6906.pdf

Media

extracted text
I
I

I

THE HEALTH IMPACT OF THE BHOPAL DISASTER :

AN EPIDEMIOLOGICAL PERSPECTIVE

I
BY
THELMA NARAYAN

I

I

Dissertation submitted in partial fulfillment of the
requirements for the MSc degree in Epidemiology.

London School of Hygiene and Tropical Medicine.
Dept, of Epidemiology,
University of London.
SEPT.1987

•r

Dedicated to the people o-f Bhopal who have suffered the
consequences o-f the disaster and to the people -from many walks of

life who have responded to this great human tragedy.

" All scientific work is incomplete- whether it be
observational or experimental. All scientific work is liable to

be upset or modified by advancing knowledge. That does not confer

upon us a freedom to ignore the knowledge we already have, or to
postpone the action that it appears to demand at a given time. "
A.B.HILL.

>*•

CONTENTS

face No.
ACKNOWLEDGEMENTS

Chap. 1
Chao.

'"J

2. 1
n n

I
i

!

INTRODUCTION

1

OBJECTIVES
Objectives
General comments
Sources of information

4
4
C7

c

Chap. 7
3.1
3.2
3.3
3.4

THE DISASTER - A BACKGROUND
The event
The UCIL plant
The manufacturing process
Carbaryl

6
6
6
7

Chap. 4
4. 1
4.2

THE AGENT
Chain of events

Methyl Isocyanate - chemical properties

8
8
10

Chap. 5

IMPACT ON THE BIOLOGICAL ENVIRONMENT

12

Chap. 6
6. 1
6.2
6.2. 1
6.2.2
6.2.3
6.2.4

THE HEALTH IMPACT OF THE DISASTER
The need for an epidemiological perspective
Descriptive epidemiology
Li mitati ons
Time
Place
Person

14
14
15
15
17
19
21

Chap. 7
7. 1
7.2

MORTALITY

Exposure related mortality rates
Demographic characteristics of the dead

Zu

Chap. 8
8. 1
8.2
8.3
8.4

AUTOPSY FINDINGS
The first week
The second week
The third week
Histopathol ogy

33

Chap. 9
9. 1
9.2
9.2. 1
9.2.2
9.2.3
9.2.4
9.2.5
9.2.6
9.2.7
9.2.8
9.3
9.4

MORBIDITY - CLINICAL FINDINGS
Ger eral comments
Acute and subacute phases
The eyes
The respiratory system
Haematology
The gastrointestinal system
The neurological system
Psychological disorders
Outcome of pregnancy
The reproductive system
Therapeutic trials
Discussion on clinical findings

36
36
37
37
38
42
42
42
43
44
45
45
49

Chap. 10
10. 1

REVIEW OF I-'IDEMIOLOGICAL STUDIES
Eye and general morbidity

52

'~icr
ZU
IT

30

34
35

cn

uz.

10.2
10.3
10.4
10.5

Chap. 11
11.1
11.2

General health status
Womens reproductive health
Outcome ot pregnancy
Watering o4 the eyes

54
61
63
66

REVIEW OF EXPERIMEN7AL STUDIES
disaster
Pre
disaster
Post

68
68
68

DISCUSSION
Summary ot findings
Outline of discussion
The exposure variable
The population at risk
The health outcome - mortality
The health outcome - morbidity
Confounding and interactive variables
Sources of bias
Suggestions for epidemiological studies
in Bhopal
12.9.1 Cross-sectional studies
12.9.2 Longitudinal study
12.9.3 Serial cross-sectional studies
12.9.4 Epidemiological Monitoring Unit

Chan. 12
12. 1
12.2
12.3
12.4
12.5
12.6
12.7
12.8
12.9

Chap. 13

i

I

Ii

1

73
74
74
78
78
79
80
80
83
84
84
88
89

CONCLUSION

90

REFERENCES

92

TABLES / ILLUSTRATIONS
Map of affected areas
Exposure related mortality rates
Symptom profile
Symptom profile - contd.
Spontaneous abortion rates
Spontaneous abortion rates - contd.
Foetal oeath ratio
I.

■7 *7

Epidemiological

studies - a summary

20
28
59
60
64
65
65
67

ACKNOWLEDGEMENTS

Many thanks to :
- the staff of the Dept, of Epidemiology, London School of
Hygiene and Tropical Medicine,for sharing their
epidemiological skills,
- Paul Elliot, my tutor, Neil Andersson and Andy Salmon of the
Dept, of Occupational Health, LSHTM, for useful discussions,

- all my colleagues on the course for very stimulating
i nteracti on,
- members of the Medico Friend Circle, India, for constructive
and critical debate,
- my husband, Ravi, for many discussions and much support.
- and my son, Lalit, for putting up with it all.

I

&

CHAPTER 1
INTRODUCTION

The world's most serious disaster in the chemical industry

which occured at Bhopal, is a tragic instance o-f the adverse impact o-f
chemical agents on human health. The health profession has
historically more experience and expertise in dealing with biological

and physical agents of disease. Experience with chemical agents was
previously limited to small groups of workers in specialised
occupations. Health problems arising therein were dealt with by

specialised occupational health teams. With rapid growth of the

I

chemical industry, there has been an increasing use of a variety of
chemicals in activities of daily life as well as in agriculture.

During the past decade there has been a growing realization that the

I

adverse impact due to human interaction with chemical agents is now
spreading beyond the confines of the workplace, to the environment and

(

the public in general. Incidents at Bhopal (methyl isocyanate + ?),

I

Seveso (dioxin), Mexico (butane) and Vietnam (Agent Orange) have

*

events are often considered unfortunate, uncommon, freak events or

demonstrated dramatic instances of these adverse effects. These

accidents. However, they represent the tail end of the distribution of

3

the more common, less serious, leaks, injuries and minor health

effects resulting from exposure to chemicals. The long-term effects of

ft
|

lower dose and/or chronic exposure to several chemicals in use are at

fl

present unknown. However there is

accumulating evidence of the

adverse effects of some of them eg. the impact of pesticides on the

3
1 I
I
i

environment and their effects on animals and man through the food
chai n.

