kg (Mean ± SD)
Height, cm (Mean ± SD)

OFX Group
rr-32

CHLORO ^roup
n-32

1227 (±8.1)

12.00 (±8.1)

22(68.8%)
31.91 (±22)
18.81 (±33)
51.03 (±9.8)
I64.I7(±8.1)

20(62.5%)
27.47 (±113)
18.91 (±22)
48.53 (±8.4)
159.73 (±82)

I;ifty-fivc were culture positive; Salmonella typhi in 47, and
Salmonella paratypji A, in 8. There were 28 patients in-Group-I
and 27 in group-H. In Group-I, 18 patients had in vitro resistance
to chloramphenicol detected after randomisation, and as blood cul­
ture remained positive, 16 were switched to receive ofloxacin, 1
continued to receive chloramphenicol and 1 dropped out Four (2
in each group) did not complete the period of observation in hospi­
tal and were excluded from the analysis, which had 10 patients in
Group-I and 41 in Group-II (Table 2).
Table 2 : Summary of patients enrolled, culture-positive and
culture-negative cases, changes to alternative study medication and
drnpouts/withdrawals

Chloramphenicol
(CHLORO)
Group-i

Ofloxacin Total
(OFX)
Group-ll

Patients enrolled at
randomisation
Culture-negative cases
(withdrawn)
Total culture-positive cases
In-vitro resistance to study
drugs

32

32

64

5

4

9

27
18

28
0

55
?
(I patient
clinically im­
proved and wa<
thus continued
in the same
group despite
in vitro
resistance)

Switched from CHLORO
group to receive study drug
♦Subtotal
Withdrawals:
- discharged against medical
advice
- non-study antibiotic therapy
during trial

0

16

11

44

I

2

0

1

10

41

Statistical Analysis
Qualitative parameters were tested by Fisher’s exact test and

quantitative parameters by Mann-Whitney test.
Economic analysis:9'12

The following direct medical costs were considered.
Drug cost referred to approximate retail costs of chlorampheni­
col, (Chloromycetin: Parke Davis) Rs. 2.50 per 500 mgm capsule.
Ofloxacin (Tarivid : Hoechst) Rs. 46 per 400 mg tablet
Cost of antipyretics : Paracetamol Rs. 2 per day.

Standard hospital bed charge at Rs. 50 per day.

Cost of blood culture and sensitivity testing Rs. 200/Pcrspcctive of analysis was that of the individual being the
payer. The indirect cost of morbidity and mortality was not as­
sessed.
Sensitivity analysis, both one way and two way of the assump­
tions to varying susceptibility of the organism to the drug and cost

of hospital stay was done.
RESULTS
Sixty-four patients were randomised; 32 received chloram­
phenicol (Chloro) (Group-I) and 32 received ofloxacin (OFX)
(Group-ll) Table 1 shows the baseline characteristics of the pa­

• •Total

-A- <;
51

Despite 1:1 randomisation, the imbalance in the patient numbers was due
to in vitro resistance seen in 18 of27 patients assigned to CHLORO, 16 of
whom were switched to OFX. For costing purposes the 27 and 28 patients
originally allocated to the CHLORO and OFX study arms, respectively,
were considered.

Eleven patients had received ampicillin, chloramphenicol o
co-trimoxazole prior to hospitalisation but they were included ir.
the study, as organism could still be isolated from blood.
The time required for defervescence of fever, was shorte
among patients who received ofloxacin (81.95 ± 36.39 hrs) com

tients at randomization.
JAP! 1998. VOL 46, NO. 1
992

}

«p

cceived chloramphenicol (125.33 ± 56.00 hrs)
pared lo those who rccei
patients (76.61%) who received ofloxacin had

who received chloramphenicol, (p < 0.05) (Table 3).

Side effects
Two patients who received ofloxacin complained of nausea
and vomiting. There were no untoward effects reported in the
Chloro group. In the OFX group one patient developed perforation
of the bowel for which she was operated.

Table 3 : Response to therapy

Clinical parameter

CHLORO OFX
n=4l
n=IO

Defervescence by day 5, n (%)
Mean time to defervescence. ± SO
in hours
Mean no. of signs and symp.
by day 4

2 (20%)
125.33
(±56)
1.80
(±1.1)

P value

31 (75.6°/o)
81.95
(±36.69)
0.59
(±1.3)

p = 0.001
p = 0.005

pathogens.
p = 0.02

Micro-organisms
All 55 isolates (100%) were susceptible to ofloxacin whereas
17 (31%) were susceptible to chloramphenicol. 21 (38%) to cotrimoxazolc and 25 (45%) to ampicillin (Table 4). In the Chloro arm
of study 18/27 (66%) was resistant to chloramphenicol. In the
ofloxacin arm there was no resistance. The diflerencc was signifi­

cant p - < 0.0002).

Table 4 : Microbiology-antibioiirnm of blood isolates
Number susceptible (at baseline)

Ofloxacin Chloramp- Cotrimoxa
-zoie
•henicol

Study
group

Organism
isolated

OFX
(n=28)

S. typhi
(24)
S. paratyphi
(4)
S. typhi
(23)
S. paratyphi
(4)

CHLORO
(n=27)

55

Total

Follow-up
All
All 64
64 patients
patients were followed up for safety parameters. Forty
four of the evaluable patients were followed up for 14-32 weeks
(mean 14 ± 12.6) after discharge from the hospital, and stool sam­
ples collected after cessation of therapy were negative for enteric

Ampicillin

24

4

7

S

4

4

4

4

23

5

6

9

4

4

4

4

55 (100%)

17(31%)

21(38%)

25 (45%)

Economic analysis
The details of the costing based on typical case treated by each
arm of the study is given in Table 7. The average cost per case
cured in the chloramphenicol arm is Rs. 1163/- while the average
cost in ofloxacin arm is Rs. 964/-. The incremental cost tor treat­
ing one case by chloramphenicol and subsequently switching over
to ofloxacin is Rs. 199/- per case cured with the assumption of bed
cost being Rs. 50/- and 67% showing resistance to chlorampheni­
col with the extra hospitalization of 15 days each.

The robustness of these assumptions were further tested using
one-way sensitivity analysis by changing drug resistance to
chloramphenicol (Ref. Fig. 1). At 40% drug resistance the cost ex­
ceeds Rs. 964/- which is the mean cost fixed for treatment with
ofloxacin. Thus if there is more than 40% resistance to chloram­
phenicol. use of ofloxacin is justified. In Fig. 2 a two way
sensitivity analysis is done by changing the drug cost and degree
of resistance to chloramphenicol. If the bed cost is Rs. 200/- even
a resistance level of 20% to chloramphenicol will justify the

choice of ofloxacin.
The only untoward effect observed with ofloxacin was nausea
which occurred in 2 patients. Perforation of bowel was more likely
to be a complication of the disease as it occurred on day 1 and
with discontinuation of the drug ofloxacin. It needed to be in­
cluded for documentation of untoward effect and as this was not
attributable directly to a side effect of the drug, it was not costed in
the ofloxacin arm.

Minimum inhibitory concentration (MIC) of ofloxacin and
chloramphenicol were estimated for 36 strains of S. typhi and 6
strains of 5. paratyphi (Table 5). Peak and trough serum levels

were estimated for both drugs (Table 6).
Table 5 : MIC of ofloxacin and chloramphenicol

S. typhi

S. paratyphi

(Range)
(Mean)
(Range)
(Mean)

OFX pgm/ml

CHLORO pgnVml

0.015-0.62
0.016 = 0.01
0.07-0.62
0.24 ±0.18

1.87-240
157.7 ±97.7
1.87-240
42.9 ±96.9

Table 6 : Serum drug levels
Drug ievei pgm/ml
Peak
Trough
Ofloxacin
Chloramphenicol

Range
Mean
Range
Mean'

0.124-31.74
5.05 ±7.37
15-240
64 ±52.6

0.124-1.98
0.99 ± 0.62
0.248-15
5.6 ± 6.9

DISCUSSION

Ofloxacin given for 10 days in doses of 400 mg
once daily was successful in treating all the 41 patients
with typhoid and paratyphoid fever. Ofloxacin also
produced more rapid defervescence of fever compared

to chloramphenicol. There were no relapses and none
became chronic carriers during the follow-up period.

Ofloxacin also cured the acute infection in the 8 pa­

tients with paratyphoid fever.
The MIC 90 of the isolates 5. typhi and S. paratyphi

obtained ranged from 0.016 ± 0.01 mcg/mi (range
0.015 - 0.62 mcg/ml). The mean serum levels of
ofloxacin, 5.05 ± 7.37 mcg/ml (peak) and 0.99 ± 0.62
mcg/mi (trough), seen among patients were several
folds greater than the MIC of S. typhi or 5. paratyphi
A (Table 6).

Serum levels of the drug achieved in our patients,
even at the trough levels, were at least 15 times more

993

JAP! 1998. VOL 46, NO. 12

I able " : Costinu nl lhc clinical trial li" Indian rnpccM

Days

Number sensitive
Cost of antibiotic:
Duration treated
Cost for one day
Total cost
Number resistant
Extra cost of CHLORO
Extra cost of OFX to treat
resistant eases
Cost of antipyretics:
Duration in days
Cost per day
Total cost
Cost for culture and
sensitivity tests for
resistant eases:
Cost per culture and
sensitivity
Cost per 18 cultures
Cost of hospitalization:
Hospital charge per day
Hospital charge for drug­
sensitive cases
Hospital charge for
CHLORO resistant eases
Total (A+B+C+D+E+F+G)
Cost per ease cured
Extra cost per ease on
CHLORO

No.

CHLORO (n=27)
rZ
Total
Rs.

lOx

(B)
(C)

2x
10 x

10

9x
18
18 x
18 x

10
10 =
10 =
46 =

lOx

900

5x

28x
0

46
46“

12,880
00
00

360

8,280

I

2

5
(D)

OFX (n=28)
Total

28

10

(A)

Rs.

Rs.

9

27 x

2
2=

270

2x

28 x

2
2=

lOx

28 x

50
50

112

200
(E)

18x

200

3,600

(F)

lOx

9x

50
50 =

4,500

(G)

15 x

18 x

50 =

13,500

than the MIC of the organism.

We have also confirmed that there is resistance to
ampicillin, chloramphenicol and co-trimoxazole with
various rates, but all isolates were susceptible to fluroquinolones.5’6
Ofloxacin allows once daily dosing unlike twice
daily with ciprofloxacin and its oral dose of 200 mg
was equivalent to the oral dose of 500 mg.

All patients needed hospitalisation for 10 days as
the study was primarily designed to evaluate efficacy
and safety. All of the three quinolones we have stud­
ied ciprofloxacin,5 pefloxacin,^ and ofloxacin in this
study cleared organism from the blood within 24
hours. It is likely that the duration ol hospitalisation
could be less in the OFX arm as defervescence was
achieved sooner. A five day short course of ofloxacin
was effective for treatment of multidrug resistant ty­
phoid with such high serum levels of drug being
obtained.15-18

Indirect costs’which include cost of travel, food and
994

No.

Days

31,410
1,163
199

'

(A+D+F)

26,992
964

accommodation expenses of attendants when a famil} .
member (patient) is hospitalised would account foi 1
more than 50% of total expenses but is also excludec
as they were common to both. It is also assumed that
morbidity and mortality due to multi-drug resistant or
drug susceptible strain would be the same, if left un­
treated. Since typhoid is a self-limiting fever, those
likely to respond to chloramphenicol therapy despite
demonstrating in vitro resistance needs further sf y.
Costs of inefficacy due to non-compliance ir King
chloramphenicol, which needs a dosing frequency, ol
four times daily vs ofloxacin once daily, and the long
term costs of changes upon the intestinal microflora
and the ecological problem related to the inappropriate
use of the drug ofloxacin where chloramphenicpi
would have sufficed was not addressed. Typhoid fever occurs in populations subject to high
unemployment and the effect on productivity of thif
intervention cannot be calculated as it is assumed that
an individual who is absent from work can easily be
replaced. Changes in man-days gained or lost due tc
JAPI 1998. VOL 46, NO. L

CHLORO e

w

(D
0)
Q.



a:

c
tn

o
O

1500
1400
1300
1200
1100
1000
900
800
700
600
500
400

a
O'

©

964

OFX

.O'

r

i

300
200
100
0
0

10

20

30

40

50

60

70

80

90

100

Resistance to CHLORO (%)
. . .

line shows the threshold mean cost for treatment using ofloxacin at 40% drug
resistance to chloramphenicol



Shows the varying cost for different levels of resistance to chloramphenicol

Fig.
One-way sensitivity
sensitivity analysis
analysis showinn
showing
costs of ofloxacin and chloramphenicol at different levels of bacterial resistance to
Fig. II :: One-way
chloramphenicol and at a hospital bed cost of Rs. 50 per day.

early or late return to work of patients in either group
need to be considered. We could equate the average
daily wage of a labourer in India at Rs. 30/- a day.

Our conclusions in the phannacoeconomic analysis
relating to such broad variable factors as the, suscepti­
bility of the organism and drug costs related to
treatment.
This study indicates that initial drug costs alone
should not be the basis of choice of drugs. The cost ef­
fectiveness assessed with this reduction of 43.34 hr
(i.e. 2 days) in defervescence time among those
treated with OFX, translated into lower overall costs
despite the drug ofloxacin, being 4.5 times more
priced than chloramphenicol (Table 3).
In view of the recently emerging multi-drug resis­
tant strains of 5. typhi in India.1It is important to
make every attempt to establish the diagnosis by blood
culture and determine the antibiotic susceptibility pat­
tern of all isolates. The choice of the drug to be used.
JAP! 1998, VOL. 46. NO. 12

should be guided by these results or based upon
epidemiological studies of drug resistance. This cost
would need to be added to the chloramphenicol group
as susceptibility studies need not be done among those
receiving ofloxacin as quinolone resistant strains have
not been reported from India. An extra total cost of
Rs. 199/- for the Chloro group with the prevailing re­
sistance of 69% to chloramphenicol makes
justification for replacing chloramphenicol with oflox­
acin at a hospital bed cost of Rs. 50/- a day. At any
hospital cost less than Rs. 20/- a day. at 50% prevail­
ing drug resistance to chloramphenicol (point E Fig.
2), the use of ofloxacin becomes as cost effective as
chloramphenicol. The only untoward effect observed
with ofloxacin was nausea which occurred in 2 pa­
tients. Perforation of bowel was more likely to be a
complication of the disease as it occurred on day 1 and
with discontinuation of the drug ofloxacin. It needed
to be included for documentation of untoward effect
and as this was not attributable directly to a side effect
995

- ' 1
• i
r

600 -i

Rs. 5054

A

500 -

ci

CC

Prefer ofloxacin

co
CC

400 -

TJ
O>

cu
20

</)

k.


300 -

o

o
TJ
0)
^2

o
X

0)
U)

Rs. 2544

cu

♦—

200 -

c

75
’cl
cn

O

Prefer
chloram­
phenicol

w
o
o
*c5
o
H

Rs. 1713

♦100 -

D

0
0

----------- ------------ -----------10
20
30
40

Rs. 1044
Rs. 6^

T
50

T
60

i

70

80

90

WO

Resistance to chloramphenicol


Points A B C D and E .how the threshold points of total cost according to changing cost of
hospitalization and resistance to chloramphenicol

cured by chungiiig (he levels ol chloramphenicol
Tig. 2 : Two-way sensitivity analysis showing the cl led on variation in cost per case
resistance and hospital costs.
REFERENCES
of the drug, it was not costed in the ofloxacin arm. If
1. Mary Jcsudasan. Jacob John T. Multidrug resistant
the costs of toxicity of chloramphenicol, fatal pancy­
Salmonella typlu in India. (Letter) Lancet 1990: 336 (8709)
topenia (1 in 50,000), poor compliance, potential for
252.
relapse (4%), carrier state following treatment (3%)
2. Agarwal S. Madhu SV, Guleria JS. Talwar. A problem of
and complications are calculated, there would be fur­
emerging chloramphenicol resistance in typhoid fever,
ther savings using ofloxacin, as a quinolone is the
preliminary report JAPI1991: 39 (6): 443-4.
drug of choice for all of these clinical situations. From
3. Chakravarty 13. Jain N. Gupta 13, Rayvanshi I’. Sen M‘~
Krishna A. Chloramphenicol resistant enteric fever. J Indu
the view point of both society and the patient as payer,
Med Assoc 1993; 91 (I): I -3.
• if the other direct and indirect costs are also to be
4. Dupont II. Quinolones in Salmonella lyphi infection. Drugs
added, it would make chloramphenicol a less attrac­
' 1993; 45 (suppl 3): 119-24.
tive option than this newer quinolone ofloxacin to the
5.
Mathai D. Kudva GC. Keystone JS. Kozarsky P. Jcsudasi
practising physician in developing countries. Where
M. Lulitlui MK. et al. Short course Ciprofloxacin therapy l~.
multi-drug resistance similar to ours may be a prob­
enteric fever. JAPI 1993; 41 (7): 428-30.
lem, we would recommend quinolones as the drug of
6. Mathai D. Martin S. Jesudason MV. Lalilha MK. Thomas
first choice. Ofloxacin given once a day would ensure
Zachariah A. Abraham OC. Short course of Pciloxacm
better compliance and would ensure total eradication
typhoid fever. Indian Medical Gazette 1997; CXXXI (I) •
of the organism and particularly the multi-drug resis­
9-12.
7.
Gruneberg KN. I'elmingham DO. Hare MD. Robbins h
tant strains.

JAP11998. POL* 46. NO. .J

996

Journal of Health-System Pharmacy 1995; 52 (19 Suppl 4) :
S5-8.

Perry. Walla RA. The comparative in vitro activity oi
ofloxacin. J Antimicrob Chemotherapy Sep. 1988; 22 suppl C

:9-l9.
8. Helen Giamercllou. John Tsagarakis. Efficacy and tolerance
of oral ofloxacin in treating various infection. Drug 1987; 34

14. Souetre EJ, Qing W. Methodological
pharmaco-economics.
Fundamentals
Pharmacology 1994; 8 (2): 101-7.

(suppl I): 119-23.
9. Narendranalh M. Paulose George T. Chloramphenicol and
ciprofloxacin in enteric fever cost effectiveness. JAP1 1992;

15. Smith MD. Comparison of ofloxacin and ceftriaxone for
short course treatment of enteric fever. Antimicrob Agents
Chemotherapy 1994: 38 (8): 1716-20.

40(3): 176-8.
10. Deborach A. Freund Robert S Dittus. Principles ol
pharmacoeconomic analysis of drug therapy. Pharniaco
Economics 1992; 1 : 20-32.
11. Deborah MF, Stoddart GL. Torrance GW. Methods for
economic evaluation of health care. Oxford Medical

Publication. Oxford. 1987.
12. Drummond MF. Jefferson TO. Guidelines for authors and
peer reviewers of economic submission to the BMJ. Br Med J

1996:313 :275-83.
13. Reeder CE. Overview of pharmacoeconomics and
pharmaceutical outcomes evaluations (Review). American

approaches to
of
Clinical

16. Tran TH. Belhell DB. Nguyen TT. et al. Short course of
Ofloxacin for treatment of multi-drug resistant typhoid.
Clinical Infectious Disorders 1995: 20 (4): 917-23.
17. Alam MN. Haq SA. Efficacy of ciprofloxacin in enteric fever
comparison of treatment duration in sensitive and
multidrug-rcsistant Salmonella. American journal of Tropical
Medicine and Hygiene 1995: 53 (3): 306-1 1.
18. Tran TH. Bethel! DB. Short course of ofloxacin for treatment
of multidrug-resistant typhoid. Clinical Infectious Diseases
1995: 20 (4)': 917-23.

announcement
National Asthma Update 99, organised by Department of Medicine, SP Medical College, Bikaner

will be held on 2nd and 3rd January 1999 and followed by
Patient’s Education Programme on 4th January 1999.

Co-sponsors - Medical Council of India and Bikaner Asthma Care Society.
For details contact: Dr. M Sabir. Organising Secretary, Assoc Prof and Head. Resp Div.,

Dept of Medicine, SP Medical College. Mohalla Choongaran, Bikaner 334 005.

Tel.: (R) 0091 151 200663, 528766

Fax: 0091 151 201334

997
’ini ’ nno

Lepr Rev (1996) 67, 208-216

Cost-effective footwear for leprosy control
programmes: a study in rural Ethiopia
GIRMA SEBOKA & PAUL SAUNDERSON
ALERT, P.O. Box 165, Addis Ababa, Ethiopia

Accepted for publication 13 February 1996
Summary A randomized, controlled trial of commercially available canvas
shoes was carried out in a rural area of Ethiopia. Subjects with deformed and
anaesthetic feet, most with ulceration, were given either canvas shoes or
plastazote/moulded shoes and followed up for one year. Seventy-five percent
of subjects with ulcers who used canvas shoes had no ulcer at the end of the
study, while no significant change was noted in the plastazote group. The
durability and acceptability of the shoes were also examined. Clients in remote
areas who have no access to an orthopaedic workshop, but who have anaesthetic
feet, with or without deformity, should have access to canvas shoes with an MCR
insole. Two pairs are needed per year at a cost of USS6-7 per pair.

Introduction

Plantar ulceration is a major complication of leprosy, which can lead to chronic
infection, bone destruction, deformity and eventually amputation, often with prolonged
periods of hospitalization.1
The cause is repeated trauma to a foot rendered anaesthetic by leprosy neuritis,2
while deformity may exacerbate the problem by causing an abnormal distribution of
pressure during normal activities.3
It is well known that immobilization alone, which breaks the cycle of repeated
trauma, leads to healing of most simple ulcers;4,5 this process may be augmented by
simple wound care.1 Many programmes, however, find that it is difficult in the long-term
to assist people affected by leprosy in keeping themselves ulcer-free.6 This may be due to
lack of knowledge in the care of feet, but is more likely to be due to their socioeconomic
status which precludes rest for prolonged or even short periods.
The provision of special footwear can help to overcome this problem by spreading
the pressure more evenly over a wider area of the foot, reducing the trauma to specific
pressure points. The person can then effect a compromise between normal activity and
complete rest, by reduced activity, the use of protective footwear and wound care.7
While the provision of footwear should be an integral part of any leprosy control
programme,8 it may be very difficult for financial and logistic reasons to make this
service available in practice, on a continuing basis.910
208

0305-7518/96/067208 + 09 $1.00

Lepra

Cost-effective footwear for leprosy programmes, Ethiopia

209

Man, ieprns, control progrades ha«
^"XXoXwcared^-made boots. However, there are

major problems with special footwear, including.
;; this means that many do not
Cosmetic: clients are easily identified as leprosy patients.

are

PrSn^adequTte nmbSthis is impossible for most programmes and often only
hospitalized patients are served. In other words, the shoes are not used as a preven
measure, until severe damage has already occurred.
r>
f
5n China11’12 using canvas shoes with an MCR insole showed an

healed ulcer prevention. At ALERT, we have used the

I

I

i

deep enough to take an additional MCR insole (shore 15°, thickness 8 mm) and we have
started to provide these in our control programme. The major advantages are the high
acceptabilky to clients and the possibility of providing adequate numbers of shoes
ShOmsOsiudy addresses certain important questions regarding the provision of footwear

on a routine basis to people affected by leprosy:
1 Given the difficulty in supplying moulded sandals in the required number^would the
provision of canvas shoes to people with deformed feet give acceptable resuite.
2 Can the provision of canvas shoes lead to the healing of eiustmg uiarafion and the
prevention of further ulceration, in people with anaesthetic feet (and even deformed
3 How acceptable and durable are the canvas shoes under field conffitions?
4 How cost-effective are canvas shoes as compared to other methods of managing
plantar ulceration in the long-term?

1
i

Methods

<1

STUDY DESIGN

)

A prospective, randomized1 controlled trial was carried out near Sheshemane, Ethiopia
from November 1994 to November 1995.

£

SUBJECTS

i
7

>1
S

a

210

G. Seboka and P. Saunderson

Randomization was by day of attendance at the clinic. Subjects were examined at the
start and subsequently at 2, 4, 6 and 12 months by one of us (GS), together with one of
two local supervisors. The majority were farmers living in and around the village of
Kuyera.
Health education had been given in the past to these people, but no additional
educational measures were taken during this study.
There were two exclusions from the plastazote group (one was admitted to hospital
and the other refused to attend for follow-up). Results are reported for 68 subjects.
OUTCOME MEASURES

Ulcer size was measured at each visit and the area of ulceration was calculated according
to the following formula: 0-785 x length x width.13’14 At all follow-up visits, the shoes
were examined for wear and tear and the subjects were asked a series of questions
concerning the acceptability of the footwear and how helpful they found the shoes in
assisting with their foot care.
COSTS

The cost of providing both types of footwear was also examined in order to provide a
cost-effectiveness analysis (CEA). The canvas shoes were sold at the wholesale price to
us, namely USS6-7 per pair. The true cost of manufacturing the moulded sandals could
not be ascertained, but is likely to be more than US$20 per pair, the materials alone
costing USS12-7 per pair. Distribution costs were not examined.

Results
SAMPLE CHARACTERISTICS ON ENTRY

Table 1 shoes the sample characteristics at the start of the study.
AREA OF ULCERATION

Figures 1 and 2 show the area of ulceration found at the start and at subsequent follow­
up visits, for the plastazote sandal and canvas shoe groups, respectively.
Three of 28 subjects in the plastazote group never had an ulcer during the period of
study, so 25 are included in Figure 2. Five subjects who were initially ulcer free,
developed ulcers, at least one because the new plastazote shoes did not fit well. Twelve
of 40 subjects in the canvas shoe group never had an ulcer during the period of study, so 28
are included in Figure 1. None who were initially ulcer free developed ulcers.
The geometric mean size of ulcers over time for the two groups is shown in Figure 3,
with confidence intervals for each value. There is a highly significant difference between
the two groups at the end of the study. All ulcers in the canvas shoe group decreased in
size (the majority, of them healing completely), except in two instances. In one case, the
only one in which a large ulcer showed no improvement over the year, a biopsy showed
epidermal hyperplasia (requiring surgical excision); in the other case, a new ulcer
appeared at the one-year follow-up, said to be due to inexpert trimming of dead skin
by the person himself.

Cost-effective footwear for leprosy programmes, Ethiopia

with the ground on standing or walking.

Total analysed
Age <45
>45
M:F
Foot pathology:
Complete loss of protective sensation
Adsorbed toes (1 or more)
Ulceration
superficial ulcers (<5 mm deep)
deep ulcers (range: 5-15 mm deep)
Inversion of both feet
Inversion of one foot
Footdrop without inversion
Adsorbed forefoot: bilateral
unilateral

DURABILITY

211

___________________

Control group
plastazote shoes

Experimental group
canvas shoes

28
16
14
8:20

40
21
19
20:20

28
28
20
7
13
4
8
5
10
7

40
40
28
14
14
2
9
6
4
7

'=z.,E.zr.:.=

s- s

^wXXmp.ed, in conjunction with the Canvas Shoe Factory, to strengthen the
canvas uppers and the initial results are encouraging.
ACCEPTABILITY

Clients were asked how they themselves, their families and. their neighbours likedjthe
shoes and also how suitable the shoes were for their work. These questions were asked at

canvas shoes found them exceUent and more^^0% ported

rtctracsce^

requested a pair of plastazote sandals for use at home and a. pair of canvas shoes for

i

W°Approximately^% of clients using plastazote sandals foundthem excellent, but full

G. Seboka and P. Sounderson

212

1800
1800

14004 \ \

12Oo4\\
□ 12 months

10004 \
Area
of
ulceration
mm.sq.

8Oo4\
■ 8 months

600 4

□ 4 months

i

□ 2 months

200 4

I

□ start

o4
i

2 3 4

5

• 7

» 9 10 n

12 13

" 15

18 " 18

Patients

n[.„ !. Uta
JKSiSSS

ta taa..!» ~

19 20 21
22 23

’ -u

24 25

. • - J •......

— « 2.4. 6 - 12 —4u 4

to

same patients.

• *•
• Was reported by no more than 20% at any time.
acceptance by family and neighbours
sStyfor work was between 30% and 60%.
COST-EFFECTIVENESS ANALYSIS

Canvas shoes
We have shown that 21 (75%) of 28 subjects with deformed feet and chronic ulceration
showed healing of the ulcers during a one-year period of regular use of canvas shoes
The cost of these shoes is approximately USS 6-7 per pair and two pairs
arXiS
X
year. pS comparison MDT for MB paucnrs costs about

"^Z'^lbulcerswerehealedtn^Brstbmon^^

6 months The first 16 ulcers were healed at an average cost of 28 x 6-7/16 - 11 7UM>. ine
XX. 6 were healed at a cost of 12 x 13-4/6 = 26-3 USS. The average cost per ulcer
healed was 28 x 13-4/22 = 17-1USS, over a one-year period.
Cost of ulcer prevention: 12 clients without ulcers but with anaesthetic and deformed
feet, did not develop ulcers during the year; and 15 out of 16 chents with healed ulcers at
6 months remained ulcer-free for the second six months.
Ulcer prevention was therefore attempted for 40 subject/half-years at a a»t of
40 x 6-7 = 268USS. Ulcers were prevented in 39 of these half-year penods. ITie cost per
ulcer prevented was therefore 268/39 = 6-9USS.

i

I

------- --------------------

Cost-effective footwear for leprosy programmes, Ethiopia

213

1800

1600
1400
1200

Area
of
ulceration
mm.sq.

10001\\ ■

□ 12 month#

wo 4
4o°4

■ 6 months

□ 4 months

zotU Vw

□ 2 months
□ start

1 2 3

K. -

< 5

8 7

8 9 ion

12 13

1*15 16 17

Patients

18

19 20 21

22 23

24 25

28 27
97 28

Figure 2. Ulcer size over time for the canvas shoe group. The patients arc placed in order of ulcer size at the
start of the study and numbered 1-28. The area of ulceration for each patient at the start, is shown at the back
of the diagram; moving towards the front, the area of ulceration at 2,4, 6 and 12 months is seen for the same
patients.

Plastazote sandals

Plastazote sandals showed no overall benefit in healing and preventing ulceration. Eight
of the group started without ulcers and 7 were ulcer-free at one year. Only 3 clients
remained ulcer-free throughout.
LABORATORY TESTING OF CANVAS SHOES

Neuropathic plantar ulceration develops over areas of high pressure associated with
deformity or joint limitation. Laboratory methods have been developed to show how
effective different shoes are in reducing peak walking pressures.15 A sample of the
Ethiopian Canvas Shoes (ECS) used in this study has been tested at the University of
Liverpool, UK, and compared with a range of products available in different leprosy
control programmes around the world. Pressure was measured at 10 points on the
plantar surface of the foot during normal walking. Forty-one different shoes, sandals
and insole materials were examined (including the ECS).
Table 2 compares the results for the ECS with the means and ranges for other
samples and the results from walking barefoot. No shoes were consistantly at the low
end of the range across all measurements, but the ECS was one of about 6 pairs to have
consistently below average pressures.

214

G. Seboka and P. Saunderson

geometric mean ulcer size (sq mm)
Shoe Type
1,000

— Plastazote •’Canvas

100

r-.
10

(vertical bars show 95% confidence intervals)
1

initial

2 mo

4 mo

1 yr

6 mo

time
Figure 3. Geometric mean size of ulcers (mm2) over time for 25 Ethiopian subjects with plastazote sandals and
28 subjects with canvas shoes.

Discussion

Previous studies have shown that it is possible to achieve high rates of ulcer-healing
through various techniques, such as: good wound care and immobilization in a centre of
excellence—94% healed;4 podiatric orthoses—57% healed;14 and the use of canvas
shoes in China—84% healed.11 The challenge for control programmes is to achieve and
maintain ulcer-healing on a wide scale at the lowest possible cost and by a method
straightforward enough to be applied through the general health services.
This study was carried out under field conditions in a rural part of Ethiopia, 250 km
from Addis Ababa, where a large number of people affected by leprosy have settled. It
Table 2. The maximal force and peak pressures recorded at various points on the sole of a normal
foot during walking, in the Ethiopian Canvas Shoe (ECS), in 40 other shoes or sole materials
from other leprosy projects and barefoot.
Mean and range for all 41 samples

ECS

mean

low

high

Barefoot

Maximal force (Newtons)
mean of readings for metatarsal heads
mean of readings for all 10 sites
reading for the heel

50
76
394

75
88
394

44
68
290

100
105
461

93
103
472

Peak pressure (N/cm sq)
mean of readings for metatarsal heads
mean of readings for all 10 sites
reading for the heel

16-1
12-8
19-2

16-4
13-3
19-7

9.2
9-5
13

22-1
17-3
27-5

25
20-7
31-5

I
Cost-effective footwear for leprosy programmes, Ethiopia

215 r

was prompted by the awareness that:

• People affected by leprosy have been taught how to carry out self-care, but a large
number (which can only be guessed at) are unable to prevent chronic or recurrent
ulceration of their anaesthetic feet, without appropriate protective footwear.
• The large numbers of former patients with anaesthetic feet (whether deformed or
not) cannot be supplied with special footwear made in orthopaedic workshops. It is
logistically impossible at present.
• Various commercial footwear manufacturers can make shoes which are protective
for anaesthetic feet and are socially acceptable.
The study has shown that commercially produced canvas shoes are beneficial for clients
who have deformed as well as anaesthetic feet. They are a cost-effective method of
achieving ulcer-healing and or preventing new ulcers. Probably the most important
aspect of the canvas shoes is their ready acceptance by both clients and community,
while the specially made plastazote sandals immediately stigmatize the person as a
leprosy case. It appears also that the canvas shoes are preferred for farm work and for
walking on dusty and stoney grounds.
The study was unable to investigate why subjects with plastazote shoes showed no
improvement overall. However, it is our impression that because of poor acceptability
by the families and neighbours of clients, these shoes may not be worn on many
occasions. However appropriate as a technical solution, plastazote sandals and even
open MCR sandals, appear to be socially (and often functionally) unacceptable in
Ethiopia.10 Conditions in different countries must be examined closely; for example,
canvas shoes may not overcome problems such as working in paddy fields and not
wearing shoes in the house, which are issues in India.6
At ALERT, we are trying to move away from the traditional monthly care clinic,
where patients come for soaking, trimming and oiling, but then may do very little else for
the rest of the month. We have recently started a pilot study of community-based selfcare, in which a group of clients living near each other meet weekly to assist each other in
self-care and to discuss problems. A supervisor and foot-care specialist have been
visiting monthly in the initial phase and early results are very promising.
We would therefore advocate a foot-care programme in which self-care is promoted
and commercially available footwear is provided twice a year. In the long-run, most clients
would only see a health worker or supervisor twice a year. People who still have an
apparently simple ulcer after 1 year of using canvas shoes would require further
investigation to discover the reason and may need referral for surgery. There may be
epidermal hyperplasia, as in one of our subjects, or even a malignancy; surgical correction
of deformities and reduction of pressure points may be indicated. This would also be the
most appropriate stage for the provision of special footwear, after discussion between
surgeon, orthotist and technician. It may be that two types of footwear, for use on
different occasions, will be the best solution for some people with deformed feet
While the annual cost of providing footwear is noted to be very similar to the cost of
MDT for MB patients, the provision is not limited to the two years of MDT. Thus there
are many more chents requiring footwear than are registered for MDT and they will
require it for many years. It may be that some manufacturers can produce such shoes for
a lower cost, and part of the cost can be recovered from clients, but this will still be an
expensive programme, requiring further long-term commitment from donors.

G. Seboka and P. Saunderson

216

Acknowledgments
We thank Dr Diana Lockwood, Ms Heather Currie, Ms Catherine Benbow and
Dr Roland Kazen for helpful comments; Dr Peter Byass for the statistical analysis;
Dr Peter Byass for the statistical analysis; Mr Keith Linge for testing the shoes in the
laboratory; Ato Gabre Yesus, Ato Sahelu and the Shashemane workshop for their
cooperation; the ALERT field staff for their continuing care of patients and Wzt Almaz
Mengesha for typing the manuscript.
ALERT is supported by ILEP members through NSL, the Netherlands Leprosy
Relief Association.

References
1 Kazen RO. Management of planter ulcers—theory or practice? Lepr Rev 1993; 64: 188-89.
2 Brand P. Insensitive feet. A practical handbook on foot problems in leprosy. 1981 The Leprosy Mission,
London.
3 Patil KM, Manoj R, Srinath MS, Oommen PK, Babu M, Malaviha GN. Analysis ofstatic and dynamicfoot
pressure in leprosy patients. Abstract RE11.14th International Leprosy Congress, Orlando, Florida. 1993.
4 Birke JA, Novick A, Patout CA, Coleman WC. Healing rates of planter ulcers in leprosy and diabetes. Lepr
Rev 1992; 63: 365-74.
5 Srinivasan H. Do we need trials of agents alleged to improve healing of plantar ulcers? Lepr Rev 1989; 60:
278-82.
6 Krishnamoorthy KV. Protective footwear for leprosy patients with sole sensory loss or ulceration of the
foot. Lepr Rev 1994; 65:400-2.
7 Srinivasan H. Deformities and disabilities—unfinished agenda in leprosy work. Lepr Rev 1995; 66:193-200.
8 Brand M. Pre-congress Workshops: Recommendations on rehabilitation. Lepr Rev 1988; 59: 304-5.
9 Becx-Bleumink M, Berhe D, Mannetje WT. The management of nerve damage in the leprosy control
services. Lepr Rev 1990; 61:1-11.
f
10 Becx-Bleumink M. Priorities for the future and prospects for leprosy control. Int J Lep 1993; 61: 82-101.
11 Xu S, Watson J. Comprehensive protection effects on leprosy sole ulcer and simple sole sensory loss. Abstract
RE58. 14th International Leprosy Congress, Orlando, Florida. 1993.
12 Xu S, Watson J. Development and application ofprotective shoes for leprosy patients. Abstract RE59.14th
International Leprosy Congress, Orlando, Florida. 1993.
13 Kundin JI. A new way to size up a wound. Am J. Nursing 1989; (Feb.) 206-7.
14 Cross H, Sane S, Dey A, Kulkarni VN. The efficacy of podiatric orthoses as an adjunct to the treatment of
planter ulceration in leprosy. Lepr Rev 1995; 66: 144-57.
IS Birke JA, Foto JG, Deepak S, Watson J. Measurement of pressure walking in footwear used in leprosy. Lepr
Rev 1994; 65: 262-71.

--------

7^'

fc,

_HEAL7
healT;j;

v ’ X-A

'

xfe ---- '“A--

+

X
The Cost-effectiveness of Three Smoking Cessation Programs
David G. Altman, PhD, June A. Floha, PhD, Stephen P. Fortmann. MD

___________________ _____

AND John W. Farquhar, MD

Abstract! This study analysed the cost-effectiveness and dlstribulion of costs by program stage of three smoking cessation
programs: I) o smoking cessation class; 2) an Incentive-based quit
smoking contest; and 3) a self-help qdll smoking kit. The self-help
. program hod the lowest total cost, lowest per cent quit rale, lowest
lime requirement for participants, and was the most cost-effective.
fhc most effective program, the smoking cessation class, required
the most lime from participants, had the highest total cost, and was

Inlioduciiuii

■ 5V,“‘:’;,rccUveness analysis (CEA) and cosl-benefil anal­
ysis (CEA) arc used by some public health professionals as
one aspect of evaluation.1-3 Accordingly, one of the goals of
the Stanford Five City Project (FCP) is to conduct costimalysis of Ils healtli interventions. The FCP is a compre­
hensive, long-term, quasi-experimenlal community health
education study designed to reduce heart disease risk, mor­
bidity, and mortality in two intervention cities (total popu*P»^00). The Findings reported here examine the
cosUeHecliveness • of three smoking cessation programs
and are the results from experiences gained during tliejirsl
five years of the FCP. The study does not attempt to quantify
the monetary costs or benefits to society of a person who
quits smoking.
Cost analysis has not been used equally to analyze
different types of health interventions. For example, there are
fewer studies examining Uie cost-effectiveness of prevention
FrPI?rILnis than of treatnienf-orfented programs,2 In particulur, the cost-effectivehess oT smoking prevention/cessation
programs has rareiyljeeh studied? In one of the few CEA
studies of smoking tessation programs, a secondary analysis
of 43 published studies, program costs were estimated on the
basis of’. . .the dollars expended on contacts with smokers,
bused on the duration and number of contacts multiplied by
the national average hourly salary or fees of workers of the
kind used in the contact.’®This melbod of cost estimation
probably underestimates the costs of these programs because
only rough estimates of a single cost category (i.e., person­
nel) were made whereas the lo£al costs include many othercategories (e.g., o-verhead, benefflih volunteers, supplies,
travel, data analysis).
This study examines in detail the cost-effectiveness_or
three smoking cessation programs while addressing some of
the limitations of previous cost analyses. Specifically, a
comprehensive analysis of program costs is included, the
distribution of costs in program stages is analyzed, and
cost-effectiveness is examined ovser the lifespan of a program
and under different quit rale assumptions.