I
1

The increasing use of potentially harmful chemicals, the

conditions in which they are manufactured, the health and safety
regulations adopted and implemented are closely linked to economic and
political -factors operating at a national and international level.

People living and working in conditions o-f material poverty bear the
brunt of the most adverse effects, because for them employment under
any conditions is a necessity tor survival. The poor also carry a

double burden - suffering from the diseases of poverty viz.
malnutrition and infections along with the modern diseases of
industrialization. A balance sheet approach of costs and benefits to

people has to be used in evaluating the role of chemicals in this

context. The following quotation by Albert Einstein is relevant and

must be kept in mind,
" Concern for man himself and his safety must always form the chief

interest of all technical endeavours. Never forget this in the midst
of your diagrams and equations".

I

Responsibility for the safety of workers and communities and

the prevention of technological disasters requires the involvement of
professionals and decision makers from several fields. More

importantly it calls for committment

and political will by wider

forces in society. Public health physicians could contribute by using
epidemiologic skills to study the health effects of these agents. This
would help provide a firm basis of knowledge about possible effects on

health, which could then be used for creating awareness and for policy
making. This has already been recognised and interest in environmental
epidemiology, which includes the above group of agents has been

growing. Being a new area of enquiry, the limitations of the existing

tools of research, have also to be kept in mind.
The chemical accident at Bhopal has been an experience of a

public health emergency caused by a technological disaster. The World

2

Health Organization has defined a disaster as " an event that suddenly

overwhelms the capacity of the normal system to respond” (Wasserman,
1985). Though used to describe natural disasters like earthquakes.

■Floods and volcanoes the definition could also apply to technological

di saste'*s.

In Bhopal, the disaster

overwhelmed the capacity of

individuals, physically and psychologically. And because of the
magnitude of the numbers involved and its complexities it also
overwhelmed the capacity of the macrosystem le. the social and

administrative structures to respond.

Technological disasters are

very complex and require specialist intervention. Besides the

possibility of causing external bodily harm, they may also act at a

cellular and biochemical level causing a disruption in

physi ologi cal

functioning leading to unknown pathological states.
Exposure of the population to toxic chemical vapours during the
Bhopal disaster has resulted broadly in three adverse outcomes :

mortality, morbidity and disability. An epidemiological perspective
and method of study is vital to understanding the pattern and
distribution of the three adverse outcomes in the community.

1i
I

I

Epidemiological studies and population profiles can provide data which

could be used for various objectives viz.
a) to substantiate/support, disprove or provide clues for
aetiological hypothesis.

b) following

from the above, to

help indirectly in deciding

upon rational therapeutic interventions and in initiating secondary
and tertiary preventive measures where possible.

c) to plan health services, including rehabilitation, for the
affected people.

d) to provide supportive evidence in court, regarding extent of
injury, in claims of compensation for the victims.

3

I
1
:HAP

OBJECTIVES

2. 1

A)

to review the available literature concerning the health

impact of the Bhopal disaster viz.

a brief review of the event itself,

mortality -following the exposure,
morbidity -following the exposure.

The latter two would include clinical observations and

I

I
1 I
I
^9'

The objectives o4 this study are :

findings reported by medical professionals and the findings of a few
epidemiological studies that were conducted.
B)

to use an epidemiological perspective in the appraisal of the

above information.
C)

to review literature regarding the chemistry and toxicology

of the "agent” ie the chemical substances that were released during
the event which produced the adverse health effects.

D)

based on all the above information, to discuss methodological

issues relevant to an epidemiological approach to studying and

understanding the adverse health outcomes following exposure to the

toxic chemicals in Bhopal.

4

2. 2

(jeneral cowngntt :

In the context o4 an epidemiological perspective of the health
impact, one must initially consider all available information in broad

categories of the epidemiologic triad of agent, host and environment
and also in terms of time, place and person. In the Bhopal situation,
there were and still are several limitations in attempting this, si nee
there are big “knowledge gaps” about the chemical agent/s of exposure
and also about the possible outcomes on human health. These will be

considered in greater detail later.
The situation in Bhopal is also complex in other respects. The

incident has enormous medico-legal implications. There has been a
delay in publication of results of ongoing studies. It has therefore

not been easy to get adequate published references from standard
scientific journals. For the purpose of this study a wide variety of
secondary sources of information have been used to build up an

epidemiological profile.

n
t
x. ■

Sources of information psed
These could be broadly classified as follows :

a) Articles published in journals. Many of these are toxicological

studies and a few are reports of epidemiological studies.
b) Reports of the Indian Council of Medical Research (ICMR) that

were available in the first 6 months following the disaster. Some of
these are minutes of meetings at which many of the researchers

presented their interim findings. Furthur reports / results of studies
I

conducted are not available.
c) Reports of studies and investigations carried out by research
teams not linked with the ICMR.

t

CHAPTER 3

THE DISASTER - A BACKGROUND

I
3. 1

The event

The ’’Bhopal disaster" as it has come to be known, was a major

I
I

leakage o4 toxic chemicals from a pesticide plant in the city o-f
Bhopal on the night of 2/3 December 1904. The plant. Union Carbide

India Ltd (UCIL), was the Indian subsidiary of a

multinati onal

chemical company. Union Carbide Corporation Ltd.(UCC), with

headquarters in Danbury, Connecticut, USA. The leak occurred from a

I

tank containing methyl isocyanate (MIC). It resulted in death and

injury to human, as well as to plant and animal life.
3.2

The UCIL plant

The plant was set up in 1969 and gradually expanded over the

years into a large manufacturing facility for the formulation of

i

carbaryl, an insecticide, commercially known as Sevin. Till 1979,
methyl isocyanate (MIC) and alpha naphthol, intermediates used in the

manufacture of carbaryl, were imported from the parent company.
1

Subsequently, from 1980, MIC was manufactured locally with knowhow and
basic design supplied by UCC.
?■