From the Center for Research in Disease Preventiom Stanford University.
Address reprint requests to David G. Allman. PhD. Center for Research in
Disease Prevention. Health Research and Policy Building. Stanford Univer­
sity, Stnnford. CA 94305. This paper, submitted to the Journal January 31,
I9R6. was revised nnd accepted for publication October 10, 1986.
Editor’s Notei See also related article p 153 and related editorial p 140 this issue.
01987 American Journal of Public Health 0090-1)036/87$1.50.

162

the least cost-effective. The smoking contest was in-between the
other two programs in total costs, per cent quit rale, mid cost-cCfeetiveness; it required the same time commitment from participants
as the seir-lielp program. These findings arc interpreted within the
context of community-bused intervention in which the mpumcnl is
made that cost-effectiveness is only one ofscvernl factors that shmild
determine the selection of smoking cessation programs. {Am J Public
Health 1987; 77:162-165.)

Methods
The three interventions selected fur this study wute
community-based smoking cessation proginins developed for
use in the two education communities of the Stanroid Five
City Project.

Smoking Cessation Programs
Smoking Cessation Class—This program was designed
by Stanford Five City Project staff and implemented by it
coupty health department. The classes included eight onehour.session.s., the first five offeredJn consecutive weeks nnd
IheJasMhree offered every othcr week; attendance ranged
participants.iQuittmgtccfihfques included behavioral problem solving, self-monitoring, tapering, deep muscle
relaxation, goal selling, and group social, support., BtHurcQ class, instructors received two to three hours of v/
training and attended a class led by another instructor.
Tnceniive-based Smoking Cessation Contest—'Fins pro­
gram was a six-week communily smoking cessation contest.
Smokers enrolled in the contest provided verification of their
smoking status and then attempted to quit by a p^edclcrm••
day. Following jL_six-week period, a random drawing v.as
lieldTor the grand prize (a trip for two to Hawaii) and 21 other
donalecLprizes. All winners had to verify their non-smoking
status by submitting to a carbon monoxide assessment. One
month Following the drawings, a questionnaire was sent to all
entrants assessing smoking status and habits. A large major­
ity of the participants quit on their own even though self-help
materials and smoking classes were available to them.* The
contest was promoted through lelevision,.radio, newspapers,
posters, schools, .and-word of-mQUtlu.Phone-surveys indicated lfiat60.per.cent of the communily populaliun-was.awurc
of the contest, v
Self-help Quit Smoking Kit—The third program was a
four-step self-help quit kit7’® containing four "lip sheets’’ and
an introductory page. Each lip sheet had two sides, one
providing general information and the other providing lips on
quilting and specific action steps to lake. Tips include use of
subslilulfifi-for smoking, social-support, public cjumniilnicnl,
planning, record keeping, and goal.selling. A heart-shaped
magnet is includeffTn the kit for use iir posting each of the
lipsheels on a refrigerator or other prominent place. ThcJdt
can be distributed through a variety of channels.,..includinglibraries, health agencies, and physician offices.

•Kinp AC. Flora JA. Clark M. er al: Smokers’ challenge: evnluniitin of a
communily smoking cessation contest. Paper presented al the annual rncciing
of the Society of Behavioral Medicine. New Orleans. I9K5.

AJPH February 1907, Vol. 77, No. 2

s

1983. B_v ..using present value anntysis lechniMues.2,,> cost
outlays were compared in 1981 dollars. A discount rale of 5
The seven mnjor categories of costs are: I) stair and stair
percent was applied to cost outlays in each year of a program.
neHts (Stanford, community, and volunteer slnll, and
The sum of these dis^oyjiLcd values represents 1981 dollars
nsukunts): 2) overhead; 3) rent; 4) supplies and materials
needetHrrimpIemenl eacIryfogfamT oveFTls ‘e'xjpeetcdliveriming, advertisements, telephone, postage, prizes, supyeeHrfeypSnr;
ies); 5) trnvcl; 6) data analysis (keypunching, computer
nie); and 7) linuLXeQUilClLpLs.<nokers for participation in
Program Effects
ich program. In addition, program costs were divided into
The primary outcome variable was post-program smoking
tree project stages: 1) development (costs for program
status (smoker or non-smoker) defined as complete abstinence
^velopmenl and planning, lender training, formative refrom smoking at the time of assessment. Pariicjputi.lsjnJlicj:lussJ
eurch); 2) proinolion/implemenlntion (promotion of proquif,at thelatest, by ihe iffilfweeko'fan IJ-weckclass. Smoking/
ium, entry form distribution and collection, program implestatus was ass^gd pn avei^e-six^^mxksjxller the final duje furl
lenlation including the costs for prizes, and community
cessation. Participants in the contest quit, aFlhe latest, by tlnr
rgnnization); and 3) evaluation (pre- and post-test surveys,
fourtluveek of a six-wegkcnnlesl. Smoking status was assessed
ala entry and analysis, and feedback to participants),
four to six weekraftenheTmai'date for cessation. Partictpants
e v o hippie ii I n I rntts-wirv liniil'etkWHh^se di reel I y Telate<f to
in lhe sclf-lieip kit prognun quii/al IheTalest, by the third week
ic smoking cessation programs understudy rather Uian^lo—
of a four-week program. Smoking status was assessed on i
dated icsunrces required Id^eVSfopThe program (e.g., costs
average five weeks after lhe final dale for cessation. Thus^j
>Pthrory development and intervention researgOxLai^F
assessment of smoking status for all three prognms uccurrciLaI(t
ivestignlors). To account for I fie iacllhal many health
aininimutafrDnL4-76_wcck.s and al amaximumJF-_l f
^ofe^rmnih-ttne-intervenVwrts-developed by other people
cessation. In addition, it was assumed that participants who |
alHeHhan'deveibbing*theirownTwe also conducted analyses
dropped out of a program (slopped attending classes, failed to |
inder the~assumption-tirat^there were no developmental
return post-test questionnaire) between pre- and post-test mea­
Q&li. In these" analyses7~briiy promotion, implementation,
surement were still -smoking al post-test. Because only the
nd evaluation costs were included.
contest used biochemical validation of self-reported smoking
Assumptions of the Cost Estimation—EaclL-program
, status, comparisons of lhe threSL-Drograms are based on self­
'as analyzed on the basis of one-year and five-year lifespans,.
report measures. It was assumed that smoking relapse rales
.ie projection of a five-year lifespan is based on experiences
would not differ between programs.
rom the FCP in which the smoking class has been conducted
Quit Rates—The quit rate and number of participants on
or five consecutive years, the smoking contest for three
which th
thee quit rate was based for each program is as follows:
(iijsetmiiye yenrs, and the self-help quit kit has been used for
self-helpf(N = JQU, quit rale 21 percent); contest (N = 498,
ive years in a variety of settings. All of these programs are
quit rate =^22,^£rxenl);-class (N = 541, quit rale ^JJ5_per
still being implemented and it is conceivable that their
cenl)*.Thequit rale for lhe self-help intervention is based on
lifespan would be greater than five years although such
aTampie of 101 people who contacted lhe Stanford commu­
factors as program obsolescence, program saturation due to
nity olfice for materials after they were described in a weekly
a finite number of interested smokers, and general degradanewspaper column. Approximately 46 per cent of lhe house­
lion of intervention quality certainly limit the lifespan of
holds in lhe study area subscribed to the newspaper. During
community programs such as these. Quit rates were project­
the period of program implementation, however, lhe self-help
ed to remnin-lhr. same over the life of the programand it was
materials were actively distributed and widely used in many
assumed that each program would attracX_equal numbers of
other sellings. Thus, it is misleading to consider lhe costs for
people each year of implementation. These assumptions are
developing lhe self-help program materials to be limited to
also based on experiences from the FCP which indicate that
lhe particular community program that attracted 101 partic­
quit rales remain the same or improve slightly each year a
ipants. Therefore, lhe cost-effectiveness ratios for lhe self- ,
program is implemented and recruitment of participants
help program are based on_a more realistic sample size of 5011
decreases slightly until program saturation in the community
in order to reflect the actual community-wide distribution qf
is reached, nt which point recruitment drops significantly.
materials. It is reasonable to assume that in moderale-lo-large \
The classes-required about nine hours of participant time
sized communities, a self-help program would be of interest
while the contest and self-help quit kit each required about 25
to even greater numbers of people.
minutes. The value placed jan each hour-of.participant time
Data Analysis
was assumed Jq he 4X0.
The cost per quilter was estimated by dividing lhe total
After the first year of each program, we assumed that the
cost of each program al one year and five years by lhe number
developmental costs in each subsequent year is 15 per cent of
of people who would be expected to quit. To examine lhe
first year costs. This estimated figure was based on previous
robustness of this CEA, a sensitivity analysis was conducted.
experiences repeating these and other community programs,
The quit rate was changed to account for recidivism after lhe
and it accounts for the minor changes that are made in the
two-monlh quit dale as well as for differences in quit rales
design of a program when it is implemented in different
obtained, either lower or higher, when lhe program is
settings, with different people, and at different times. We
implemented in different settings. Nine. qujj._.rales were
assume, however, that the basic intervention remains un­
examined: 5, 10, 15, 21, 22, 25, 30, 35, and 40 per cent.
changed throughout its lifespan. Evaluation costs in subse­
Results
quent years are also assumed to be 15 per cent of first year
Demographic Data
costs since the programs were evaluated extensively in the
As Table 1 indicates, participants in lhe self-help pro­
first year.
gram (mean age = 45) and class (mean age - 45) were older
The smoking class and self-help prograinsj^ere imple­
than those in lhe contest (mean age=38). There was a greater
mented in 1981 while the smoking contest firslocgurred.
J EMlimillvn

AJPH February 1987, Vol. 77, No. 2

163

?

j-

.tman, et al.
table 1-P.m.,r.phloPI„rll,ul|on or P.rtllll|,.n|, |n 8moWnB c<Bial|ori proBriimj
Program

Class (N - Ml)
Sell-Help (N - 101)

Contest (N - 498)

Mean Age

Mean Education

% Males

% Females

44.8
(13.4)
47.0
(139)
38.2
(12.6)

13.5
(2.9)
not
available
13.6
(2.8)

33.7

68.3

32.7

67.3

44.6

55.4

lhc CPnlesL<45 P" cent) than in either
(he class (34 per cent) or the self-help kit program"(33 ne?
reUes
PartlciPanls also s"'°ked slighll^fewei cigantlP 1° y "l pre'lesl Onean-25) than did participants in
the other two programs (both means=27). in addition data
Uml'n r"8 rnt“ !" l'’e.FCP etlucalion communities indicate
981^82 whiter'? °r 116 poPulalion n8ed 25-74 smoked in
1981-82 While 26.6 per cent smoked in 1983-84. Likewise the
wasU20 3^198°!per day amonB smo^rs
was 20.3 in 1981—82 and 19.1 in 1983-84.* These differences
tn the population and in the smoking rates at the time of the
froi?'lh"s1st'u’dy.’0U d b' n°led in i,,lerpreli"8 Ule ^"83

1 olal Costs by Program Stage
(Hist dd^1 flr8.4ff«r oC.°,S.lS ror eacl1 proBran’ ar=: self-help
($15,144), contest ($25,832), and class ($75,632). If develop­
mental costswe .excluded and only the costs for impKHJnlmg, promoting, and evaluating the programs are considered,
trsl year total costs drop as follows: self-help ($4,698),
con esl ($17 671), and class ($50,383). Development costs
made up a higher proportion of total costs in the self-help
program (69 per cent) than in either the class (33 per cent) or
the contest (31 per cent), reflecting the substantial time
needed to develop self-help materials but the limited time
necessary to implement a self-help program, the reverse of
the smoking class and contest.
Cosl-clTecliveness Ajialyab
The cost-effectiveness ratios (cost per quitter) for the
three programs are presented in Table 2. The relative ranking
of cost-elfectiveness ratios across the three programs under
the different conditions is consistent: the most cost-effective
program is the self-help program, thel^t Toit-effective
progranr »s the class. Depending on the*assumplion$ made
-aboub-program lifespan and costs included the following
upper and lower range of costs per quitter by program were
1 f found’- c,ass ($399 to $235), contest ($236 to $129), and
self-help ($144 to $22).
Sensitivity Ajialysls

Sensitivity analyses were conducted to examine the
impact of different quit rales on the cost-effectiveness of each
program in year one (see Figure 1). Inall programSjlhe cost
per quilter decreases as the per cent quit rale increases.
Comparison of the ratios across the three programs is
informative. The class, even at a 5 per cent higher quit rale
than actually found (i.e., 40 per cent), is approximately as
cost-effective as the smoking contest al a 7 per cent lower quit.
rale than actually found (i.e., 15 per cent) and is less
cosl-effeclive than the self-help program at a 16 per cent
lower quit rale than actually found (i.e., 5 per cent). The point
al which the contest and self-help programs become equally
‘Internal report, Stanford Five City Project, 1986.

Mean Cigarettes
Smoked, Pretest

26.8
(12.3)
27.0
(15.2)
24.9
(12.6)

2400 r=-

I
2200 —
2000 —

I
.G
c5

I

1600 —

I

1600 —

I

1400 —

1200 —

O
1000 —
Q.

In

3

800 —

600 —

400 —r
200 —

0 L-l

liliirliLiIrJrJ rp| rirl J
h

5%

10%

15%

21%

22%

25%

30%

35%

40%

Per Cent Quit Rale
Key (Quit Ralos)

d) SELF-HELP (21%)
ES CONTEST (22%)

na

CLASS (35%)

FIGURE
Ye>r Cosl l’er Quitter under DlUerenl Quit Rate Assumptions
(excluding developmental costs)

cost-effective occurs when the contest achieves a 35-40 per
cent quit rate and the self-help program drops to approxi­
mately a 10 per cent quit rale.
Discussion

I

I

The total costs as well as the'Costs, per qui Iler reported
in the current study are generally higher than those reported
by~-Green and.. JohnsonJlJThls Is due, in part, to their
incomplete estimation of program costs, which is understand­
able given that their study was an archival analysis of
published smoking cessation studies.
Our findings are apparently robust within a number of
varied assumptions. Moreover, if the number of participants
recruited to smoking cessation programs, particularly selfhelp programs, is greater lhan the numbers assumed in the
present analysis, the cost per quitl9X.dlQps-due^lD econoiuics
oLscale.
Because it is. unusual for health professionals or organi­
zations to have resources available lo develop all.of their owiv
progranrs, ratios with and without developmental costs were
reported. In many, if not most, instances, previously devel­
oped programs or program components are adopted. Health
professionals should carefully consider Ihis issue since de­
velopmental costs can be substantial. Possible reasons fur
developing new programs include unavailability of interven­
tions for a specific population, unacceptable effectiveness of

164

AJPH February 1987, Vol. 77, No. 2

(

/'n

gw/
,1.

COST EFFECTIVENESS OF SMOKING CESSATION PROGRAMS

'S

TABLE 2—Coat-elfectlveneas Ratios In Three Smoking Cessation Programs

Total Costs Excluding
Developmental Costs

Total Costs

Smoking Cassation Programs
Class
• Total cost
Number ol participants
N ol quilters (al 35% quit rate)
Cosbenecllveness ratio*
Contest
Total cost
Number ol participants
N ol quitters (al 22% quit rale)
CosFellecllveness ratio*
Sell-Help
Total cost
Number ol participants
N ol quilters (at 21% quit rata)
Cosl-ellecllveness Ratio*

One Year

Five Years

One Year

Five Years

$75,632
541
169
$399

$261,589
2,705
947
$276

$50,383
541
189
$266

$222,911
2,705
947
$235

$25,632
498
110
$236

$82,925
2,490
548
$151

$17,671
498
110
$161

$70,423
2,490
548
$129

$15,144
500
105
$144

$26,190
2,500
525
$50

$4,698
500
105
$45

$11,498
2,500
525
$22

'In 1881 dollar a. r ■ 5%.

previously developed interventions, or an interest in evalu­
other more visible and belter understood programs (c.g., a
ating a new technology of intervention.
smoking cessation class) are offered. Combining cessation
The findings from this study must be interpreted within
strategies across programs (incorporating seff-hclp materials
a broad community context.3,9 in a given community, there
into a smoking contest) may further increase the effective­
are multiple demographic groups with different preferences
ness. Findings from the current study may help professionals
for health services, and diverse smoking histories. Different
determine the most effective approaches to reducing cigarette
smoking cessation programs may attract different types of
smoking in free-living community populations.
people and preferences for cessation programs may shift over
time. In this study, for example, the data indicate that the
ACKNOWLEDGMENTS
contest attracted a slightly younger population and a higher
. ..T,‘!c rc«nrch reported here was supported by a Public Health Service
percentage of males than did either the self-help program or
institutional National Research Service Award (5T32 HLO7O34) from the
ihc ciass FuUjr^catduhoul^Ue.^
National Heart, Lung, and Blood institute to Dr. Allmnn and Public Health
O” l.he uiffrrpntiaUillraetftMfef cessation-programs for pop­
Service grant N1H HL2I906 to pr. Farquhar. A version of this paper was
presented al the annual meeting of the American Public Health Association.
ulation subgroups. Moreover, the target population for com­
Washington, DG, November. 1985. The authors thank Anthony J. Evans.
munity smoking cessation programs may vary over time due
Christine Jackson, Nathan Maccoby, and Abby C. King for their comments on
to changes in community social norms and health services
.
iflnd Pe,cr Bartok for hc,P in library research. Special thanks to
utilization. For these reasons, the absolute cost-effectiveness
Judith J. Altman Cor her help with data analysis.
should not be the only evaluative criterion employed. It may
be important for a community to offer a range of integrated
REFERENCES
1. Banta HD, Luce BR: Assessing the cost-effectiveness of prevention. J
smoking cessation programs in order to meet the needs of
Community Health 1983; 9:145-164.
diverse groups of citizens. In short, if the goal is to reduce
2. Warper KE, Luce BR: Cost-BenelU and Cost-Effectiveness Analyses in
smoking in the community-at-large, it makes little sense to
Health Care: Principles, Practice, and Potential. Ann Arbor: Health
limit the smoking programs offered to only the most effective
Administration Press. 1982.
3. Allman DO: A framework for evaluating community-based heart disease
or cost-effective if in fact they only attract a small fraction of
\ preventioni programs. Soc Scl Med 1986: 22t479-4ff7~^
the population in need. Moreover, if the cost of smoking Zk
SP, Maccoby N, tt at: The Stanford Five City
H-oject: design and methods. Am J Epidemiol 1985; 122:323-334.
cessation programs are below their value, the cost-effective­
5. Levin HM: Cost-Effectiveness: A Primer. Beverly Hills: Sage Publications,
ness of a specific program may not be as important because
1983.
a higher cost-effectiveness ratio implies only that the cost for
6. Green LW, Johnson KW: Health education and health promotion. In:
Mechanic D (ed): Handbook of Health, Health Care, and die Health
achieving a given effect is higher in one program than in
Professions. New York: Free Press, 1983.
another. Thus, a higher cost-effectiveness ratio does not
7. Sallis JF, Flora JA, Fortmann SP, Taylor CD, Maccoby N: Mediated
necessarily imply that a program is undesirable.
smoking cessation programs in the Stanford Five-City Project. Addict
Behav 1985; 10:441-443.
Similarly, there may be a need at the community level for
8. Sallis JF, Hill RD, Taylor CB. Flora JA, Killen JD, Telch MJ. Girard J,
sequencing or combining intervention strategies. Because of
Maccoby N: A controlled evaluation of a self-help smoking cessation
program. Am J Prev Med.
the unique nature of some cessation programs (e.g., a
9. Allman DG, King AC: Approaches to compliance in primary prevention. J
smoking contest), the community may not fully accept it until
Compliance Health Care 1986; 1:55-73.