Bp.

t
Iw

T

T

The manufactur;nq process

Very briefly the process for the manufacture of carbaryl i s as
follows : carbon monoxide is made to react with chlorine to yield

phosgene which is then made to react with monomethylamine to produce
methyl carbamoyl chloride. This is pyrolyseo to yield methyl
isocyanate and hydrogen chloride. Chloroform is used as a solvent and

caustic lye for the neutralization of any toxic material. Finally MIC
is made to react with alpha naphthol, in the presence of a catalyst
and carbon tetrachloride as a solvent, to produce carbaryl. Different

I

6

I

concentrations of this are used to formulate the end products.
3.4

Carbary)
Chemically known as 1-naphthalenol methyl carbamate, carbaryl

i s an insecticide which is

moderately hazardous” accproing to the

WHO classification of pesticides.

In 1976, the Environmental

Protection Agency of the U.S.A placed it on the list of compounds that

are " hazardous to man and environment”. Carbaryl has been found to be
toxic in some animal studies (Phoon,1986).
Alternative processes not requiring MIC, are available for

the manufacture of carbaryl. They are in use by other companies who

wanted to avoid the use of MIC, which was known to be hazardous
chemical. However, they are more expensive and also produce larger

volumes of waste material.
The MIC plant at Bhopal has had a track record of previous

spills and accidents ever since its inception. In late 1981, one
worker was killed and 3 injured from exposure to phosgene.

In 1982, 18

workers were exposed to a mixture of three chemicals including MIC
during a leak. The issue regarding inadequate safety

measures

both

in design and management practices - had been raised by the workers

uni on, a local journalist, a safety commission from the US and even in
the local legislative assembly (Agarwal, 1986 and Weir, 1986). However

no action was taken.

7

I
CHAPTER 4
THE AGENT

4. 1

Chain of events
The leakage occurred from an underground, stainless steel

isocyanate 'MIO had been
tank, in which 42 tonnes ot liquid methyl
stainless steel tank was
stored -from Oct 1984. Post disaster, the huge
Part of the tank
found to have bulged at the end and between the ribs.
had risen out of the ground fracturing the concrete in which it was

embedded. All this is evidence of a tremendous build up of pressure
and temperature inside the tank - with temperatures possibly as high

as 250 to 350oC and pressures upto 11-13 kg/cm2 (Varadarajan, IS’Bu) .
Knowledge of what triggered the explosive "runaway" reaction,

the type of chemical reaction that occurred and what products were
formed is partly speculative.
and Industrial Research
The Indian Council of Scientific

chemical
(CSIR) have produced a detailed report about the possible
reactions that could have occurred. This is based on -

existing though limited knowledge
an analysis of contents of the tank;

of the chemistry of the substances involved and on laboratory
experiments.

Some of the factors that contributed to the chain of events
that took place are as follows (Varadarajan, 1985):
- the inherent chemical properties of MIC, especially its high
reactivity and volatility,

which will be described later.

the storage of large quantities of MIC in very large size

containers for long periods of time.

-insufficient caution in design, choice of materials for
construction, and in the provision of measuring and alarm instruments.
on quality of stored
- inadequate controls on systems of storage and

8

materials.
- inadequate guidelines and practices in operations and maintenance.
The probable chain of events that took place was :

- the entry of water and the presence ot catalysts initiated an
explosive, runaway, trimerization reaction with a rapid rise of
temperature. A large portion of the stored liquid MIC vapourised with

a rapid rise of pressure.
- carbon dioxide produced by the reaction of hydrolysis caused a

I

mixing of contents in the tank leading to rapid chemical reactions.

- secondary chemical trans-formations occurred at the high
temperatures reached, leading to the formation of a complex mixture of

chemical substances. The following 12 products were found as residue
in the tank : methyl isocyanate trimer, dimethyl isocyanurate,

dimethyl urea, trimethyl urea, dione, trimethyl biuret,
monomethylamine, di and trimethylamine, chloride and metallic iron.

A description of the properties of MIC only will be given, for
i

the purposes of this report, to indicate the hazardous nature of the

chemicals involved. MIC was probably present in large quantities i n
the cloud of toxic gas and it has also been the focus of most of the

research effort. Hence some information about its chemical properties

and toxicology is available. However it must be kept in mind that
other chemicals could also be playing a causative role in the

morbidity manifesting in the affected population. This could occur due
to:

other chemicals having been formed during the runaway reaction

itself; and/or by the breakdown products of MIC in contact with the
atmosphere; and/or

by the reaction/biotransformation products of MIC

in the human system.

I

9

4.2

METHYL ISOCYANATE (MIC),

isocyanic acid-methyl ester, CH3-N=C=0,

Chemical propgrtiy* :
MIC is mainly used as an intermediate in the production ot

Carbamate insecticides and herbicides eg. carbaryl (Sevin), aldicarb
(Temik), carbofuran (Furadan) and propoxur

(Baygon). MIC has a known

hazard potential because of its ready volatility, low boiling point.
■flammability and high reactivity. Traces of metallic impurities act as
catalysts and stimulate a violent exothermic polymerization. The heat

generated can volatilize a large proportion of MIC. The gas has a very

high inhalation toxicity. Hence extreme precaut-ons are necessary in
the production and

handling of MIC

MIC has a low boiling point of 39.IOC at 760 mm Hg. It is
slightly lighter than water, but the gaseous form is heavier than air.

Its vapour density is 1.97 as compared to 1.0 for ai r. It has a high

vapour pressure of 348 mm Hg at 20oC. Because of these properties MIC
vapour will be concentrated near ground level and have a greater
impact on people, plants and animals.
The Mol.wt.

is 57.05 and Sp.gravity 0.96 at 20oC-

Explosive limits :5.3 7. and 26 7. by volume air.
It is an extremely reactive substance. In the presence of water,
hydrolysis occurs and monomethylamine (MMA) is produced ; one mole of
MMA then reacts with another mole of MIC to form 1,3-dimethyl urea

(DMU) with the generation of heat, or 1,3,5-trimethyl biuret (TMB) if
there is excess MIC. At 73 7. relative humidity MIC can decompose into

monomethyl amine.
Isocyanates in general react with carboxylic acid to generate a
peptide bond with the liberation of C02. They are also known for

carbamyl ation at the biochemical level.

MIC reacts with itself and is known to poly erize on storage

10

for long periods.In the presence of a large number of metals and other

catalysts a violent exothermic polymerization reaction occurs.

It also

reacts with hydroxyl compounds, amines and acids.
On heating and burning it generates several toxic decomposition

products viz. nitrogen oxides, hydrocyanic acid and carbon monoxide.

The ACGIH (American Committee ot Government Industrial
Hygienists) have set the Threshold Limit Value (TLV) tor MIC to be

0.02 ppm (0.05 mg/m3) tor skin. The OSHA (Occupational Safety and

Health Administration) Permissable Exposure Limit (PEL)is 0.02 ppm.
Salmon et al

(1985) found that MIC vapour reacts only

slowly with water vapour. Hence substantial amounts of MIC would be
present in a vapour cloud released even into moist air. This would

allow time for MIC to react with its own hydrolysis products.

In air MIC hydrolyses to C02 and a relatively inert amine.

It

reacts in sunlight to form alkanes, nitrogen and carbon monoxide.

If

moisture is present it converts to MMA, DMU and TMB. With moist soil

it forms methylamine which strongly adsorbs to the soil surface until

degraded.
On contact with plant, animal and human tissues MIC has severe
corrosive or irritant action.

It is thought to compete or interfere

with C02 during photosynthesis .

In animals and humans it is thought to

transform to methylamine, C02 and other carbamyl derivatives.

Thus, the chemical composition of the agents causing the
adverse health effects in Bhopal are not yet completely known, nor

their mechanism of action fully understood. More work needs to be done
regarding breakdown products of MIC in conditions of high temperature

and pressure and of those formed by hydrolysis. Their properties and

immediate and long term toxic effects need to be studied. Experimental
studies of animal exposure to MIC are described in a later section.

Epidemiologic data could also provide clues regarding etiology.

11

I

CHAPTER 5
IMPACT ON THE BIOLOGICAL ENVIRONMENT

Ine toxic gas had an impact on anima] and plant life as well as

on soil and water. Most data quoted in this section is from the report

by Agarwal 0986)

Official sources put the number of dead animals

cattle.

goats, sheep etc at 1,047, while about 7,000 received therapeutic
care.

An investigative team from a research institution found that
animals had died within 3 minutes of inhaling the gas. They were
frothing from the mouth, lacrimating and breathless. Many cows

mi scarried. In clinically ill animals, there was a drying of milk

after the exposure. Milk production came down from 8 - 10 kg per day
to 0 - 0.5 kg per day. Poultry were relatively less affected for
(

unexplained reasons. Many birds escaped death as well.
The effect of MIC on plants and soil was studied by the Central

Board for the Prevention and Control of Water Pollution and the Indian
Council of Agricultural Research. The "neem" tree commonly growing in
the area was found to be a sensitive atmospheric indicator. A

vegetation damage contour map was drawn. Vegetation in an area of 3.5
severely affected, 10.5 sq km beyond

sq km around the -factory was

that was badly affected, a further 6 sq km moderately and 5 sq km

I

mildly affected. Leaves bore the brunt of the dc-mage. Scientists found
that there had been instant deafh in exposed tissues. Another study by
the Benaras Hindu University found genetic defects in locally grown

vegetables. The same species of plants which were otherwise badly
affected were unharmed when found growing near the lake. Plants
submerged in water were also unaffected.

Some clues for human epidemiological research from the above

12

account would be to study the effects on lactation in mothers, as 11

is ot vital importance to the nutrition and health ot intents. ano to
check mortality and morbidity near the lakes and beyond. Plant studies

also confirm the gradient in damage at different distances from the
factory, reflecting the gradient in exposure. The area affected, as

given in the contour map, is much wider than the area considered to be
exposed for human health effects.
The Health Dept, even reported a bright side of the disaster.
Gas affected areas have recorded a reduction in the incidence of

malaria. Mosquito breeding has apparantly been

I

13

affected by MIC.

CHAPTER 6

THE HEALTH IMPACT QF THE DISASTER

6.1

The need For »n wpideniioloqical perspective i
When studying the health impact oF the Bhopal disaster. we are

faced with a situation where the cause oF the presenting complex of
symptoms and signs is partially unknown even at present. The long term
effects on human health are also unknown. There is an urgent necessity

to understand the toxicology and pathogenesis oF the agent/s involved
so as to be able to provide rational therapeutic care and iF possible

to initiate secondary and tertiary preventive measures. Data needed
would include Facts about the clinical presentation ot the disease
complex, the pathophysiology of the affected and related organ
systems, the analytical chemistry and toxicology of the agent/s.
The role of the epidemiologist would be to provide information

about:
- the distribution of the new disease complex in the population,
the characteristics

of people who manifest these adverse health

effects,
- the various circumstances which may predispose to the development

of adverse effects.
- the morbidity attributable to the exposure as separate from pre

existing levels of morbidity in the community by comparison with

control groups of similar age / sex structure and

ocioeconomi c

status,

- to provide data relating different symptoms and signs to different

degrees of exposure.
- to understand the natural history of the morbidity over time,
I

I shall commence by describing available data in the

U

I

epidemiological categories ot time, place and person recognizing that
the three are intrinsically interrelated.

6.2

Descriptive epidemiology
An attempt has been made to build up a picture ot the

descriptive epidemiology ot the event and its aftermath from available
sources of information.