AJPH February 1987, Vol. 77, No. 2

165

© Oxford University Press 1993

HEALTH POLICY AND PLANNING; 8(2): 136-142

Cost-effectiveness of ambulatory surgery in Cali,

Coiombia

« XRr« USA
USA

wn* B. SM. «.

ee _x _ cvetpm of Simplified ambulatory surgery against traditional in-

HSSSS t—-« ~ --s™VIX
sooner man the hosp.ta pat.en

ESS

<34 versus 52 d^y^. the hospital - a four-fold difference. Intermediate

“■

“"’“v ” “

•od

-4 *

-it.

r-J

.yiifl' -sb ;?u;;L

Introduction
Among 29 developing countries for wh,c^0^sistant data were available, an average of 57 ,
public recurrent health expend'tures^sP'g
on hospital care. Colombia, according t
study, used 67^o of these funds on hospita
All curative services typically consume 8O^o of
health resources in developing countries.- Faced
wi”’ shrinking public sector budgets and m;andL .• to expand preventive and promo.ive health
services, policy makers are always searching
ways to deliver curative services more efficientb •

To enhance efficiency, the types and sophistica­
tion of curative services have to be arefuUy
matched to patient need. In Cali, C°'0,™a
third largest city with 1.5 million mhabitan ,
previous studies of surgical services have shown
that reorganizations offered substannal opportunities for greater efficiency,
operating rooms, one of the ^ost. expensi
hospital resources, had a mean utilization of only
42^o in 1974; other expensive resources were also
used inappropriately. In the university hospita^
the tertiary hospital for the city and surrounding
region, 69°7o of the surgical operations were m

the two least complex levels, whereas only 2.^0
were in the most complex category.In response to these data, university and governmem officials in Cali developed a new type of
health facility in 1983, an mtennediate health
unit (IHU) or ‘health centre-hospital , and in’
stituted an innovative surgical technology called
‘simplified ambulatory surgery’, which is prac­
tised primarily in IHUs.-” Primary care facilities
(health posts and centres) refer most surreal,
and obstetric and selected medical cases to IHLs,
which in turn refer cases they cannot treat to
secondary and tertiary hospitals.’

Simplified ambulatory surgery was designed to
diminish the costs of low to intermediate-risk
surgical procedures without decreasing the
quality of care. This surgery provides pre­
operative patient education, early ambulation
and discharge (generally on the day of surgery),
and family home care. The operating rooms are
furnished only with equipment needed for a
carefully-defined set of uncomplicated surgical
procedures. Few medical personnel are assigned
to each operation. Furthermore, operating

...............

-A-.T-’.’

t 137

j

hospital, due to previously established patterns
of referral or utilization.

j

Ambulatory surgery in Colombia

rooms are equipped with tv/o operating tables,
which could permit two surgical procedures to be
performed simultaneously in the same operat­
ing room, under the supervision of a single
anaesthetist. A 1975 study predicted that the
system would lower costs by 75(7o, compared to
' traditional care in the university hospital with a
3-day hospitalization. It would also be more effi­
cient than standard ambulatory surgery, in which
the surgery uses the same facilities and personnel
as traditional inpatient surgery. A leading public
health journal praised the system’s potential to
use existing resources more efficiently.14 This
study is a controlled evaluation of the costs and
effectiveness of simplified ambulatory surgery as
it is routinely practiced in Cali, a decade after its
implementation.

Methods
Study setting
When this study began in 1989, Cali’s public
health system had three secondary hospitals and
four IHUs. The study sites were Hospital San
Juan de Dios (SJD), a 127-bed secondary-level
hospital, and Centro Hospital Joaquin Paz
Borrero (JPB), an IHU with 20 beds. Both
facilities were representative examples of their
type of institution.15 In 1988, the IHU performed
approximately 1100 operations of all types, while
the hospital performed 3500. Both facilities
covered their operating costs through a combina­
tion of patient fees and government subsidy.

Selection of procedure and patients
Inguinal herniorrhaphy was selected because of
the frequency with which it is performed at both
facilities, its moderate degree of technical com­
plexity, and the existence of standardized indices
of surgical risk at both facilities.16-17 Eight per
cent of each facility’s operations were hernior­
rhaphies. This study included all patients aged 16
years or older with a low surgical risk,16-17 who
received an elective inguinal herniorrhaphy
(which was not for a recurrent or incarcerated
hernia) in the three months from January to mid­
April, 1989. Thirty-two patients met these
criteria: 17 at the hospital and 15 at the IHU. In
theory, all uncomplicated cases should be treated
as the IHU, and only complicated cases referred
to the secondary hospital. In practice, un­
complicated cases were also treated at the

r

AW

AM

We compared the two groups of patients on
demographic and medical characteristics (sex,
age, per capita income, education, underlying
pathologies, previous hospitalization, household
size, and preoperative pain expectation). The
auxiliary nurse who admitted the patients interviewed all available patients on the day of admission (the day before surgery at the hospital,
and the day of surgery at the IHU) about these
characteristics.

i

j

'
!*
K

Measures of effectiveness
Using indicators similar to those in previous ?
evaluations of surgery,18"23 we compared//fee- 1
tiveness of the two settings based on con^ica- '
tion rates, patient satisfaction, and duration of T
postoperative disability. Trained Colombian f
health professionals assessed complication rates ’
until the eighth post-operative day through ;
systematic abstraction of patients’ medical
records. We assessed- several components of b
patient satisfaction through a second and third ’
survey. The second survey was generally con­ t
ducted at the facility in which the surgery had I
taken place, one week after surgery, when the
patient returned for his surgical follow-up visit.
If the patient did not report for his follow-up
.................
visit,
however, we tried to interview the patient at I
home. At this stage, we also asked patients
about their out-of-pocket costs.
• The third interview took place at the patient’s
home in August, 1989, four to seven months
after the surgery. The second and third inte^ews
were conducted by Colopibian medical sta^yhts
who were not employed by either study facility.
We assessed the period of convalescence at the
second and third survey - the patient’s expecta­
tion at one week after surgery, and the actual ex­
perience four to seven months later.
Average cost per herniorrhaphy
To fully capture health system costs related to
the site of care, we measured each patient’s
health care costs, from the patient’s admission to
the facility for surgery untiL the eighth post­
operative day. We excluded costs of diagnostic
tests and examinations, which were done prior
to admission in both settings. We assessed direct
service costs in each department of each facility

f
r-

i

i

i

i
I

k•

Donald S Shepard et al

138

that provided care and the associated overhead
costs (such as administration, cleaning, and
building maintenance). We determined direct
service costs in the operating and recovery rooms
for both facilities, and also in the surgical ward
for the hospital.
Direct service costs consisted of medical supplies
(sutures), general supplies (disinfectant, gauze),
which could not be readily assigned to an in­
dividual procedure, personnel, and the finance
for medical equipment. Costs of medical supplies
were obtained by multiplying the utilization of
each item for the surgeries in the study (obtained
fr . abstraction forms completed by nurses and
technicians in each facility, and direct observa­
tion) multiplied by their unit costs. Costs oi
eeneral supplies, personnel, and depreciation
were allocated to a single hernia repair, based on
the ratio of the average time for a single hernior­
rhaphy (estimated as the mean for study patients
in each facility), compared to the total time that
the operating room was in use for all procedures
during the year (derived from each facility s
ooerating room log-books). Financial records of
each institution provided financial data. All
monetary amounts were valued in July 1988
prices converted to US dollars at the then exchange rate of 307 Columbian pesos (COP) per
dollar.

Results

ient characteristics and ciinicaf outcomes
Although patients at the hospital tended to be
older, there were no statistically significant ditferences between the means of the two groups on
any of the preoperative patient characteristics exO. d“C' PrOb“,Uity


eaS

°

The study provided no evidence of inferior
quality at the IHU. The inpatient group actually
had higher rates of both surgical and anaesthetic
complications. Among the 17 hospital patients,
three had surgical complications (two w°una
infections and one haematoma) and four had
anaesthesia complications (two failed anaes­
thesia, one intraoperative hypotension, one per­
foration of duramater). No complications were
reported among the fifteen IHU patients.

Table 1.

Characteristics of study patients by facility

Variables*

Hospital
(N = 13)*

IHU
(N = 15)

Gender
Male
Female

69<7o
31<7o

66%
34%

Age (years)

57.4 ± 6.1

45.0 ± 4.8

Years of education

4.4 ± 0.7

4.6 ± 0.7

Household size

5.1 ± 0.7

4.7 ± 0.7

Total monthly house­
hold expenditures:
72843 ± 14785
Colombian pesos
237 ± 48
US dollars

77056 ± 12634
251 ± 41

•Statistical significance of differences between the two
groups revealed P >0.10 for ail variables shown. Means _
standard error of the mean.
* Data unavailable for four hospital patients who were not
interviewed.

Patient satisfaction
Patients treated at the IHU tended to be more
satisfied than those at the hospital at both
assessments (Table 2). At the eighth post
operative day, patients at the IHU expected to
spend less time away from work than those
treated in the hospital. The follow-up survey
confirmed that IHU patients had, indeed, re­
turned to work sooner. Because of the small
sample size, none of the differences are
statistically significant, although they con­
sistently favour the IHU. No complications re­
quiring rehospitalization were reported among
the 26 patients followed post-operatively from
both groups. Conservatively, these results in­
dicate that the IHU outcomes were at least as
g00(j as those in the hospital. -----Penonnel

While the total number of personnel present dur­
ing the hernia operations at the two types of
facilities was comparable (5.5 in hospital and 6.1
in IHU), the mix differed. The hospital had
almost twice as many physicians including
residents for surgery and anaesthesia, as the IHU
(3.5 versus 2.0), but only half as many nurses and
technicians (2.1 versus 4.1). Thus the IHU relied
more heavily on less costly personnel. The IHU
had almost twice as many nurses and technicians
as the hospital.

k-i

.c:

r

r
<

t

I

139

Ambulatory surgery in Colombia

Table Z

Postoperative results
IHU
(N =

Hospital
(N = 17)

Variable

Zir:e.siLs
Hero—ZHU)
>

At eighth postoperatative day

Perceived stay was too long
Satisfied with appearance of scar
Would have preferred alternative type of facility
Postoperative perception of severity of pain*,**
Expected number of days from operation to
resume usual activity**

15%
46%

80r»

3.1 ± 0.8

2.8 x 0 “

52 ± 14

35 = 9

At 4-7 months*
56 ± 7
49 = 6
Days after surgery to return to regular activity**
^ = 23'
23 = II
53 ± 19
Days after return until all limitations ended**
72= :x
Sa =
108 ± 23
Total days with any limitation**
Patient satisfied with length of hospitalization** 64% ± 15% 77% = 12.% —13% = L9«r*

*On a scale from 1-5 of increasing severity
**Mean ± Standard Error of Mean
* 5 hospital and 1 IHU study patients had moved from Cali region or were cczervise zet
for interview
* * Borderline statistical significance (P = 0.12 with Yates continuity ccrrernnt
•Difference not statistically significant.

L’

£
Average cost per herniorrhaphy
The average cost of an inguinal herniorrhaphy
was US$39.12 (12010 COP) at the IHU and
US$148.76 (45 668 COP) at the hospital - almost
a four-fold difference. When the cost of hospital­
ization was excluded, the cost of the procedure is
three times greater at the hospital than at the
IHU. The largest differences occur in three
categories: overhead costs, personnel costs, and
surgical ward costs (Table 3). The average length
of the operations (from incision to closure) per­
formed at the hospital was 37 minutes compared
to 26 minutes at the IHU. Because computations
of costs in both facilities involved allocating
shares of total operating room and surgical ward
costs to sample patients, standard statistical tests
on cost differences would not be meaningful.

Table 3. Medical care
type of facility

zer nzumai asriorrhaphy by

k


IHU

'-S3

Cost category

Operating and recovery r^ems
Direct service costs:
Personnel
St*
General supplies
Medical supplies
Equipment deprecanoc
Overhead costs
Subtotal
ur:8

3-2

■ .

Surgical ward



USS

r9

49.4
19.31
3.59
1.5
2.21
C.0.4
4.07
•X- 12.93
33.1
“x -9.12 100.0

tI

-it*-

100

0.0

39.12

100.0

r

Patients’ costs

Total

The facility’s charge to the patient for the
operation and follow-up care was US$38.96
(12000 COP) in the hospital and US$16.23 (5000
COP) in the IHU. In addition, patients had to
pay for some of the drugs and supplies, raising
their total medical charges to US$44.07 and
US$18.77, respectively. Thus, patient charges
represented 30 and 48of costs, respectively.
Patients’ transportation expenses were also

greater in the hospital (USS2.J61 :zan the IHU
(US$1.38), so out-of-Tcckic ccsts^ere 2.5 times
higher in the hospitalyrx :han for the
IHU (US$20.15). For
patient in each
setting, these out-of-rocx-r □zszs represented
21^o of monthly
expenditures for
hospital patients, but zuiy •’’r zorZHU patients.

*

J

I

ri

f

r;

I.
!

.1

Donald S Shepard et al

140
A worker earning the minimum wage of USS3
per day would have to work 16 days to pay for
the heavily subsidized hospital operation, com­
pared to only 7 days for the less subsidized IHU
procedure.

Discussion
This study has found that an eier.ive inguinal
herniorrhaphy costs 74% less at an .HU than at a
secondary level hospital. The IH s .ower cost is
due to both its shorter time for the surgery
(probably due to its standardizes protocols), and
its lower cost-per-minute. The average direct
service cost-per-minute o: operating room time
USS1.99 (610 COP) at the IHU compared to
uSS3.14 (965 COP) at the hcspral. Both the
savings in operating room costs and the overall
proportional savings from simpLfien ambulatory
surgery are consistent with ths previous!)
mentioned predictions from 19'5.8 Shorter
hospitalization and devaluation have also
lowered the cost of inpatient stzrgery compared
to 1975.

The generalizability of the climca. outcomes is
somewhat limited by the modest sample size in
each facility, the absence of randomization, and
the inclusion of only one facility of each type.
These design characteristics coiud not, however,
change the direction of our findings. Although
eligibility criteria for the study prcduced gener­
ally comparable samples of parents in the two
settings, hospital patients tencec to be older
(though not significantly so) anc might have had
note undocumented risk factors ior complica­
tions. Even if adjustment for such characteristics
could have explained ail, or most, of the com­
plications in the hospitalized patients, it would _
only equalize the risk-adjusted complication
rates between the two settings. .As no complica­
tions were observed in the IHU. its estimated
complication rate would always remain the
lowest possible value, zero.

Sensitivity analyses showed that neither sample
variation nor the absence of randomization
could explain the lower cost of the IHU. To
examine the impact of sample variation, we ob­
tained independent estimates of the duration of a
hernia repair from separate samples of 19 con­
secutive herniorrhaphies taken from operating
room log-books during three to four randomly

chosen weeks at each institution. The mean time
was 43 minutes for hospital surgeries versus 34
minutes taken for IHU operations. As com­
plicated hernias were not excluded from these
samples, the times were longer than those of
study patients. Recalculating costs using these
times, the hernia repair still cost IWo less in the
IHU than in the hospital.
To estimate the largest possible bias that the
absence of randomization would have intro­
duced, we assumed that in a randomized study,
hospital patients would have spent as little time
as IHU patients in the operating room (26
minutes) and in postoperative ward care (0 days).
Even under these implausible assumptions, an
operation in the IHU would remain 525Fo less
costly than in the hospital. The IHU retains its
cost advantage largely because its cost-perminute for the operating room (both direct ser­
vice and overhead) is considerably lower.

Previous studies have reached conflicting conclu­
sions regarding the relative costs of ambulatory
and traditional surgery. A randomized clinical
trial found ambulatory surgery to be costeffective for certain types of surgery.24 Most
studies found that when surgery is performed at
a hospital and the. patient is hospitalized, the
costs exceed similar procedures performed on an
outpatient basis, because in the latter case, the
cost of hospitalization is saved.22’31 Studies which
compare the cost of performing ambulatory
sureery at a hospital with the cost at a free­
standing clinic, however, have not always found
savings.32’33

With increased utilization, surgery in both an
IHU and a secondary hospital could become
more efficient. Because virtually all operating
and recovery room expenses except those for
supplies are fixed,.a higher volume of services
will lower the average cost per procedure. In
1988, the occupancy rates of Cali’s surgical ser­
vices were 56% in IHUs and 60% in secondary
hospitals.34 If both occupancy rates were raised
to 90%, the average cost of an elective hernior­
rhaphy would drop to US525.4O in the IHU and
USS111.67 in the secondary hospital.
*
While an IHU might have.been expected to ex­
hibit lower quality than ai more sophisticated
hospital, the complication rates demonstrate the

•**

/



5
-i
r

I
£
i.

‘i
>•
•»

to

I
F

v”'

r

i

V

I

/

•A

■r

. "V*

r-’

Ambulatory surgery in Colombia

opposite, probably a result of the systematic pro­
tocols for simplified ambulatory surgery. The

complication rate in 1431 consecutive ambu­
latory surgeries through 1988 (Wooley, u
published data). Assuming this same rate aPP
to hernia repair in JPB, 0.3 comphcauons would
have been expected among the 15 ^ernl°
rhaphies studied: in fact, zero were observed
The IHU’s advantage in patient satisfaction adds
to findings in other populations showing one-day
surgery to be comparable to traditional surg y
in quality, and acceptable to consumers.

From a social viewpoint, ambulatory surgery at
the IHU entailed considerable advantages over
inoatient hospital surgery. An average IH
patient saved US$3.68 in transport costs, gained
7 davs of his time (through earlier resumption ot
his usual activities), and reduced the time and ex­
penditure his family and friends spent visiting the
hospital and assisting in the patient s care. These
are slightly offset by the one extra day a family
member was estimated to spend caring for an
ambulatory surgery patient at home (approxi­
mated by the length of stay ot the bospitatod
patients). On balance, the patient and his family
gain at least 6 days from ambulatory s“r8eryMost importantly, the easier access to. and lower
charges at IHU’s, may encourage patients to
have hernias and other health problems diag­
nosed and treated more promptly.

IHUs) in appropriate locations.

Conclusion
Reflecting its commitment to IHUs. Cali
officials opened a fifth IHU after this stu y
beaan. and plan to open a sixth one this- year
Building on Cali’s experience, Mexico has started
to establish similar units. A WHO consultation
which reviewed this experience--* recommended

r
i

In conclusion, uncomplicated hernia operations
in intermediate health units cost only a Quarter of
what they would in a traditional hospital. Out­
comes, complication rates, patient satisfaction
and return to work were comparable, if not
IHU,. Call’, IHU, provide =
organized around similar principles fo other
surgical services and obstetrics with, presumably
comparable results.

References
Barnum H and Kutzin J. In press. Public hospitals in

i

r
countries. Washington DC. World Bank.
1 de Ferranti D. 1985. Paying for
ing countries, an overview. Washtngton DC. World

* VeleTcA et al, 1980. El perftl de la cirugiai y los cirujanos
en d Valle dd Cauca. Acta Medico del Valle 11. 64- J.
Vdez GA. Galarza MT, Guerrero R et al. 1983. Surgeons
and operating rooms: underutilized resources. Atnencan
Journal of Public Health 73: 1361-5.
6

5
i
i

r
i

I

system for surgical operations: the econom.cs ot
9

h’ospnti orThome? In^°o^nal of Health

Iv■i-

r-

Services 1’. 101-10.
• _
If
Guerrero R. Salazar C, Galarza MT and Velez A 1988.
__ care
Increased efficiency in a regionalized surgical*
Mimeo^'.ili.
I ■
model in the city of Cali. Colombia.......