6.2.1

I

At the outset icwe pf the limitation* encountered by groups

involved in carrying out medical work with the victims or in studying
the health impact must be considered viz.

a) There was a lack of authentic informatiom regarding the

I

chemical composition of the cloud of vapours that escaped that night.
The leak occurred from a tank containing methyl isocyanate following

an explosive runaway reaction. Very little was written (and possibly
known) about MIC, its reactions in different circumstances or its

toxicology,

in standard textbooks or journals. The company did not

di sclose information that it had accumulated in the process of
registering for the commercial production of MIC. Thus medical

professionals and research workers were handicapped by ignorance about
the identity and properties of the agent whose effects they were
trying to study and treat.

Speculations regarding possible agents ranged from MIC, phosgene,
carbon monoxide, cyanide, cyanogenic substances and a combination of

all the above and/or other unknown chemicals. These drawbacks in

f

carrying out medical work with the victims have been reflected in the

reporting of the clinical findings and also in the treatment given.
b)

Details of the number of fatalities and of those affected

are not precisely known. Because of the suddeness and magnitude of the

disaster, mass burying/cremation of bodies (human as well as animal)

was carried out on an emergency -footing, to avoid -furthur public
health problems.

In the conditions of disaster and panic. with many of

the staff themselves affected, routine administrative structures could

not cope with the need for rigorous documentation vital for future

treatment and compensation of victims and for the understanding of the
epidemiology of the aftermath. Different sources, therefore, give
affected.
varying estimates of the numbers who died and who were

c) Because of the medico legal implications of the disaster a
certain degree of administrative overcaution built up in the months

following the exposure and results of studies of morbidity, toxicology
etc. conducted by various research groups of the Government and the

company were not made available to the medical community for
scientific debate.

d) other factors which affected studies in the early phase
were
mass exodus of people from the affected areas away from Bhopal,

when the
during •'operation faith" twelve days following the exposure.

remaining MIC in the plant was "neutralised". The people subsequently
returned to Bhopal over a period of time. Hence there is a variability
i n the baseline population in the different studies.

migration outside and into the affected areas altering the
population at risk. Most of the affected areas being shanty towns,
many people had their roots in villages and towns outside Bhopal. A

proportion of people migrated back to their hometowns or villages or

I

elsewhere for treatment, jobs or for c_her reasons. Similarly,

relatives of the victims also came into Bhopal to care for their

f ami lies. Thi s sort of social support during times of stress is very
common in India. It has been claimed that a large number of unaffected

people migrated into the exposed areas to claim compensation. Studies

have not substantiated this claim. The occasional case reported must

16

i

i

be the exception rather than the rule.
- invasion by lawyers and other people created confusion and
distrust among the people. This may affect the reporting of the

history / symptoms to extents that would depend on the degree of trust
/ rapport created between the people and the research teams.

6.2.2

Time
The leakage occured on the night of 2/3 December 1984. The leak

was first noticed at 11.30 pm in the MIC production area. Workers

noticed some dirty water spilling from a higher level in the MIC

I

production structure. They also felt the presence of MIC in the

atmosphere by a sense of irritation in their eyes. Due to experience

with previous minor leaks they were able to recognize the presence of
MIC by its irritant effects. Water was sprayed around the point of

leakage, as in the presence of water MIC converts to less harmful
i

products. At 12.15 am indicators showed that pressure in MIC tank 610
was shooting up and by 12.30 am it went beyond the maximum on the

scale ie. 55 psig. The temperature indicator was also beyond its range

ie. + 2u°C. The safety release valve popped out and a gaseous cloud
was seen coming out of the stack which was 120 feet high. The siren

was sounded around 12.30 am for a short while after which only the

internal factory alarm was continued according to the routine practice
.4

(I

followed in the factory. Water was sprayed to neutralise the MIC but

could not reach the height from where the gas was emanating. Around
f

03.00 am the safety release valve of the tank is reported to have sat

back and the gas stopped coming out of the tank.
Meanwhile, around 12.45 am people in JayaPrakash Nagar 100

yards south of the plant, woke up choking, coughing and with a burning
sensation in their eyes, nose and throat - reported by the victims

17

"as if chilli powder was in the air".
At Bhopal’s 1,200 bed Hamidia Hospital, about 3 km -From the

factory, the first patient reported at 1.15 am and then they came in
thousands. It was around 3.00 am that the first deaths due to the gas

leak were reported. There was a stampede as the populace began -Fleeing
the city. People died in their homes, in the hospitals, and on the

roads even upto some distance away -From Bnopal.
This was the starting point -For continuing morbidity and
mortality in the exposed population over the next few days and weeks.

Standard medical therapeutic practices were adopted. However, i t was
the experience o-F local medical practitioners that the exposed people

came repeatedly to the clinics with a variety of symptoms and signs
affecting several organ systems, apparently unrelated to each other.

This continued beyond weeks into months.
Ten days following the disaster it was announced, that as a

safety measure, remaining MIC in the two other tanks 611 and 619 were
to be neutralised by conversion into carbaryl.This created a panic and

despite assurances of complete safety, there was an exodus of people
from the vulnerable localities. On December 13, 1984, 100,000 people

left the city. Many took their animals too. By the next day a quarter
of the city’s population had fled. Many hospitalised patients also
left. Cases of injuries and accidents in the virtual stampede during
the fleeing of the city, occurred on both occasions, adding to the

morbidity related to the disaster. The second exodus furtnur

aggravated the situation as it interrupted treatment. resulted in
physical and psychological stress in those already affected and was

also an additional financial burden.
Generalizing broadly, the event could be described as an
exposure of a population to an extremely toxic air borne chemical

agent, with sharp localization in place and time. It resulted in a

18

I

large excess of deaths and disease frequency in the exposed
population. This is characteristic of an explosive. acute, point
epi demic. However there is evidence of continuing or persistent

morbidity and suggestions of an excess mortality in the exposed
population. Reasons for this are not completely understood. Several
possible mechanisms have been proposed and will be considered later.

6.2.3

Pl»ce
Bhopal is the capital of the State of Madhya Pradesh i n

central India. The city has a population of about 800,000. The UCIL
plant was located in the northern fringe of the city, adjacent to an

existing residential area and barely 2 kilometers away from the
railway station. Some squatter settlements did grow up around the
Carbide plant after it was opened in the late 1960’s. But some of the

areas worst affected by the gas leak had been inhabited for many years

before the Carbide plant opened.
The high vapour density of MIC along with the conditions of

atmospheric inversion that winter night, caused the cloud of gas to
move down and stay close to the ground. The vapours spread slowly in
the atmosphere due to the low wind velocity. All these factors
i

resulted in the population being affected greatly. The gas is reported

to have spread over an area of about 40 sq.km and affected people

seriously as far as 5 to 8 km downwind (see map). Classification c-f

areas into seriously affected and less seriously affected were made on

the basis of crude post exposure death rates in different localities.

This was done by the State Govt. The most seriously affected areas
were JayaPrakash Nagar, Kazi Camp, Choi a Road, Chandbad, New Kabbad

Khana, Sindhi Colony and Railway Colony (see map). The two lakes of
Bhopal are said to have averted a larger tragedy as MIC converts to

less I- 'rmful substances in contact with water.

19

(
I
I

GAS AFFECTED
ZONES of

Uni On
CARBIDE

FIRDQS
MAGAR

SULTANIA RttIMfNTi
«*CMl

JAMAL PVWA

3

hamna4*r

BHOPAL

J.R M
•••HAT

RLY COiOMY

*AZ« CAm»

(D J.P. NAC>AK

3/7

KAZI CAmP

TILA JamalPuRA

CMANOBAD

SHAHJAHANABAD
^.MOOtL GROUND

BUS AT ANO CHHOlA *GAl
•eOO^
NATALI

KCNCHI

11

STaDium

MAMA
MAMM

STATION BAJAPlA

KU

K3.CHANO BARH

cv

U noormamal

BARKNCRi

UPPER LAKE — Z

MA&JNO

SlLOWERZi
^^LAKEy

ARCA

4J, PATEHGARH

area

CONTROL

AREA

»G .

-

CeNT^A.U
iZUooL

BIRLA
ManOIR

MIWISTTRS

PoTef fl a..d
.Itxn-H-Ih

7V|e. It cb

(

Zcr<A

"TO

BumGAlOw

<^_?C vO <

('7^)

tu. <>p.

ir-.r.'I S’K«.-\bo

1

HOU:, d ’"i£.? to be •~ons

factor 7.
on

4s

r,E~ A?er-

b

COP'• •,

d

d i ■ t •?

• , i

pi

wonl d

r-• ■

.1 '"’i •*. ] ?1 ] ••-

DO-J - ■ i.



epidemiolngi ■ U

i nf or 'ip- '. •; or ,

i’A

2(1

I

!?• F

be

Ul ! 1
arid

(J-:/

thr

i

• ddl. V ..

i: b'..r_'.

AP£A

12. hawamahal

714,15,16 L<
/control

JAHAN&IRa
Bao

SouRtS 1 J-tMix

CHHOLA

RAILWAY COLONY

AIShsagh

could be drawn. Relating these numbers to the denominator viz. the
population who were exposed and hence at risk, would give crude death

I

rates -following the exposure, according to geographical area or
locality. Baseline data of the population available from the census,

electoral rolls or even by the surveys conducted a-fter the disaster by

I

Governmental agencies, could be used to calculate age and sex
standardised mortality rates tor ditterent areas.

6.2.4

Person
This aspect will be discussed in terms ot the population at

risk, its demographic a d other characteristics. The picture ot

mortality and morbidity will be described in later chapters.

r

a) Population at risk

Estimates of the total population exposed to the toxic gas or
the "population at risk" for the derivation of mortality and morbidity

rates are 200,000 according to the State Government and UNICEF.
I

(

Results from epidemiological, community based studies (Patel et
al,1985 and Nagrik study, 1985) using control groups 10 kms away from
the -factory, suggest that even the control groups were mildly affected

by the exposure. This indicates that the basis for defining the

"exposed population" needs furthur substantiation. This will be dealt
with again in the discussion on the exposure variable.
The factor of post exposure migration into and out of the

I

exposed localities would have also affected the composition of the
population at risk. Andersson et al (1985) observed that the more

seriously affected had gone back to their hometowns or villages.
Sathyamala C (1986), found that in a population of 8159 surveyed in

Sept. 1985, 43 people (0.527.) moved in after the gas leak and 41

(0.507.) had moved out. Be > the above movements would result in an

21

I

overall dilution e-f-fect or underestimation ot morbidity . Though
quantitatively this may not be o4 a large order
study)

<17. in the above

it must be kept in mind that the qualitative di-Herence may be

important if those who were seriously ill had gone away.
New births into the population and deaths (-fully, partly or

not attributable to the exposure) would also alter the baseline
population

Since households have been general 1y used as the sampling

units in the studies conducted so -far, this -factor has not been
considered at the sampling stage. However it would alter the number

and composition o-f the baseline population and would af-fect the

calculation o-f rates.
Patel et al

(1985) extrapolating -from prevalence rates of

morbidity have estimated that, o-f the exposed population about

70,000 would be su-f-fering -from serious health effects while 45,000

would be suffering moderate to mild effects.

b) Demographic characteristics

UNICEF officials (Agarwal A 1985) estimate that, of the
affected population :

i

- 757. are slum dwellers ie. from the lower socioeconomic strata;

Ii

- 807. are Muslims;

- 407. are children below the age of 15, 107. are elderly and
- 207. are women in the reproductive age group.
The basis of this analysis is not known.