J
Colombia. Carvajal Foundation.
/!
II Lopez MF. 1983. Casto de insumos del programa de
curugia simplificada en el Centro Hospittd Joaquin Paz i-'
Borrero’. Cali. Colombia. Universidad del Valle.
12 Amu J. 1988. Actualizacidn de costos en el bloque de
cirugia en el ultimo trimestre de 1987: evaluacion
de la eficacia en lerminos de cahdad humana y oahdad
tecnica para usuarios del serviao de cirugia de
diciembre de 1987 a enerro de 1988. Cali. Columbia, »
universidad del VaUe^
de
13 Auuddo CA. Costa F and Kevcio w. 1^03.

\esti6n del programa de cirugia simplificada del ceniro
■Jocpuin Paz Borrero'. Call. Columb.a,
r
Universidad dd Valle.
wrtrirf
» Yankauer A. 1983. Lessons in surgery for the third world.
^can/oM^
American Journal of Public Health 73: 1359-60is De Movaes Novaes H. 1990. Accidnesintegrados en los
sistemas de salud. Cuademo Ticnico
Ticmco No. 31. F

10

Simplified ambulatory surgery and other IHU
functions can be adapted to various organiza­
tional and physical settings. Although surgeons
in Joaquin Paz Borrero receive salaries trom the
local government, in another IHU. the surgeon i
paid fee-for-service directly by patients. While
the newer IHUs were specially built, older ones
were upgraded from health centres.

141

f •

i
i

CLINICAL. ECONOMICS,

MODULE 7_
•**

COST-BENEFIT ANALYSIS

you should understand:
this
module,
After completing
benefit analysis and the
--j
between
costthe difference
modules;
a.
in <earlier
-techniques introduced
-—
necessary;
are and why they are
shadow
prices
what
b.
financial, economic and social
between
the difference
c.
analyses.

1

COST-BENEFIT ANALYSIS
INTRODUCTION
The analytical techniques introduced in Modules 4 to 6 were
designed to compare alternative ways of using resources where
the aim is to ensure that the resources produce the greatest
The techniques cannot be used to determine
possible benefits.
if a single project is worthwhile, This is because costs and
costs in terms of
benefits are measured in di f f erent units:
money, and benef i ts in terms of physical health outcomes,
sometimes adjusted f or changes in the quality of life. This
makes it di f f icult to decide if the benefits of a programme
justify the costs.

In <common with these techniques, cost-benefit analysis (^BA) can
be
used to compare alternative interventions
interventions but -It also
provides
a
means for deciding if a single^
single programme Ts
worthwhile.
It (does this by requiring that benefits and costs
both be measured in terms
ofmoney..
L
The
relationship between CBA and the other techniques is
in
Figure
1.
Cost-minimisation,
costrepresented
ef fectiveness, cost-utility and cost-benefit analyses all relate
benefits to costs, and costs are measured in the same way in all
cases.
However, the techniques differ in their methods of
Cost-minimisation and cost-effectiveness
measuring benefits.
analysis are based on physical efTects, cost-utility analysis on
utilities, and cost-benefit analysis on economic benefits.

RESOURCES
CONSUMED
>-

*

HEALTii
CARE
PROGRAMME

COSTS
Direct costs
Indirect costs
(production losses)
Intangible costs

HEALTH IMPROVEMENT
>

EFFECTS
Health
effects:
natural
units

UTILITIES
Health effects
in QALYs

(Source: Drummond et al, p2)

BENEFITS
Economic
benef its:
Direct,
Indirect,
Intangible

2
MEASURING BENEFITS

The economic benefit of an intervention is the reduction in the
The costs
cost of iliness that results from the intervention.
divided
into
direct,
indirect
and~ihVangT5Te
were
of illness
so the benefits of arT^hteT^veifFion are
costs
in Module 2,
typicaTTy’“categorised in the same way.' The method of estrilffSTing
the ihdTrecF costs of illness described in Module 2, and hence
the indirect benefits of an intervention, is known as the human
capital approach.

DECISION RULES
in the future must be
Costs and benefits expected to accrue
3).
The present value of the costs
(see Module
discounted
(see
divided by the present value of the benefits is called the
If i t is less than 1, the pro ject ^s
cost-benefit
(CB)
ratio.
v i abLe--QX^^wQ_r-thbwhile

NP'V

alternative
indicator of viability is the net present worth
(NPV)
of the intervention.
The^NPV is
net present value
the present value of the benetLts^jniJiu.s^he present
.s. ^r^defined as
interve
ntion
costs;
The J-"---- is
•- viable if the NPV
of the
value
that discounted benefits exceed
____
?whTch
imp
lies
exceeds
0,
Tf
'" the NPV is positive, then necessarily the
d i scourrted~ costs.
CB Tatio wTtl” be less than 1,
1 , so either indicator can be used
However, if you wish
when assessing an individual intervention,
there are technical
rank
alternative
p
roje
cts
using
CB
A
to
reasons why“you“should use the NPV•
~

An

. *
- J''’°

i.

or

c

>1

As with the techniques discussed in earlier modules, it is
to specify whose viewpoint is being considered in CBA.
crucial
the viewpoint of a health care
common^ y takes
The literature commonly
provider,
for
hospital or a government, whose costs
f or example a
are the money costs of providing the care.
In this case there
is no need to^cpnsider either the costs borne by the patient, or
whether tKe prices paid reflect opportunity costs.
Benefits are
simply the Increase in revenue resultfrig from The intervention.

a
a

2.

— </€ N

/

WHOSE VIEWPOINT?

This approach is not appropriate if society’s viewpoint is being
considered
in which case all
costs and benefits affecting
=s
society must be taken into account.
Moreover, the prices used
to value costs and benefits will often differ from those used
when considering the viewpoint of the provider.

(J

SHADOW PRICES
You wiH recall from Module 3 that the cost to_ society of using
a resource is its opportunity cost\
Economists believe that the
best
indicator of benefit is society’s willingness to pay for a
product or service.
In many cases, market prices do riot reflect

I

3
either economic costs
(willingness to pay).

economic benefits
(opportunity costs) or

be affected by taxes
market prices can —But-there are other
because
This
is partly
3)7
transfer payments (see” Module
human
capital approach the
and other
in
the
1.
For
example,
increases
in
reasons as well.
life are the
of saving a
that
this
benefits
indirect
It is often assumed would earn
result.
which
production
to the wages the person
There are many
increased production is equal
life.
of their working
contributions
to
remainder
1
reflect
over the
not
wages do
is
a
good
why
market
reasons
unpaid housewives
and
the case of
productiont
illustration.
/
,
analysis
from
society
s
in order to conduct the
Accordingly >
values for cost' and benefits
economists must
impute
than financial
financial values.
This
viewpoint,
economic
rather
which were introduce^ in
which reflect
the techniques
These imputed values are called
applies to all
as to CBA.
modules as well
shadow prices.
,.j shadow prices have been
Two major methodologies for calculating T and require a fairly
plicated to
-- **pp^y
developed.
Both are complicated
the help
Both
,
You should
thorough knowledge
of
ec
°"
o
^
u
^
e
°.
ry
,
knowledge
shadow prices.
of an economist if you wish to use --I. > Janalysis of
xtensively in the
To date
they
nav.
b
^"
us
^
/
has
date
they have not
be because the literature
-—
economists
interventions
.
health
1?te!‘^ntJJ°nS^orth American examples,, and many
dominated by I—
economic and financial prices
that the <.
_
If this
is true “the"eTfort of
is not
__
larie

Ln

_
the

not largenot be justified.
caf^UTating shadow prices may
North America, it is not
Regardless
of
r J^f1 the'world"
In any case, the effort
true
in much of the
o
not great in many countries.
of ascertaining ®had° J pd from publications; by
___ internaj^ionaj.
Planning
an
d
the P
lanning
They can __of ten _b? .Hank
BSHlT,
and__the
regularly
organizations
such—riovaloning countries
pf
many
developing
Off ice/Departments_
caTcurate_them_fpr thei r _own purposes^

Sevrt^V^^

SOCIAL ANALYSIS.

is called financial
us
i
ng_
_
marke-t

prA.ce
s
called economic.
appraisal
Project
shadow prices are us^d’ it is , become popular
When
analysis.
third form.
social
anal)sis»
form,
-A
analysis *
recently.
It seeks to
a form of efficiency analysis.
How
analysis
is
society from a particular investment.,
Economic
SoQial
the gain to
maximise
,
within
."'aKsFSliSffiSP
gain
is
distributed
ere
this
is an <-- --- _
analysis
issues%re ^relevant and„>r.e

4
Intra- and inter-temporal distribution can be incorporated.
The
former gives a higher weight
to
benefits which accrue to a
particular group of people
(usually 1the apoor) in. ’ the _c.urx.ent
time , .period.
' TKe latter gives Jaljgher ’weightsto_benefitsrTH^ich
will not be consumed immediately, feut j£ill be saved.
Savings
can then provide benefits
in “a subsequent time period.
The
rationale
is
that society does not save enough to provide for
future generations,
so preference should be given to projects
which will
relat i vely high savings in relation to
result
in
relatively
consumption.

Sometimes
this
is extended
to give preference to government
savings _____
~ overjprivate
savings
___ ___
__
_
on t he _ground
_________
___________
s Tha __e-~gove
r n in ent
would use them to benefit society whereas individuals may not do
soT
As yet,
social
analysis
has not been applied to health
projects and it
is not intended that you should attempt to use
it.
It was
introduced so
that
you will be familiar with a
concept which appears in the literature.

REFERENCES

Major References:
Drummond et al (see Module 4), chapter 7.

Gill M.
et al,
"An economic appraisal of screening for Down’s
Syndrome
in pregnancy using maternal age and serum alpha
fetoprotein concentration",
Sc
i ence and Medicine 1987,
Social
S
cience
24(9):725-731.
Other Reference:

Olsson G.
et al,
’’Economic consequences of post inf arct ion
prophylaxis with B bl ’ kers:
k e r s : cost effectiveness of metoprolol”,
BMJ 1987, 294:339-342.

QUESTIONS

1.
a.

b.

Read the article by Gill et al.
Whose viewpoint was taken in the analysis and is the
conclusion likely to differ if another viewpoint is taken?
Comment on the way that costs were calculated, including
whether any important costs were omitted and whether
mar^jnal or^average costs were used?
Have all the important benefits been included and measured'^
correctly?

2.
Mindful that economic costs are opportunity cost's, how would
you measure the costs to society of using:
a.
unskilled labour;

R1TISH MEDICAL JOURNAL

753

27 MARCH 1976

1

MEDICAL PRACTICE

I

I

t

Contemporary Themes
Preventing the birth of infants with Down’s syndrome:
a cost-benefit analysis
SPENCER HAGARD, FELICITY A CARTER
British Medical Journal. 1976, 1, 753-756

Summary
The costs and economic benefits of providing routine
prenatal diagnosis of Down’s syndrome with termination
of affected pregnancies in older pregnant women in the
west of Scotland were examined. The potential econo
benefits would be greater than the costs for women age
40 and over, probably about equal to costs for those aged
35 ai iver, but less than costs if the service were e en

from terminating pregnancies affected with Down s
we considered two situations, one in which women ao not become
pregnant again after termination (no replacement), and the other
in which termination is followed by a further pregnancy
(replacement). The outcome of replacement pregnancies is
assumed to be normal. Since a programme testing for Downs
syndrome could also diagnose fetal myelocele, the costs and

economic benefits of this are also taken into account.

Method

to Wtjien under 3o.

Introduction
Down's syndrome accounts for between a quarter and a third of
all moderate and severe mental handicap (intelligence quotient
(IQ) < 55) in children of school age.1 = Its birth prevalence is
currently being slightly reduced by providing prenatal diagnosis
and selective aboition to older pregnant women on request, lo
make a major impact on the problem, however, would require
mass prenatal diagnostic programmes directed, in the first
instance, towards all older pregnant women—that is, those at
higher risk. We consider here the relation between the likely costs
and economic benefits of establishing such a prognunme in the
west of Scotland. In evaluating the economic benefits resulting

rates and degree of handicap of survivors; (c) the costs to society o
caring for survivors; (d) the characterisucs, including number o
affected births prevented, of a prenatal diagnostic programme, an
(e) the costs of running such a programme.
The savings in resources made by preventing affected births, m
both replacement and no replacement circumstances, were otaila‘e“;
The costs and economic benefits of testing all specimens for ieta.
myelocele were also considered.
_

All costs were standardised to a value for July 1974, using the
Price Index,3 and future costs were discounted at 10%, the rau
currently used by the UK Treasury, to obtain the net present values^
The variations of costs with benefits under different conditions were
examined.
Results

University Departments of Community Medicine and Medical
Genetics, Glasgow G20 9NB
SPENCER HAGARD, mb, dph, employee of Greater Glasgow Heaitn
Board on secondment to University of Glasgow
Health Services Operational Research Unit, University of Strath
Clyde, Glasgow G4 ONG
FELICITY A CARTER, msc, research fellow

epidemiology and natural history

Differences in methods of ascertainment prohahly:acc0'Jnt

“52

754

BRITISH MEDICAL JOURNAL

TABLE :‘
birth prez^lencc of Down's syndrome by five-year maternal age
■'tsv Lindsjb^
t
groups cfter
.
Maternal age (years):
Birtx prevalence, 1 in:

<19 I 20-24 -29

1685 i 1352 ! 1133

-34 < -39
687 j 267

-44 ■ ^-45
67 ! 16
-

population and age-specific fertility predictions.5 6 This gave estimates
of the number of births of infants with Down’s syndrome in each
five-year age group of women in each, of the years 1975-94. We
calculated that total births to women over 40 would vary a little about
a mean of 550 a year while to women aged 35-39 the number would
vary between 1780 and 2340 a year. Over the 20 years an average of
5-4% oi all births but 29% of births of infants with Down’s syndrome
would be to women aged 35 and over; 1-1% of all births but 16%of
births or infants with Down’s syndrome would be to women aged 40
and over.
Survival—An increasing proportion of infants with Down’s
syndrome survive.' Estimates of likely survival to various ages were
calculated from the results of two comprehensive surveys8 8 and are
shown in tables II and III.
Mental handicap—Using data collected in north-east Scotland10 we
estimated that 20% of children with Down’s syndrome would have an
IQ in the range 50-69, 75% in the range 20-49, and the rest below

20.
ble II Estimated proportion of cohort of 100 people with Down's syndrome
rviving a: various ages

Age years):
No of survivors:

1 2 3 4 5 10 15 20 25 30 35 40 i 45 50 60.70
76 73 /I 69 69 65 60.55 49 44 39,33 1 27 22 11 ! 0

USE OF RESOURCES BY THOSE WITH DOWN’S SYNDROME

The economic benefit of preventing the birth of handicapped people
is the cos: to the community of their care. In the case of an abortion
followed by a successful normal pregnancy (replacement) this is the
difference between the cost of caring for a handicapped person and
that of caring for an average person. When there is no further preg­
nancy (no replacement) the cost is the total cost of caring for a
handicapped person. In estimating these costs the use of resources by
a nominal cohort of 100 people live-bom with Down’s syndrome was
assessed. An estimate of the costs that would be incurred in the
replacement situation is illustrated for representative years in table
III. The derivation of these costs is indicated below; more detailed
consideration is available elsewhere.11
Permanent care—We estimated the percentages of survivors in
permanent care using projected Down’s syndrome survival rates and
the results of a recent west oi Scotland survey.12 We considered that
out a quarter of patients would be in permanent care by the age of
id, half cy the age of 25, about three-quarters by the age of 35, and
all by the age of 45. The additional cost of permanent care over
residence at home is higher for children than for adults: the childhood
costs inc.ude those of education. For a birth cohort of 100 the costs
of permanent care were calculated to lie in the range £3000-£16 000
a year (replacement and no replacement).
Education—In general children with Down’s syndrome with IQs
over 50 20°0) attend special schools, while the remainder attend

27 MARCH

junior occupational or day care centres.13 14 For those continuing t
live at home special education was estimated to cost £100-4400 i r
per child per year than normal schooling—£10 000-£25 000 j rt
per year for a birth cohort of 10Q (replacement and no replacement'
Lost maternal income—As their children get older an increasin
proportion of women go out to work, but many of those with ha j
capped children stay at home to look after them. We assumed
lt
labour force participation among mothers of children with Do..„’<
syndrome would be half that of average mothers with children of th<
same age. Hence, using published wage and employment data,1 -c
estimated that for a birth cohort of-100 six mothers would forgo
L
potential earnings of over £6000 a year when their children ) _e
young, and 11 to 15 mothers would forgo around £14 000 a year later
on (replacement and no replacement).
Additional costs—Since no useful data are available on the additi Li
services required to cope with the greater physical morbidity oft] e
with Down’s syndrome, we omined the possible costs of such services
from our calculations. The inability of most people with Dow-’j
syndrome to work, however, imposes a considerable economic bui q
on society. We assumed that only those with IQs over 50 could w
and then only with half the productivity of an average person. Using
average lifetime earnings and consumption data,1410 we calculated that
in the replacement situation for a birth cohort of 100 this would im; e
costs rising from about £45 000 a year at age 16 to a maximun f
about £110 000 a year at age 25.
Total costs Estimates of total annual costs in each age group were
obtained by adding the estimates under each heading for each y .In the replacement situation and for a birth cohort of 100 they v j
calculated to increase from under £5000 a year in infancy to Over
£120 000 a year in middle life. A discount rate of 10% was applied to
each year s total, and the discounted individual year totals ad> I
together to give the net present value df caring for a birth cohor f
100 people live-bom with Down’s syndrome. In the replacem^:
situation this was £415 000—that is, £4150 per person, and for no
replacement £10 620 per person. The lifetime consumption c i
available for calculation were not entirely appropriate to no repla
ment circumstances, however, and they probably exaggerated the
present value of costs.
Myelocele The figure of £3940, estimated as the economic ben
from preventing the birth of an infant with myelocele,17 was use'. >
calculate the economic benefits of preventing such births in the cou
of a Down s syndrome programme.



-Tr t >. • .

r

.

PRENATAL DIAGNOSTIC PROGRAMME
We calculated the costs of a programme to examine 550 women ap-4.
40 and over each year; these women were estimated to be at risk
having 91 live-bom infants with Down’s syndrome. We assun
that 90°rt <495) of the women would attend antenatal clinic at the
appropriate time for diagnosis (14th-18th week of gestation)
that this would not entail extra antenatal clinic visits, that the chraci
istics of attenders and non-attenders would be the same, and that, ai
genetic counselling, all attenders would accept amniocentesis (done
on condition that affected pregnancies diagnosed would be terminated).
All would receive ultrasonographic examination and amniocentei
About l°0 '5) of amniotic fluid chromosome analyses would requ
repetition fcalculation based on data supplied by Glasgow and West o*
Scotland Genetic Advisory Service), which would entail repeat
attendance, ultrasonography, and amniocentesis. A diagnostic succ

TABLE in—Costs of canng jor cohcr: of 100 people live-bom with Down's syndrome in the replacement situation
_
Permanent care

Age
1
2
3
4
5
10
16
20
45

No of
si^vivors
76

1

No ("..) in
care

Cost

4'5)

4588
4588
4588
3441
5735
9176

MJ

65
59
55

s (13)
14 (23>
13 (23)
27 (100

(£)

3528
3276
6804

i

;
Cost of
education

(£)

23560
20913

~
Maternal income

No (•'„) of
No of children
mothers
living at home unable to work
69
67
66
64
57
44
42

\

6 (9)
6(9)
6 (9)
.12 (19)
11 (19)
12 (27)
11 (27)

I

I

j

j

!