It was probably done as an

estimate for planning and administrative purposes. The demographic

This
data appear to be direct extrapolations from national figures.

would be very inadequate data for a thorough understanding of the
aftermath.
Banerji et al

(1985) have described the socioeconomic profile

of the study population as follows :

22

Muslims : 307.,

Lower castes : 207.,

Backward castes : 187.

Income of Rs 150/head/month ( le relatively well off) : 107.
Housing : Kutcha (without brick and cement) : 707.

Pucca (well bui 11/concrete) : 307.
Presence of holes in the structure : 507

(allowing air entry)
Patel et al (1985) found the population to De predominantly

Muslims and Harijans in JP Nagar and Tamilians and Maharashtrians in

Anna Nagar.They belonged to the lower socio economic class

the

percentage of skilled workers being less than 10 7.. The range of
occupations included daily wage labourers, construction workers, beedi
rol1ers, cobblers, railway and factory employees, domestic workers,

self employed artisans and potters.
The profile is of a population, the majority of whom belong to
the lower socioeconomic class. The poor housing conditions would have

offered no protection from the toxic chemicals in the atmosphere.

c) Community awareness of the hazard potential of the plant

I

Most people had no idea about the hazardous nature of the
plant operation. Banerji et al (1985) report that the population were
not told earlier of the potential hazard of the plant. Nor were they
aware of preventive measures to be taken in case of a leakage eg. use

of a wet cloth, moving in a direction away from the wind, not running

etc. Unfortunately most of them ran in the direction of the wind
carrying the gas and were furthur exposed. A Patel et al (1985) found
that 8.37. of the population in JF Nagar and 0.087 in AnnaNagar took

safety measures at the time of the disaster. Most of these reported to
having used blankets/wet towels over the face by instinct rather than
due to prior knowledge. The workers in the plant knew of these

precautions and all those in the

23

ight shift escaped unhurt, except

one.

d) Pattern of utilization of health —rvices

Banerji et al (1985) found that the uti?ization o-f health
services by the study population after the exposure was as follows :

Hospital

40.2 7.

Camps

46 7.
2.5 7.

Dispensaries
General practitioners

25 7.

Registered medical practitioners

2.5 7.

Institutions outside Bhopal

9.1 7.

There is an overlap as some people utilised more than one type of
service. These findings are an indication that when studying the

j

distribution of disease in the community, data

collection would need

to be population based to get a true picture of morbidity. When using

hospital or clinic based data one would have to keep in mind selection

bias caused by selective attendance of patients. Possible determining
factors for this could be severity of illness, accessibility of

service, social class of the affected persons and so on.

e)

Impact on income

Patel et al (1985) found that 65 7. of wage earning individuals
in the exposed group experienced a drop in income ranging from 20 7. to

100 7. with a median of 50 7.. In the controls, only 9 7. reported a drop
in income in the range of 20 - 55 7.. This occurred due to physical ill

health resulting in occupational disability.

24

CHAPTER 7

IORTAI TY

7.1

Exposure related mortality rttgs
There is uncertainty as to the exact number o+ people who

died. In its petition in Court, the Government has claimed 1,700 dead.
This is based on death certi-ficates issued/deaths recorded by
Government authorities.

The Indian Council o-f Medical Research report (1985) states
that about 1,200 people died in hospital wards. They estimate the
total death -figure to be about 2,000. The maximum number o-f deaths
were recorded in JF' Nagar, Kazi Camp, Kenchi Chola and Railway Colony.

These areas accounted -for 777 deaths.
They quote a study done soon after the disaster, in which 300
■families consisting of 968 males and 863 females were surveyed.

47

deaths in males and 35 deaths in females were recorded ie. the crude
post exposure death rate for this population was 48.55/1000 population
i

I
I

for males and 40.55/1000 population for females. The maximum mortality

was i n the 0 to 5 year and above 60 year age group. Most deaths
occurred within 46 to 72 hours of the disaster. No details are given

I

of exactly how long after the disaster this study was done, from which
populat i on the families were or awn, what the method of sampling was or
details of the age structure of the population or the dead.

Andersson et al

(1985), in an epidemiological study, conducted

during the first 10 days after the disaster, found the crude death
I

I

rate

derived from households in the worst affected population based

cluster to be 37. or 30/1,000 population,

(death rates being calculated

as number deceased/number exposed). They have stated that this data

25

would confirm estimates of a total of 2000 - 2500 exposure related
deaths. Random sampling methods were not used and exact location of

study areas have not been mentioned. However extrapolation of rates
from localised study areas to the total exposed population should be

made with caution. Reasons for this will be apparent when summing up
the results from the various studies conducted.

Patel A et al (1985), in another population based.

cross-sectional study, using statistical methods for sample size

determination and random sampling methods, found that the crude death
rate was 86.6/1,000 population in JayaPrakash Nagar (01) and 7.9/1,000

Mkr

in the contrrl area of Anna Nagar (15)(refer to the map). The Dept, of

Information and Publicity of the State Govt. had documented the death

rates to be 23.4/1,000 and 3.2/1,000 respectively in the same areas.
It appears that there is a gradient of mortality according to the

degree of exposure. This is another reason why direct extrapolation of

rates from any particular study/area to the total affected population

may be misleading.

Banerji D et al (1985) conducted a survey in the affected

areas, between Jan 6 to 15,1985. Using a sampling frame of 68,000,

1

they randomly selected a 6.667. sample (1 in 15 households), and
administered a semistructured questionnaire by trained interviewers.

They enumerated 82 dead and 5 missing (presumed dead) in 700

households in the severely and moderately affected areas. The exact
denomi nator has not been mentioned in their preliminary report, to be

able to calculate rates. They extrapolated the number to their
sampling frame of 68,000 to yield lc>05 dead. From this study the crude
death rate for the combined population of severely and moderately

I

i

affected communities is 19.19/1,000 population. The exact area covered

by the study has not been mentioned.
Beside'

the caution mentioned earlier regarding extrapolation

26

ot rates, it appears that combining rates ■from di-Her ent localities

eg. severely and moderately exposed to give an overall rate may mas)
important differences in mortality rates.
Sathyamala C (1986), surveyed 3 exposed localities to study

the impact of the exposure on the outcome ot pregnancy.