Cost

(O

6660
6660
6660
13320
12210
13320
12210

GC)

j Present value
1 (discounted
at 10%)
|
GO

4588
11248
11248
10101
42615
37419
60787
76765
96130

4171
9295
8451
6899
26460
14425
13227
9514
1317

Additional

78’

—4880
43939
61279
89326

Total

I

j

I

I

Total

£415000

755
27 MARCH 1976

BRITISH .MEDIAL JOURNAL

rate of 99v, would be expected.1’ Thus 8-1 cases (99 of 90 „ of
911 would be identified. In addition, 1-4 cases of anencephaly and l-_
cases of myelocele (1 infant born alive) would be identified m this
P°C«n-Costs to rhe Health Service.and participants were calculated
and are shown in table IV. Costs relating to ume lost. from 'vor^
were derived from government data1’ and those to medical ^n«ics
Provision From local data (supplied by the Glasgow and West of
sTotiand Genetics Advisory Centre). Publici.y,
general practitioners, obstetricians, and the pregnant
estimated to cost the serv.ee £2500 a year with an add. tonal ,C-500
in the first vear. Attendance for genetic counselling would impose costs
^wo^.^d-2

r£?«VaX). T2r”“s vv^-uld aH bTborne by the service.

TXBLE iv—Cost of prenatal diagnostic programme for Down's syndrome for
zeomen aged 40 and over in zeest
zvest of Scotland

Annual recurring
costs

Capital costs
Initial
Cost
to
Health
Service

Recurring after:

To
patients
and or
husbands



(X)

2871

372

1448
2500

I
9929

2500

35 700
102

81
3704

20

. 25 054

20

25 054 35 302 20128

5371

Capital: 63 376; recurrent: 25 499

“ruipment,’ which would need

“placingfronr tirAe to time, and £81 a year for its
£3704 a year for disposable materials, utilities, rates, and
^on Ae total cos/of the first year of the 2
given by adding the imtialrapitrdaists (£6
*
annual recurrent costs (£20 128 to tne service
*,

and £20 in^e 16di. Accommodation and certain equipment cosu

(dtOt^raTofapi^

;US-

counted individual year totals added together to_give: thenet present
value of total costs of establishing and maintaining a 20-year diagnosn
programme—£311 855.

TOTAL BENEFITS AND COSTS

1

Discussion
Economic benefits were divided by costs to give benefitxost
>s. For women aged 40 and over these were: 1 13 (replaceratios.
it) and 2-58 (no replacement). If, after genetic counselling,
^Ke;t^a==’aA^0Ssible temtination the
benefit most ratios would be reduced by 25%
25°(to
jo 0-84
0 84 in the
replacement situation, for example). But the no replacement
situation would probably apply to almost all women of this age,
so the overall benefit :cost ratio would considerably exceed unity,
even were participation reduced so considerably and even after
allowing for the probable exaggeration of the no replacement
hv the
the inappropraite
inaonropraite nature of the available lifetime
ratio caused by
consumption data. This suggests that there would be net
pro­
economic benefits from providing a .prenatal diagnostic
.
gramme for this age group. This view is remtorced by new
evidence from several centres, including Glasgow, that the
birth prevalence of Down’s syndrome among women over 40

were: 0-63 (replacement) and 1-25 (no replacement) If after
genetic counselling only half accepted amniocentesis and possible
termination the benefif.cosr ratios would again be reduced by
about 25% (to 0-48 in the replacement situation, for examp ).
But were this 50% participation confined to those under 40,
perhaps as a result of their acting on self-recognition of their low
risk of fetal abnormality, the benefit:cost ratios would be slightly
higher than 0-63 and 1-25. The availability and inclusion of such
resource use data as those relating to morbidity from Physl“l
illness or to social services would probably also raise the benefit,
cost ratios; incorporation of better Ufetime “nsu”P“°nh^“ “
the no replacement situation, however, would probably have: the
opposite effect. Furthermore, although no replacement could be
expected to apply to most of the age group, the bias in its favour
vXld presumably be less pronounced than among the over 40s
With all these factors taken into account, the costs of a prenand

2094

at least 20 years hfe

I

replacement situation).

m^of6 the complete oveJsS age group the benefit most ratios

Start

Totals . .
diagnostic

To
Health
Sennce

2500

2500

Accommodation . •
Equipment
Running costs

Total cost of
programme

20

vears years years
’ C • • (X) • <X)

(X)
Publicity
.•

Genetic Counselling ’,493
consultations)
Staff costs:
5 Senior registrar
sessions week
Patient costs:
at £160 (women)
at £4 20 i men i.
Ultrasonography (500
examinations)
Start costs:
2 Midwifery sister
sessions, week
Amniocentesis (500
examinations)
Start costs:
4 Registrar sessions
week . ■
••
Patient costs:
at £5 00
Laboratory costs

io-

~~5

to give a net present value for the total economic benefits of the
programme, amounting to £351 699. The corresponding va ue of

The programme described, costing £311 855, would p-venr -te
bmhs of 8 1 children with Down’s syndrome each year and 1 0^
mXle, and thus in the replacementsttuauon produceannual
economic benefit of (8-1 X £4130) + (1 0

economic grounds, this programme too could probably be
justified. Its development might follow initial provision to the
over 40s, from which the collection of appropriate epidemio­
logical and resource use data would provide for more accurate
economic calculation for the younger age group.
It has been suggested that it would be economically justifiable
to extend prenatal diagnosis of Down’s syndrome to women
under 35 1 ’ The data and calculations presented here maxe tm
seem unlikely. Such a decision would therefore have to rest on
other considerations of the priorities for health expendmne. _
any case the view that amniocentesis should be offered to women
whose risk of diagnosable severe fetal abnormality is remote
neOmCcTnXs^nsX^t1oTtide assumption that those responsibk

for health planning embrace the cost-benefit analysjs approa x
—perceiving their responsibilities beyond the narrow framewor
of health services accountability, and accepting a
economic perspective. The findings also re-emphasise tha.
society’s respome to the problem of Down’s syndrome canne
rest solelv on consideration of economic costs and benents.
Down’s syndrome is socially unacceptable provision ot a pre
gramme to reduce its birth prevalence by scarcely a third woul
be an inadequate response. Conversely, fadure to imp em n ■
programme for aU maternal age groups would imply mut me

756

. *•

were ether, perhaps more appropriate, responses to the problem
of'Down’s syndrome. Since this would call into question any
«£rog:*mme directed at identification and termination of affected
pregnancies, it would be logical to resolve this dilemma before
any programme was started.
The authors thank members of staff of the Glasgow and West of
Scotland Genetic Advisory Centre for providing detailed information
about their work, Mr R G Milne for several original ideas, Dr F A
Boddy, Professor M A Ferguson-Smith, and Professor G T Stewart
for giving helpful advice, and Mrs F Sinclair for secretarial help.

References
1 Drilhen, C M, Jameson, S, and Wilkinson, E M, Archives of Disease in
Cteldhood, 1966, 41, 52S.
* Mental Handicap, p 17. London, Office of Health Economics, 1973.
3 Economic Progress Report. London, Information Division of Treasury,
1914.

4 Linds: o, A, A eta Paediatrica Scandinavica, 1974, 63, 571.
5 General Register Office, Scotland, Report of Registrar General for
Scotland, Edinburgh, HMSO, 1973.

BRITISH MEDICAL JOURNAL

27 MARCH

r

5 Genera; Register Office, Scotland, Report of Registrar General for
Edinburgh, HMSO, 1973.
ar'
■ Penrose. L S, and Smith, G F, Dozen's Anomaly, p 152. London, Cru
1966.

11
8 Collma.-_-., R D, and Stoller, A, Journal of Mental Deficiency Rei
1963, 7, 53, 60.
c
’ Fabia, J, and Drolette, M, Journal of Mental Deficiency Research, 197

10 Ross, H S, Innes, G, and Kidd, C, Scottish MedicalJournal, 1967, 12 5
11 Carter, F A, and Hagard, S, Calculating the Costs of Down's Svm r
University of Strathclyde Health Services Operational Research Um
Mimeograph, 1975.
12 Milne, R G, Institutional Provision for Mongolism and Spina Bifi
Scotland, submined for publication.
13 McKeown, T, and Lowe, C R, An Introduction to Social Medicine r
2nd ed. Oxford, Blackwell, 1974.
U SCHA1SO d1961iOn DeparrmCnt’
of Mental Handicap. Edinb—7

15 Central Statistical Office, Social Trends, London, HMSO, 1973.
18 Central Statistical Office, Annual Abstract of Statistics. London, HMaC
17 Hagard, S, Carter, F A, and Milne, R G, British Journal of Preventiv
Social Medicine, in press.
18 Ferguson-Smith, M A, et al. Health Bulletin, in press.
12 Stein, Z, Susser, M, and Guterman, A V, Lancet, 1973, 1, 305.

n

Problems of Childhood
Wheezing children
J K SARSFIELD
British Medical Journal, 1976, 1, 756-759

The wheezing child is a common clinical problem. Asthma is by
far the commonest cause of wheezing, but other conditions must
be considered, especially in the young child. These include
respiratory tract infection, expecially viral bronchiolitis, inhala­
tion of a foreign body, cystic fibrosis, primary tuberculosis, and
congenital anomalies. Clinical and radiological examination will
sually identify these conditions.
The role of infection in wheezing in the young child is not
clear. Acute bronchitis with airways obstruction may produce a
wheeze with associated respiratory distress, fever, leucocytosis,
and general malaise. If these attacks recur terms such as
“wheezy bronchitis,” “asthmatic bronchitis,” and “pseudo­
asthma"’ are applied. Antibiotics are often prescribed but with
doubtful benefit. Viral infection may be responsible for some
attacks, but several studies have shown that viruses can be isolated
from only about a third of patients and their role in the patho­
genesis is uncertain. Williams and McNicol1 made a valuable
contribution to our understanding of this group of young
wheezing children who have appareiit preceding infection.
After a prospective long-term epidemiological study they con­
cluded that they could not separate recurrent “wheezy
bronchitis” from asthma. Their evidence suggests that both
conditions exhibit a common basic asthmatic disorder, but the

Department of Paediatrics and Child Health, University of Leeds,
Leeds LS14 3ET
J K SARSFIELD, aid, mrcp, lecturer in paediatrics

spectrum of severity varies greatly from a few mild early attacks
that abate to established severe asthma persisting into adult:.
Asthma
In the absence of any widely accepted definition of asthm t
seems reasonable to consider it, simply, as a constitutional cuaorder characterised by hyper-reactivity of the airways. Various
factors ma\ provoke this reactive state and lead to paroxysi 1
attacks of airways obstruction producing respiratory disti »
and wheezing. The basic constitutional disorder is probably
biochemical and almost certainly has a genetic basis, but
precise mode of inheritance is unknown. This familial tenden ,
however, may help the doctor towards an earlier diagnosis a I
hence more appropriate management.
An understanding of the known basic mechanisms concerr'd
in the cause of asthma is essential to proper assessment 2
management.

Allergy
Most asthmatic children have demonstrable allergies. The
incidence of hay fever, eczema, and urticaria is much higher
these children and their families than in non-asthmatics. T
fact that an acute attack of asthma may be a manifestation of
immunological hypersensitivity reaction to an external provoking
antigen (allergen) has been recognised for many years. Rec
advances have established that antibodies belonging to a nei
recognised immunoglobulin class, IgE, arc concerned. Th'
antibodies are firmly attached to mast cells in the bronchial
mucosa, and on exposure to an offending allergen the antiboc
antigen reaction causes the release of vasoactive amines fn

ft

i.. i)»<

<•» '

W 0 t L 0
.

H E A L T H

Ki



0 I 6 A H I I k T I 0 N



t U I I I C A T I 0 N S

I

Global eradication of poliomyelitis:
benefit-cost analysis

•!

L•

.Basic
Analytical Toxicology
This-book provides a practical guide to the labora*
tory analysis of over 100 substances frequently
invol ved in episodes of acute poisoning. Noting that
many hospitals, especially in developing countries,
lack the support of analytical toxicology services,
the book aims to help laboratory staff perform a
range of simple tests known to produce rapid and
reliable results for the management of poisoning
emergencies. All tests described can be performed
without the need for sophisticated equipment, ex­
pensive reagents, or a continuous supply of electric­
ity.
■- >.
The manual opens with general information about
the organization and functions of an analytical toxi­
cology laboratory, the principles of safe laboratory
practice, and the essentials of emergency medicine
and intensive care that will influence the laborato­
ry’s work. A brief introduction to the apparatus,
reference compounds and reagents needed in the
laboratory is followed by an explanation of basic
cHhical issues. Details range from a table listing the
clinical features associated with some common poi­
sons to a description of essential symptomatic and
supportive measures that can be taken before the
j diagnosis is confirmed.
Chapter three explains the most useful biochemical
and haematological tests for the diagnosis of acute
poisoning and for assessment of prognosis. The
final introductory chapter goes through; the do’s
and don’ts of laboratory practice pertaining to safety,
the performance of colour tests, the pretreatment of
samples, and procedures for the use of thin-layer
chromatography and ultraviolet and visible spec­
trophotometry.

The remaining two chapters, which constitute the
core of the manual, describe the many simple ana­
lytical tests that can be used td detect and identify
poisons, whether biological fluids or in powders,'
tablets, or other items found near the patient. The
first chapter, on qualitative tests for poisons, sets
out a three-part series of tests designed for use as a
routine, rapid screen, especially appropriate in the
many cases where the identity of the poison is
unknown.

The second and most extensive chapter provides
step-by-step instructions for the performance of .
qualitative tests and some quantitative methods for
113 specific poisons or groups of poisons. Sub­
stances covered range from pesticides and other
industrial chemicals, through compounds contained
in household products, to pharmaceuticals, plant ■
toxins, and drugs commonly abused. To assist com­
munication between the toxicologist and the clini­
cian, each monograph also contains information on
clinical signs of intoxication and recommended
treatment. All techniques and procedures have been
tested by laboratory technicians in developing coun­
tries to assure reliable performance using relatively
simple apparatus.

KJ. Bart,’ J. Foulds,'& P. Patriarca3
*

A benefit-cost analysis of the Poliomyelitis Eradication Initiative was undertaken to facilitate national and
' international decision-making with regard to financial support. The base case examined the net costs and
benefits during the period 1986-2040; the model assumed differential costs for oral poliovirus vaccine (OPV)
' and vaccine delivery in industrialized and developing countries, and ignored all benefits aside from reduc'
tlons In direct costs for treatment and rehabilitation. The model showed that the ‘break-even" point at which
benefits exceeded costs was the year2007, with a saving of USS 13600 million by the year2040. Sensitivity
analyses revealed only small differences In the break-even point and in the dollars saved, when compared
with, the base case, even with large variations in the target age group for vaccination, the proportion of case-,
patients seeking medical attention, and the cost of vaccine delivery. The technical feasibility of global
eradication Is supported by the availability of an easily administered, inexpensive vaccine (OPV), the
epidemiological characteristics of poliomyelitis, and the successful experience in the Americas with elimination of wild poliovirus infection. This model demonstrates that the Poliomyelitis Eradication Initiative Is
economically justified.

Introduction
Developing countries are confronted with destabil­
izing health problems and with a serious shortage

of resources. The prospects for per capita income
growth in many countries have deteriorated, and the
adoption of structural adjustment policies calls for a
rigorous review of public investment programmes.
The allocation of resources in the health sector in the
past has not been efficient and equitable, owing to
an emphasis on expensive urban and hospital-based
curative care, and was not'directed at the main
causes of ill nealth in the majority of the population,
. especially in the less developed countries.
Only in the past decade has immunization, one

of the least expensive and most cost-effective of all
health interventions, which has been confirmed by
cost analyses, been accorded a high priority (1). The
prospect of removing the burden of a disease and its
treatment for ever, and at the same time eliminating
the continuing costs of vaccinations, is an attrac­
tive policy alternative. Prior to the development and

Basic Analytical Toxicology
fl. J. Flanagan, RA. Braithwaite. S.S. Brown,
B. Widdop, and FA de Wolff
1995, xil + 274 pages + 8 colour plates (available in English;

French in preparation)
ISBN 92 4 154458 9
Sw.fr. 60.-AJS $54.00
In developing countries: Sw.fr. 42.Orderno. 1150437

’ .Associate Director for Medical and Scientific Affairs, Office of
International Health, Office of the Secretary, Department of Health
and. Human Services (DHHS), Room 18-74, 5600 Fishers Lane,
Rockville, MD 20878, USA. Requests for reprints should be sent to
this author.
1 Project Officer, Division of Microbiology and Infectious Diseases,
National Institute of Allergy and Infectious Diseases, National Insti­
tutes of Health, DHHS, Bethesda, MD, USA.
» Associate Director for Medical Affaire, Division of Viral Products,
Office of Vaccines, Center for Biologies Evaluation and Research,
Food and Drug Administration, DHHS, Bethesda, MD, USA.
Reprint No. 5870

WHO • DISTRIBUTION AND SALES • 1211

GENEVA 27 • SWITZERLAND

introduction of poliomyelitis vaccine, up to 32 out
of every 100000 children born in the world had
permanent lameness as a result of infection with
poliovirus (2-4).
The early successes in the Americas, through
the expanded programmes on immunization (EPI),
led the Forty-first World Health Assembly in May
1988 to adopt a resolution (WHA41.28) to eradicate
poliomyelitis from the world by the year 2000 (5).«
This goal was confirmed in 1990 at the World
Summit for Children (6). Poliomyelitis eradication
is an example of the EPI focus on the impact of
immunization on a target disease. In addition, the
progress towards such eradication is seen as pro­
viding a measure of the progress towards achieving .
the WHO goal of Health for All by the year 2000,
i.e., reaching and maintaining >90% coverage with
current EPI antigens for all children (7, 8). Under
WHO’s global leadership of EPI, an estimated 80%
of the world’s children were fully immunized in 1993
against poliomyelitis; even so, the disease still causes
paralysis in over 100000 individuals each year and
kills perhaps more than 10000 (9).
The decision to undertake eradication has eco­
nomic implications for the poorest countries and
donors.who are concerned about the potential to
divert resources from other activities with a poten­
tially greater impact or to interfere with the develop­
ment of primary health care (10-13). A benefit-cost
analysis of global poliomyelitis eradication war’*-""'
therefore recommended so that decisions aboutr
national and international financial support for this>5
effort could be made.

KJ. Bc*t «t al.

Benefit-cost analysis of poliomyelitis eradication

Methods
The model

The global poliomyelitis eradication programme has
been modelled as a unit effort from the beginning of
the first eradication efforts, with projections beyond
eradication; benefits have been calculated up to the
year 2040. The number of children to be immunized
with oral poliomyelitis vaccine (OPV) annually, the
disease incidence and morbidity, and the vaccine’s
efficacy permit an estimation of the number of cases
prevented by vaccination and the cost of achieving
these reductions. Reductions in the morbidity and
mnrtalitu
_____ ___ ___ * j

.
mortality, and the _consequent
drop •in demand
for
treatment and rehabilitation constitute the principal
, the
...base case
? (see
Z
benefits of OPV --------------vaccination in
below). In addition, in order to approach a more
complete estimate of net present value, additional
direct benefits of eradication, i.e., the savings in vac­
cine and its delivery, have been added to the treat­
ment and rehabilitation costs and compared with the
cost of
whether me
the
— the programme to determine
wciEiuiuie wncmer
programme u economically justifiable (14^19}^ -h -

TTte analy^s

m

df treatment and rehabilitation of that number of
poliomyelitis cases, the cost of the eradication effort
(the vaccine, the cold chain, administration of the
programme, the deployment of delivery teams,
social mobilization and the immunization strategy
chosen), and the net benefit (i.e., the reduced treat­
ment and rehabilitation costs). The analysis com­
pares the annual total costs >vith the total benefits
throughout the entire period from pre-eradication
till after eradication (20,2/)/
The analysis presented is a model designed to
simulate as closely as possible the global eradication
programme now being undertaken. While the niodel
simulates what is taking place, some inputs, costs and
Drocramme
Pr°gramme orni^ntinnfi
projections ran
can nniu
only be estimated ___
and■
ac
may differ from country to country within regions as
well as between unions.
Unions. For example, what propor
proper ­
tlOn
Of
&
childhood
nnniilfitinn
tinII
ka
__
tion of a childhood population will be reached during
a national vaccination day, “mpp-up
mop-up”,»
,• or outbreak
response;
iponse; and will there be just 5 years of national
vawine days or more than that, 2 years of mop-up,
and the undertaking
oftsignificant
outbreak res[K>nse
c .• .
-------------?cuy,l,e1' «timate» of costs and benefits and pro•



,

t

/



-------- —VW4MI4UIV,14

“lected from

- —----------------- * — —

.VIIWO.

J,—' ’IT" ” —~,"r

mwh ww airjcoat-

' EPI. Pdiomy^ HMveva bv’malonr EP/
Section 3, Table 3.4, July 1988 (unpuUlebed)

ftPt 1W (unpubi^ ^e^.'
d^iv9rino 0PV

R^CH pra>«

mant WHCVEP^SLw.Z, 1902.
<iocum^tWHO^/Po4kV94.1^.

Senskivuv analyses
..............
...
Sensitivity
performed
where dif-r
dif<tu<uyaca were
penormea ......
fenng programme strategies, e.g., age of the target
population for national vaccination days, discount
rot.
rate and .kthe cost. of vaccine, or where costs Vary
widely in published data between country experi­
ences, such as the cost to immunize a child or the
amount of vaccine wastage. Only one assumption
is varied at a time in the base case to test the effect
of each parameter in question. Since the estimates
of costs are only available from some countries,
data reviews and surveys for the less devel- •
oped countries ,~l and for industrialized coun­
triesy were used in addition to individual published

. Net Present Value

i

developing world —

n^itat,onl for PoHomyelitia, North Arcot Diag^Sal

M
WHO Bu»Mn OM8. Vot 74 10M

h ,

(2) fhadicatipn of poliomyelitis is taken as that
point where the transmission of the causative organ­
ism has ceased in an irreversible manner, vaccine is
po longer in use and, as a result, cases and infec­
tion have disappeared from all countries of the
world.*


(3) The estimates of global population, the global
birth cohort and the population living in industrial­
ized and developing countries in each region are
derived from the 1992 mid-year United Nations
population estimates (22). No attempt has been
made to incorporate the growth rate into the cohort
to te vaccinated.

• wiup-up
Mop-up enone
efforts are intensive
Intensive houee-to-houee
house-to-house vaccl
vaccination
lmmunt!«
•*») are at
P'^unmo on Immun-

(1) The base case includes the identification, valuation apd summation of the cost and benefits in each
year of the project’s life. Costs (C) and benefits (B)
were summed over the years, projected and dis­
counted to calculate the Net Present Value using the
formula given below. The present value of this
stream of net benefits is the sum of these individual
terms over the years of the model programme. In the
model, net benefits remain positive, but because of
discounting the benefits are smaller in future years.

*

^^cdnution to combat polio: a coat-banaM
InsUtuto de Peequlsas e Estudoe Econimlcs.
U^jrajdade Federal di" Rio Gr^de Do’ Sut’iew (hTnJllS:

' UnpuWbMd docu'

'

t
'

experiences^

SepI
79^1979 ‘0ukUUn*9- Unpuwiaheddoaimint
UnPuWI«h«<i document wh
WHO/EPV

Syetam.

reports and personal communications to establish
global birth cohort of surviving infants (133831500,
cost parameters.
of which 115272400 are in the developing world and
Two additional models were constructed, the
18559100 are in the industrialized world and Eastern
first using 1988, the year of the World Health Assem­
Europe) who receive four doses of OPV during rou­
bly resolution on eradication of poliomyelitis, as the
tine immunization (at birth, 6, 10 and 14 weeks)
base year and pleasuring the marginal additional
during the first year of life.' It is projected that 90%
■ costs and benefit-dost of moving from control of ..of the target population is reached. In addition, chil­
the disease; to eradication. In the second, benefits
dren aged 13-59 months (1-5 years) who have been
included savings in the cost of vaccine and delivery
identified with incomplete vaccination during rou­
as well as treatment and rehabilitation in order to
7
tine facility-based
vaccination sessions during their
reflect more closely the true benefits of poliomyehtis
Jirst year of life are assumed to be vaccinated at these
eradication.
"
- routine
routinecontacts.
contacts.It Itis isestimated
estimatedthat
that1/51/5of ofunvaccinated children aged 13 to 59 months are identi­
fied and vaccinated each year.
The base case and sensitivity analyses

411111

endmininw thsJjsnsfltjja^der to construct
Jhe least favourable balance in the benefit-cost rclat
tiunship of poliomyelitis eradication. Because it is
difficult and often controversial to place dollar val­
ues on many of the elements on the benefit side of
the benefit-cost equation, only the savings in treat­
ment and rehabilitation following the reduced inddence of disease have been used as the benefit in
this analysis. If eradication can be seen to have a
favourable benefit-cost ratio while ignoring all other
tangible reductions in the costs to the community
and the family, the long-term handicaps, the value of
life and income calculations as well as intangible and
external benefits; it would be expected to be even
more coet-beneficial when these other benefits are
taken into account, no matter how or at what level
they are valued. In addition, this approach enables
comparison with the analysis that was carried out for
the Americas
(19). ■ •
rvuicncas yiy).
. i.
yean of the model programme,
. For each of -the
----- fra
estimates were made of the number of paralytic poliomyelitis cases that would be prevented, the costs

1

..
<

(0) The population to be vaccinated during the routine vaccination programme is estimated as the 1992

* Worid Health Organization. CortificaUon of tha global eradica­
tion of poUomyeHtis. Meeting of a working group. Unpublished
.•document EPI (Poliol/93.1), 1993.
' Th® choice cf vaccination schedule and vaccine (OPV or elPV)
differs within and between the countries of the developing and
Industrialized worid. The WHO-recommended schedule of four
doses of OPV. In lhe first year of life Is used In developing coun­
ties. In the Industrialized worid at least five countries us elPV
solely as th® primary series and at least three countries use OPV
and. elPV.
WHO ButoWi OMS. Vol 74 1996

(5) Immunization coverage estimates are those esti­
mated and reported by the EPI as of October 1993.
Coverage estimates are for three doses of oral polio­
myelitis vaccine (OPV 3), since no systematic cover­
age estimate for the birth dose is available. Costs are
projected for four doses to all children, i.e., as if all
infants had received a birth dose, to maximize the
costs of lhe model.
(6) National vaccination days are projected twice a
year for 5 years in addition to the routine vaccina­
tion programme for all children aged ■€59 months.
These are projected as two doses one month apart.
It is projected that 90% of the target population
is reached during each national vaccination day.
Sensitivity ahalySBrare done for tafger-p&pularions
aged «£36 months and <48 months (Table 1).

(7) In addition, after 5 years of national vaccination
days are completed, in response to the continued
occurrence of cases, 10% of children aged <59
months are projected to be vaccinated in annual
“mop-up” campaigns. It is projected that 90% of the
target population is reached during each mop-up
campaign. These are projected as two doses one
month apart for two years. Sensitivity analyses are
done for target populations aged <36 months and
<48 months, with 1% and 0.1% of all children
targeted to be vaccinated.
——;
(8) Outbreak response is projected surrounding
cases in which 1% of all children aged <59 months
are immunized with two doses one month apart. It is
projected that 90% of the target population is
reached during outbreak response. Sensitivity analy­
ses are done with 0.1% of children aged <59 months
vaccinated.
. (9) Vaccination costs are expressed in 1993 US dol­
lars, and are based on the planned regional pro­
grammes of eradication (Table 2). All costs are
modelled beginning in 1986. The programme for the
Americas is assumed to have begun in 1986 with
national vaccination days and outbreak control car37

i-

I

Beneflt-co«t analysis of poliomyelitis eradication

KJ. Bert et al.

I i5

i

-

tivity analysis for p^omyeUUs^^d^thon^pnh
oranwne f0 WcSter? PfaC^C Region (WR) the programma model
gramme is assumed to have begun tn 1991 with
--------------- ----- J---------------- ---------- - ----------- J.
‘subnational vaccination days" (20% of the children
Ceeto
{ ■ • ; '------------------- ’------------------- less than 5 years of age) having been carried out for
'
I •«

two years prior to the first national vaccination daya
1. FlouHna vaccination
in 1993. It is projected that 90% of the target popu­
a. Population <12 months old
tXi
133830000 plua 20% of
lation was reached during each subnational vaccina­
1-S-year ago group (107084000)
"‘I
— industrialized countries
tion day. Children were vaccinated with two doseu
18560000
— Developing countries
115207000
one month apart.
b. Industrialized countries
In the European Region (EUR) only one-third
Vaccine oost-r-OPV
US| 4.18/doee (base case)
of the population are in countries expecting to con­
Delivery cost — OPV
US$ 5.00/dOM (base ease)
c. Developing countries
duct national vaccination days, i.e., the former Soviet
Vaccine cost — OPV
UBS 0.08/dose (base case)
Union, the former Yugoslavia, Turkey, Romania,
(Sensitivity analysis:
USS 0.12/dOM)
Bulgaria and Albania. This effort is projected to
Delivery cost — OPV
USS 1.51/dooo (base case)
have begun in 1992, and 90% of the target popula­
(Sensitivity analysis:
USS S.OCVdoee)
d. Wastage
tion was reached with two doses one month ’ipart.
OPV
33%
In the EastWn Meditenanean Region (EMR) .
(Scnsltivttyanalyski:
50%)
two yean of subregional vaccination days began in .
21 Ao^^t0d
»ctMtiaa (national vaccination daya
1992 to be followed by national vaccination days for
(NVD), mop-up, outbeak control)
five years. Subregional vaccination days are defined
a. Population <59 months age
669150000
as targeting 10% of the children aged <59 months
— Indu^riallzed countries
92800000
for vaccination with two doses one month apart. It is
— Developing countries
578350000
(Sensitivity analysis;
projected that 90% of the target population was
for mop-up 1% and 0.1%; reached.
,
for outbreak control 0.1%
of the target population.)
National
vaccination
days
began
in the Southb. InduatrlaHzed countries
East Asia Region in 1994 although individual coun­
Vaccine cost — OPV
USS S.IQ/doae,'
tries e.g. India, started subnational vaccination days
Delivery cost — OPV
US$ 1.48/dosS
c. Developing countries
in 1992. The African (AFR) Region began national
Vaccine cost — OPV
USS 0.08/doM
vaccination days in 1995. It is projected that there
Delivery coat — OPV
USS 0.10/doM
will be 90% coverage with two doses one month
(Sensitivity analysis:
USS 0.7WdoM
apart.
3. Fa^pel population for
accoloratad activltiaa «59 months old (base case)
(Sensitivity analysis <48 and <38 months old)

4. Oacounirata

6%

(Sensitivity analysis at 0%, 3%, 10%)

u. BMWfts
Case at paralytic potiomylitia — pre-EPI 5/100000
(Sensitivity analysis: 2/100000 and 19/100000)

J

Proportion ol paralytic caaaa mcolving traatmant and
rehabilitation
Industrialized countries
100%
Developing countries
33%
(Sensitivity analysis:
0%)

Cost ol traatmant and rahabiktation

Industrialized countries
Developing countries

L

USS 25000/casa
USS 250/csm

ried out for 5 years until 1990, with mop-up efforts
that continued for two additional years. Routine
immunization is planned to continue till the year
2005 when eradication is projected to be declared
globally.

1

s

IJliidirniilJlWi

Jr

a: <e ce ce cr

+++++

liiiiiiiiiiiiHiyH i

i

i

a: a: acr tt
+ + ++ +

CE

o

£

II

I °

1 §

8

2

z
£

f 1

•IP I j

I

g •'8

. irii
II Ihl
iFi
z:-.

+ + + ++11

I

S

Ills I®

1 hpi

I i
8
e
1

nj00»i

I i III
I
iiiiiffli
itii
hyy
i
I
a.

(10) Globally, each country will be monitored for
three years by a Global Eradication .Certification
Committee after the last case of poliomyelitis has
been reported, and after vaccinations have been
stopped-and poliomyelitis is declared as eradicated
in the year 2005. Routine immunization is projected
to continue in all regions until they are polio-free.

Hi

(E
+ tr or cr cc

a.

IE
§

(11) Although an incidence as high as 32 per 100000
has been reported, the global incidence of paralytic
poliomyelitis at the outset of the eradication effort is
estimated conservatively at 669158 or 5 per iOOOOO
surviving newborns.
•.
(12) It fs assumed for the purposes of this calcula­
tion that, Jn industrialized countries, 95% of those
vaccinated- with OPV were immunized, i.e, devel­
oped detectable levels of neutralizing antibodies
(23). In developing countries it is assumed that 80%
of those vaccinated with OPV were immunized.
From the limited data available, seroconversion
rates with OPV during national vaccination days
may be 10% higher than during routine immuniza-tion programmes, but this is not taken into account
in these calculations.

o

•4=

3
2a>

I

1

«

I>.
:

58188++

IIff*

HIHil
iiiiihlllliHiilii i still
!L
a: oratE cr

E
o

I
5o
0
C4
0)

.■8

I liHiHIlititiirilH
i

It; is;?
iHSHji

diHH

HOPif

M
WHO BuMn OMS. Vd 74 IMS

WHO Bdtotn OMS .Vol 74 IMS

M

Il
'I

KJ. Bart 'rt al.

(13) -Analysis of secular trends in countries where
substantial coverage has resulted in a significant
reduction in disease demonstrates that with rou*
tine immunization programmes cases fall at an
estimated 40% per year. With mass campaigns the
reduction is estimated at 70% per year. The pro­
jected decline in cases with the continued appli­
cation of vaccine is? projected linearly for ease of
presen-tation. At the end of the programme 669158
patient-cases are projected to have been prevented
worldwide annually,

(14) Programme costs, treatment and rehabilitation
costs, and vaccine costs are stratified by developing
and industrialized countries and are presented in
1993 US dollars. . ,
.
<
(15) The cost of acute care and subsequen/rehabilitation is conservatively assumed to be US$ 25000
per case in industrialized countries and US$250 per
case in developing countries. The base case assumes
that only one-third of all cases in developing coun­
tries receive acute care and rehabilitation; 100% of
cases are assumed to receive treatment and rehabili- ..
tation in industrialized countries. A sensitivity analy­
sis is done assuming that only 10% of cases in
developing countries receive acute care and rehabili­
tation; and that 0% of cases in developing countries
and 75% in industrialized countries receive treat­
ment and rehabilitation.
(16) The expenses of poliomyelitis vaccination,
which are derived from studies sponsored by the
expanded programme on immunization (EPI), are
the result of summing capital costs (buildings, vehi­
cles, refrigeration and the cold chain), operational
costs (including staff salaries, supervision), and the
cost of transport (including fuel and spare parts).
Only the costs to the delivery system are used in this
model. In addition to the costs of purchasing and
delivering vaccine, additional resources are needed
to effect eradication such as training, increased
surveillance activities; improvements in the cold
chain; improved laboratory support; data collection
and processing, etc. Since the largest portion of
the costs are due to vaccine and delivery, these
additional costs are not included to streamline the
model.

(17) For simplicity of the model it is assumed that all
countries use only OPV, the WHO-recommended
vaccine of choice. The cost of OPV for the pro­
gramme in developing countries at 1993 UNICEF
prices is US$0.08/dose at port of entry. Sensitivity
analysis is done with OPV vaccine costs using
US$0.12/dose (a 50% increase as an estimate of fu­
ture inflation). In industrialized countries the aver­
age cost of OPV is US$4.16/dose.

Benefit-coat analysis of poliomyelitis eradication

costs, and since the rate of vaccine-associated polio­
myelitis is so low, vaccine-associated poliomyelitis
has not been included in these calculations.
(26) The costs of treatment, rehabilitation and vac­
cination will end at eradication, and the net benefits
accrued Are estimated beyond the year 2007 for a
total of 55 yeais.

(18) The cost of vaccinating a child in the devel­
oping world with OPV is estimated at US$1.51/
dose. A sensitivity analysis is done at US$3.00.
This is based on the EPI estimates that in the
developing world the delivery system costs are
estimated to be US$15.00 to fully immunize a
child. Assuming OPV is given at 4 of the 5 visits
required to fully immunize a child, the cost of
fully vaccinating a child with OPV is estimated to be
four-fifths of US$15.00/child. In order to fully
test the benefit-cost relationship, the cost of vac­
cinating a child with all vaccines during each of
these four visits is attributed, to poliomyelitis,
i.e., US$3.00/visit. For industrialized countries, estimates for the delivery of vaccine are estimated at
US$5.09/dose.

15

*12
M

S

a. 9

1 6
Results
Th9.baa9 case

Using the assumptions and parameters described,
the net costs and benefits of the global eradication of
poliomyelitis were calculated for a period of 55 years
from 1986 to 2040. As shown in Table 3, the costs
exceed the net benefits in 2007, at a base case dis­
count rate of 6%. By the year 2040, the saving will
amount to USS 13640 million.

(19) Based on
estimates, .the cost of delivery
during national vaccination days, mop-up and .out­
break response activities in a developing country is
US$0.10 per dose. A sensitivity analysis is done at
US$0.79/dose. In industrialized countries the cost of
a contact is US$1.48/dose. A sensitivity analysis is
done at US$2.47/dose.
(20) The estimate of the benefits of poliomyelitis
eradicatiqn fakes no account of the reduced pain
and suffering or deaths due to poliomyelitis, the
greater productivity of individuals who would
otherwise be paralysed and become unproductive,
the improved quality of life, or the reduction of
other vaccine-preventable disease that could be ex­
pected to result from a successful programme against
poliomyelitis.
(21) As an additional model, eradication of the
267663 cases (2 per 100000, down from 5 per
100000) estimated to be occurring in 1988, the
year of the* World Health Assembly resolution on
global eradication of poliomyelitis, is taken to
represent the marginal additional costs to move
from a routine vaccination programme directed
al control of the disease to one directed towards
, in.:'
■' eradication,
j(22) To explore further the full net present value
of eradication, a second additional model is pro­
jected, adding the costs of vaccine and delivery to
those of treatment and rehabilitation as items of
benefit.
(23) Costs and benefits are discounted at 6% annu­
ally. A sensitivity analysis is done at 0, 3 and 10%.

8

3

2007.

? 0

I

Benefit
Cost

-3

l

-6

-9
1980

.1990

2000

2010

2020

2030

2040

Year

JTable 3: . Net costs or benefits of global poliomyelitis eradication: base case and sensitivity analyses, for the year

I

Bam case

II

Senaltivtfy analyses
•. Target age of accelerated acUvitlea (base case <59 months):
. <36 months
<48 montns
b. Proportion of target age reached by mop-up (base case 10%):
. 10%
0.1%
c. Coat of routine vaccine (OPV) and delivery (base case Is USS 0.08):

Developed countries USS 3.00
Developing country USS 0.12
d. Coat of accelerated activities/contact
Developing country USS 0.79
(Base case USS 0.10)
Industrialized country USS 2.47
(Base case US$ 1.48)
a. Access to treatment and rehabilitation (base case 33% for developing and
100% Industrialized):
10%/100%
0/100%

0/75%
f. Vaccine wastage (base case Is 33%)
!
53%
' g. Discount rate (base case Is 6%)
0%
3%
10%

(24) Vaccine
, . wastage
. , - is estimated at 33%. A .sensianalys,s w d°ne a< 50 /o.
(25) While vaccine-associated poliomyelitis can, in
individual cases in industrialized countries, rarely be
associated with substantial treatment and litigation
WHO Butetin OM8. Voi 74 10M

Fig. 1. Global poliomyelitis eradication model: base
case.

Ill Addition*) models
a. Treatment and rehabilitation plus vaccine and delivery
b. Acceleration from control to eradication (base case 5/100000)
2/100000 to zero (treatment and rehabilitation plus vaccine and delivery)

'

WHO BuM<n OMS. Voi 74 10M

Costs or
benefits in 2040
(USS millions)

Year when
benefits
exceed costs

13640

2007

16310
14970

2005
2006

14240

14260

2007
2007

13690
3790

2007
2024

11270

2010

13480

2008

13480
12870
3070

2007
2008
2023

10210

2011

86130
34 500
1850

2004
2005
2017

27360

2005
2010
2026

11440
3380

■ i

41

Une net cumulative cost or benefits of the base
case are shown in Fig. 1 in millions of US dollars.
Each data point represents the net cumulative costs
or benefits for the model programme to that date.
The slope of the line falls as a result of net costs
increasing as each of the Regions begins its acceler­
ated vaccination efforts. In the year 2000 th6 benefits
of the programme are seen as the number of averted
cases increase, and the curve begins to turn upwards.
For the base case, the year 2007 is the break-even
point, the year in which the savings exceed the pro­
gramme costa. From this point on, the benefits of the
programme exceed the costa in every year, and the
net benefits continue to increase. This increase con­
tinues after the planned end of the eradication effort
since the benefits of eradication continue beyond the
programme, i.e., through cases prevented and since
treatment and rehabilitation are no longer needed.
As seen by the plateau of the curve, although
the benefits of eradication continue in perpetuity (no
cases with the attendant costa are occurring, and
there is no longer any requirement to vaccinate), the
dollar benefits decrease ’substanUally as a result of
discounting.

developing countries (US$0.79) had the effect of

W2)ying th6 yCar

Accesa to treatment and rehabilitation. The impact of
varying the assumptions about access to treatment
and the availability of rehabilitation in both develop­
ing and industrialized countries are assessed. Fig: 2.
compares the base case (33%) with the overall im­
pact of a reduction in treatment and rehabilitation
rates to zero percent in developing world children.
Little impact is seen on the shape, slope or break- .
even year. At zero percent of the developing world’s
children and 75% of the industrialized world's .
children reaching treatment and receiving rehabilita. tion, it is.still cost-beneficial to eradicate poliomyeli­
tis, The minimaWteatment and rehabilitation rate'
at which eradication leads to savings is 0% access
in developing countries and only 60% of children
in industrialized countries receiving treatment and
rehabilitation.

Vaccine wastage. The effect of vaccine wastage on
net costs and net benefits is tested by increasing the
vaccine wastage rate from the base case 33% to 50%.

Sensitivity analysis

n

i

When alternative assumptions were tested, the
eradication of poliomyelitis was still shown to be
cost-beneficial. Different assumptions about the dis­
count rate, the proportion of cases receiving treat­
ment and rehabilitation, vaccine wastage, the age
and proportion of the target population vaccinated
during accelerated activities, and the cost of deliver­
ing immunization services may modify the results of
the base case, but do not significantly alter them.

Target age of accelerated activities. Alternative tar­
get ages of accelerated immunization activities were
evaluated. The base case assumes the target age is
*59 months. The shortage of resources to purchase
vaccine has the potential to force a lowering of the
target age group for accelerated activities as was
the case in China in 1992. For all target age groups
eradication is cost-beneficial. Each year of lowering
of age group in the target population to be vac­
cinated decreases the year of breaking even by one
year: for the base case (<59 months) the break even
year is 2007; for <48 months it is 2006, and for <36
months it is 2005.

Fig. 2. Global poliomyelitis eradication: varying treat­
ment rate assumptions In developing and Industrial­
ized countries. Base case: in developing countries, 33%
of acute poliomyelitis patients received care and rehabili­
tation; in developed countries, 100%. Sensitivity analyses:
0%/100% — no acute poliomyelitis patients receive care
In developing countries, while 100% do sb In Industrialized
countries; 0%/75% — no acute poliomyelitis patients re-'
celve care or rehabilitation In developing countries, while
75% do 80 In industrialized countries. The year in paren­
theses Is the break-even year.
16

Basa case: 33% and 100%^-

(2°07)

S 12

i
8

(2008)

9

6

0V75%

(2022)

3

8 0

I -3
-€

Cost of vaccine and delivery. The effect of an in­
crease in OPV price from US$0.08 to US$0.12 has
no significant impact on the year of breaking even or
the overall slope of the curve (Table 2). Increasing
the coeta of delivering OPV by nearly eightfold in

-9
-12
1980

1990

2000

While this increase delays the time when net costs
are exceeded by net benefits from the year 2007 to
2011, eradication remains cost-beneficial.

^rca^*n® cvcn bytwo yean

2010

Y«ar

2020

2030

L
2040

Proportion of the target population reached during
accelerated activities. Reducing the population cov­
ered by mop-up and outbreak response from 1% to
0.1% had no significant effect on whether eradica­
tion or the year of breaking even is cost-beneficial, as
compared to the base case (Table 2).

treatment and rehabilitation plus vaccine and de­
livery costs. When the combined costs are tested,
beginning with the level of 2 cases per 100000, an
accelerated programme is cost-beneficial, and moves
the break-even point to the year 2026. Poliomyelitis
eradication, whether modelled from pre-eradication
to eradication or as an acceleration from rou­
tine immunizations after the substantial initial
impact,is cost-beneficial and saves USS 3380 million.
(Table 2).;.
i



Discount rate. The effect of alternative discount
rates is shown in Table 2. At a 3% discount rate, the
break-even point is two years earlier as compared to
the base case at 6% in the year 2007. Discount rates
of 0% and 10% demonstrate a family of curves, all of
which are dost-beneficial with break-even points between 2004 and 2017. The fate of 3% is more commonly used for social sector programmes such as
poliomyelitis eradication. At 3% the net savings will
be USS '34 500 million by the year 2040.
Additional models

Addition of the cost of vaccine and delivery to the cost
of treatment and rehabilitation as benefits. The an­
nual global cost of routine vaccination with OPV
under the base-case assumptions, which includes the
cost of vaccine, delivery and wastage, is US$1774
million. If the costs of routine vaccination (vaccine
and delivery costs) are added to the cost of treatment
and rehabilitation as benefits accrued from eradica(ion, the year of breaking even is 2005, two years
earlier than for the base case (Table 2), and precedes
■tthe year of the declaration of eradication, which
makes the eradication even more cost-beneficial.
Marginal additional costs of accelerating from control
to eradication. After the World Health Assembly
resolution (WHA41.28) had established the goal of
global eradication of poliomyelitis, there was con­
cern about the diversion of resources from primary
health care development and other priority dis­
ease control activities towards eradication activities.
Roiitinp administration of OPV had reduced cases
from an estimated 5/100000 to 2/100000 in 1988,
the year of this resolution. This level would be ex­
pected to be maintained if there was no acceleration
of activities towards eradication. Reducing cases
from 2/100000 to zero represents the marginal addi­
tional cost of moving from control to eradication.
The base case is compared with the acceleration of
the programme beginning in 1988. Similar to the
base case model, the benefits accrued were modelled
in terms of treatment and rehabilitation alone, and

Discussion

i

~

In spite of systematic underestimation of the benefits
of aa poliomyelitis
poliomyelitis eradication
eradication programme,
programme, there
there isis
of
evidence of
of positive
positive and
and high
high returns
returns from
from such
such an
an
evidence
investment; the base case demonstrates
that the
______
__netI
1benefits

exceedJ *the
c net costs. of the programme only
two years after eradication is declared. By the year
2040, the savings will be US$ 13600 million. Polio­
myelitis eradication, through sensitivity analysis and
under the most stringently unfavourable test condi­
tions, is economically beneficial, and the break-even
point is always close to the date of eradication. This
is true under various assumptions — increase in cost
of vaccine and delivery, high vaccine wastage, lim­
ited access to treatment and rehabilitation, limited
target age groups for accelerated activities, and high
discount rates. The world would therefore not have
to wait many years for eradication to pay off.
The eradication of poliomyelitis is a justifiable
investmgiiLev^n-without making any allowance for
Savings (benefilsJ otKerihandhose due to real reductidnJ in expenditures to treat and rehabilitate some
of the victims of the disease. Morbidity in the form
of a post-poliomyelitis handicap affects a child’s
activity throughout life. This loss is associated with
both real and intangible costs (missed work, un­
employment, family loss of work time and income,
reduction in anxiety, pain and the social stigma of
handicaps). The cost of treating even a small fraction
of those who need treatment and rehabilitation is
large enough to pay for the total prevention of polio­
myelitis. The addition of vaccine and delivery costs,
which will cease after eradication, makes the tan­
gible saving even larger — an additional annual sav­
ings of US$ 1700 million.
Experience with poliomyelitis epidemiology
and with elimination of poliomyelitis in the Ameri­
cas has demonstrated that eradication is technically
feasible. Like smallpox, humans are the only reser­
voir of the virus, and there are no long-term carriers.
An effective and inexpensive vaccine is readily avail­
able. While the ease of poliovirus transmission and
the high infection-to-case ratio would suggest that
the virus potentially is difficult to contain, little evi-

~

I

CLINICAL ECONOMICS

MODULE 6
COST-UTILITY ANALYSIS
h

After completing this module, you should understand:

r

a.

how utility values can be incorporated into
cost-effectiveness analysis;

b.

how the resulting cost-utility analysis differs
from cost-effectiveness analysis;

c.

the strengths and weaknesses of cost-utility
analysis.

1

COST-UTILITY ANALYSIS,

introduction
Module 5 that cost-effectiveness analysis
recall
from
alternative interventions
You will
to compare either
(CEA)
can be used
or condition, or alternative programmes
for the same disease
In the latter case, the
saving lives,
are
the number of lives
which aim at extending or
indicators of effectiveness
most common
gained.
saved and the number of life-years
to save lives but
interventions do not aim purely
life
as
well.
Other
Many health
quality of
the
seek
to
improve
for some types of cancer,
such as chemotherapy
interventions,
It is not
quality
of life.
life but reduce the extensions in the quantity

of
may extend
to measure benefits by
appropriate
Some way of combining quantity
situations.
life
in these
must be
(reduced morbidity)
and quality
(reduced mortali ty)
seeks
to
do
this.
Cost-utility analysis (CUA)
f ound.

COST-UTILITY ANALYSIS
first step is to „est imate the
CUA is based on CEA,
and
The extra xears
number of life-years gained by an intervention^
for changes in the quality
of “life are then adjusted to account benefit
is called a quality
of life.
The CUA indicator
adjusted life-year (QALY) or a well-year.
set of weights called
is based on a
The quality adjustment
i
economists use to describe the benefits ^people
utilities,
a term
from consuming a good or utilizing a service.
consider they get
associated
ty“ or^fating

j an
Each possible health state has

»
compared
to other
desirability of that state
which reflects the <-----is used where 0
of 0-1
Usually a scale
possible> states.
(More complicated scales
represents death and 1 is good health.
to be Jess desirable than death
' “ j some health states
which allow
also ex 1st.)
possible states of health be
Thus,
CUA requires that the
utilities determined.
For
the corresponding
described and
Office
of
Technology
Assessment

the US Congress
example,
which may result from an influenza
f our r outcomes
described
were death, sick in bed, sick but mobile, and
They
epidemic.
Loritea the uti U t tee, ot .0 , 0 ■ V
They
healthy,
respectively,
which implied that
these states
day’s health, for example.
gave 40% of the satisfaction of


_

1 _

__ J

z-v

I T" Fl

'■ which will
these data to a hypothetical project
Let us apply
Epidemiological
healthy male against flu.
f.
vaccinate a single
who
contract
flu are
unvaccinated
healthy
men
i
----tha t
data reveal

2
conf ined
to
bed for 1 week (1/52 of a year), and are mobile but
sick
for a
further week.
They then return to full health.
If
we assume
that
the vaccine is 100% effective, the man will not
contract
flu but will
be
in good health all the time.
His
quality of
life
improves.
However, no extra life-years have
been gained,
so the benefits of the intervention would not have
been captured
in a CEA.
The patient gains [( 1/52 ) ( 1.00) (l_/52) (0.4)] + [ ( 1/52) ( 1 .00) - ( 1/52)(0.6) ] = 1/52 QALYs.

DESCRIBING HEALTH STATES

y->-

A group' of academics from McMaster University have developed a
system which can be used to classify health states for a wide
variety of diseases and conditions-.
They define health as a
function of four attributes - physical, activity, "level of selfcare ,-s social-emotional wellbeing,
and the nature_of the health
problem.
r''Earch attribute has a number of levels, and a person’s
state of health at a particular time can be identified with one
of the 960 possible health states defined by the system.
Drummond et aJ
describe—the system" Tn more detail.
(Anderson
and Moser base
their analysis on an even more complicated
system.)

To use the McMaster approach it. is necessary to obtain utilities
for each of the 960 states of health.
This generally involves
assigning utilities
to each level of the four attributes, then
mathematically modelling the way the attributes combine to form
a single utility
for the state of health.
The advantage_pf
these systems
is
that they are applicable, to a Jarge number -o-f—.
d i s eases' and conditions,
and are sens i_tiye t° small changes in
the quality of life.

A less complicated approach
is
to define a limi_ted number of
states of health whi ch are specific to a particular condition.
Each
state is then des "ibed in a way which includes information
about how a patient
f unctions
in
terms of each of the. four
attributes descri bed above.
This method is simpler to use than
the
larger system, but is less sensitive to small changes in the
quality of life,
Moreover,
it
is
less easy to compare the
ef f i ciency of
interventions aimed at different diseases
if
disease speci f ic QALYs have been es Limated.
This problem is
discussed in more detail in the reference by Donaldson et al.
ESTIMATING UTILITIES

I

Three methods of deriving utilities can be used.
The first
relies on
the judgement of either
the analyst or a group of
ei ther
experts.
\vThe second uses weights from the literature, while the
third estimates utilities directly fromi a sample Qf_sxibjects .

In the third case ,
the analyst must identify the appropriate
subjects.
In
theory
the sample
sh ^uld be representative of
society because CUA
is generally undertaken from society’s

c

3

1 '..
However, members of the public can be unaware of the
viewpoint
ramifications
of different states of health, so researchers
full
resort to taking samples
from health professionals or
often
Drummond et al argue that generally this__does not
patients.
the
different---groups
d to rate
states of health in
matter as
-- ---------*ten
__________
_________
a similar manner although
there is some evidence to the contrary
j------(Lomes & McKenzie).
i

sample has been selected, the analyst asks a series ...of
Once the
designed
to
reveal the preferences the subjects have
questions
Three systems of doing „this
different
states
of
health.
f or
____
___
___
?, sdt.jandar d_ gamb 1 e and -time
been
developed
the
rati
ng
scale,
have
studies
have
shown that -—
they
result
trade-off.
Again some i-------_________________
---—— in
similar ratings
for the same states of health vhile .others_have
shown large differences.

descript ion of the way utilities can be estimated is
A full
scope of this module which is designed to help you
beyond the
Should you wish to
the rationale behind CUA.
understand
you
will
need
to
read
the literature more
estimate utilities,
It is
extensively,
and
Drummond
et
al
is
a
good
place
to begin,
extensively
also recommended that you seek the advice of an economist.

COSTS
The CUA
Costs are measured in the same way in .both CEA and .CUA.
Like
is the cost per QALY (or cost per well year),
indicator
By
CEA,
CUA must be used to compare alternative interventions.
itself, a CU ratio rarely prdvTaies"useful information.

UTILITIES AND CLINICAL DECISIONS

' » are willing to sacrifice
people
An assumption behind CUA is that
to gain extra quality of life.
If this is true,
life
of
years
can also be used as an aid to clinical decision making
utilities
from patients
before
and clinicians could e’icit utilities
the
appropriate
treatment.
This
would
be
on
deciding
for treatments which extend life but result
particularly useful
The
in serious side effects which reduce the quality of life.
McNeil
et
al
provides
some
justification
for
this
article by
Formal methods of incorporating utilities into
suggestion.
clinical decision making are considered later Module 9.

PROBLEMS
it
is used to guide clinical
CUA is controversial whether
^individual
patients
or as a basis for resource
for
decisions
One
practical
problem
is that utilities elicited
allocation.
state
of
health
can
vary
with the way the question
for a given
A
more
theoretical
difficulty
is the implicit
is framed,
that
people

s
willingness
to
sacrifice
years of life
assumption
given
improvement
in
health
status
is
constant.
This
for a
not
be
true
if
people
valued
good
health
more
highly
at
would
times
of
their
lives
(perhaps
while
raising
children),
di f f erent

4

if they they had a rate of time preference for years of life in
a given state of health,
or if they were less willing to
sacri f ice quantity of
life
if they believed they had only a
short
time to
live.
McNeil et al, for example, found that
patients were willing
to trade years for improved qual.Lty_Q.nly
if they expected to live more than five .years.
Another problem
years.
concerns
way
the
QALYs are aggregated
in evaluating an
intervention.
A <QALY gained by a 70 year old person is given
same
weight
as
;a QALY gained by a 2C year old, but it is not
the
society
clear that
wishes
to make cecisions
about resource
allocations on this basis.
The ]problems with CUA are widely recognised and are discussed in
Loomes & McKenz ie
(a difficult but worthwhile .reference).....,.. Two
conflicting conclusions
have been drawn.
The first is that CUA
is
flawed and should not be used.
The second is that it is at
least as good as other methods of resource allocation which do
not take quality into account and should be given an extended
trial.
REFERENCES

Major References:

Anderson J.P.
& R.J.
Moser, "Parasite screening and treatment
among Indochinese refugees",
JAMA 1985, 253:2229-2235.

Drummond et al,

(see Module 4), chapter 6.

Other References:

Donaldson
of Health
Weinstein
issue, pp.

C.
et al ,
’"Should
* *
QALYs be programme-specific?", Jnl
Economics 1988, 7 ( 3 ) : 239-259.
See also the comment by
”A QALY
is a QALY is a QALY
or is it?" in the same
289-290.

Hatziandreu E.I.
et al,
"A cost-effectiveness analysis of
exercise as a health promotion activity", American Jnl of Public
Health 1988, 78( 1 1 ):1417-14 2 1 .
Loomes G.
& L.
McKenzie,
"The use of QALYs in health care
decision making", Social Science and Medicine 1989, 28(4) :
209-308.
McNei1
B. J. ,
R.
Weichselbaum & S.G.
Pauker,
"Speech and
survival:
tradeoff s between quality and quantity of life in
laryngeal cancer", NEJM, 1981, 305:982-987.

Torrance G.W.
"Measurement of health
state utilities
economic appraisal", Jnl Health Economics 1986, 5:1-30.

for

C’-

5
QUESTIONS

Read the article by Anderson and Moser.
Estimate a CE ratio in the way that you learned in Module
5 using the data contained in the article.
(Define, your own
indicator of effectiveness).
What are the relative merits of using CUA and CEA in this
b.
case?
Discuss the use of discounting in the article.
c.
Consider Table 4 and the following questions:
HINT:
In calculating column D, what assumption is made
about when a life is saved?
Is the same assumption implicit in calculating column F?
ii .
In calculating column I, what assumption is made about
iii .
when the savings (H) occur?
Can you detect any problems with the way that utilities
d.
were measured?
1.
a.

2.
What are the advantages and disadvantages of using subjects
from the following groups to determine utilities?
a. patients,
b. doctors,
c. members of the general public.
3.
Consider the following table which adjusts the results of a
number of studies by different authors to costs per QALY in 1983
US dollars.
The
table
is adapted
from Torrance G.W. & A.
Zipursky,
’’Cost effectiveness of antepartum prevention "in Rh
immunization”, Clinics in Perinatology 1984, 11(2):267-281 .
tsy Yi

' PROGRAM

--- r—

_.,.r

PKU screening

antepartum anti-D

negative

1220

coronary bypass surgery:
- left main disease
- single vessel disease (moderately
severe angina)

36300

treatment of severe hypertension (diastolic
above 104 mm Hg) in males aged 40

19100

continuous ambulatory peritoneal dialysis

47100

hospital haemodialysis

54000

a.

b.

r

COST/QALY

4200

How do you think such tables could be used to evaluate the
efficiency of a particular programme?
Can you see any problems with using such tables?

Media
8367.pdf

Position: 1183 (7 views)