The basis o+

selection was the post exposure mortality rates as given in later
These were as
unpublished data •from the study by Banerji et al..
|

■follows :
JF* Nagar - 65.3/1000, Kazi Camp - 46.7/1000 and Kenchi Choi a -

35.7/1000. Their study found the crude post exposure death rate for
the 3 localities together to be 33.19/1000.

A table of postexposure mortality rates as derived from

different studies has been compiled. Relevant aspects of the study
designs have also been given.

I

I

I

27

Post exposure mortality rates
A tufnmary from different sources
jpopulation/
Study
sample size
group/
1nvestiqator

Sampling
method

MP State
Govt.

I

Time
of
study

Area
of
study

Crudepost
exposure
deathrate

Early
post
disaster

JPNagar

23.4/1000 |

AnnaNaqar

3.2/1OO0 j

300 families
(968 males *
863 females.’

short 1y
after
disaster

Andersson
et al

? number
in each
cluster

10 days
post
exposure

worst
affected
cluster

3 7. or
30/1000

3 months
Random
sampling post
exposure

JPNagar

86.6/1000

AnnaNagar

7.6/1000

148
exposed
138
controls

Patel A
et al

r-----

1 month
700 4ami 11es Systepost
| matic
' sampling exposure

Ban er _ii
et al

i
i

81 65
persons

Sathyamala

severely -h
moderatel \{
affected
19.2/1000
areas
JPNagar
K.az i Camp
K.Choi a

I

I

48.5/1000
! (males)
40.5/1000
(females)

1CMR

9 months
;Random
isampling post
exposure

i

ii
i

I
I

28

65.3/1000
46.7/1000
35.7/1000

JPNagar
Kazi Camp
33.8/1000
K.Choi a
together

There seems to be a wide variation in crude death rates
derived from the various studies as well as in those stated by the
I
t.

Govt. Possible reasons for this could be :
a) differences in levels of ascertainment, reporting etc
b) due to factors related to study design - sample size, sampling

methods etc
c) real differences in different localities reflecting varying levels
of exposure to toxic gas.

I

The methods used in enumerating deaths would also play a rol e

i n accounting for differences between studies in postexposure death

rates. Andersson et al (1984) have mentioned

that they calculated

death rates as the number deceased/number exposed. Other studies have
not stated what they have considered and used in the numerator and

I

denominator. The different studies were conducted at varying time
intervals following the exposure. We do not know the time period used

in the different studies as cut off points in the enumeration of

deaths attributable to the exposure.

From an aetiological point of view, as well as for the
victims families to receive adequate compensation, it is crucial to
try and achieve greater precision in enumerating deaths. It is

important also to analyse exposure related mortality rates taking into
consideration age, sex, locality in which present at the time of the

disaster and degree of exposure.

f
29

7.2

Demographic ch#rapteristic^
Furthur details o-f the study by Banerji et al

(1985) are given

here as it describes the age/ sex and socioeconomic profile of the
dead.
a) Males accounted for 607. of the deaths, which they commented
was an important and unexplained finding. Majority of these were i n

the age range of 2 to 20 years. As a first step in interpreting these

differences it is important to take into consideration the sex ratio
and age distribution ie. the population structure, of the

I

population

at risk. As mentioned before, standardised mortality rates would be

more interpretable.
b) There was one death per household in 49 households, two

deaths/household in 11 households, three deaths/household in 4

I

households and one household had 4 deaths. There is a need to analyse

these deaths in terms of degree of exposure.
c) In an attempt towards determining the degree of exposure
it was found that:

- 757. of the dead were among those who ran on foot,
- 23.57. among those who remained at home,
- while none who used a vehicle died.
I

I

This is related to the finding that 737. ran, 217. stayed in their
house and 6.37. used a vehicle. This is plausible because besides
direct exposure to tne toxic chemical laden atmosphere, those who ran

also inhaled deeply and had an increased rate of respiration thus
getting exposed to more of the toxic gas.

d) Among the dead the socio economic profile is as follows :
- 567. lived in houses with large holes,
- the proportion living in kutcha houses was higher,

30

- the proportion belonging to the lower and backward

castes was significantly higher.
the proportion of Muslims was similiar to that in the

study population
i e.

those who died were the poorest of tne poor.

The few rich who came within the sweep of the cloud of toxic chemicals
did not suffer as much damage *' because of their well built houses.

healthier bodies and possession of/access to transport."
e) The study hypothesises that there would have been an
underestimation of deaths because of underreporting of deaths among

the homeless and destitute who would have been the most exposed and
vulnerable. An estimate of 3,000 shelterless in Bhopal has been made

(Agarwal 1985). Many of these people used to live around the railway

station which was directly in the line of the cloud of toxic

chemi cals.

The Tata Institute of Social Studies, Bombay,

conducted a

door to door survey regarding exposure, mortality, socioeconomic
status etc the results of which were to become the basis for relief.

compensation and long term treatment. But its total tally of
dead, even less than the officially counted bodies caused it to lose
credibility. The survey failed to cover 600 exposed families i n which

i

deaths could have occurred.

It could not enumerate 315 families who

had migrated outside the city after the disaster and 286 families who
had their houses locked. The findings of the study, even with its

limitations, would have provided some information of use. However the
results have not been published or made available.
Many say that the official numbers

are underestimates of the

true figures of mortality. Other estimates have been given ranging
from '“000 - 10,000 (Agarwal, 1985) . However most of this data is

C Tvi

Oh -1 to

06906

/vg:



anecdotal and hence difficult to interpret.

Regarding mortality in the months following the disaster,
but attributable to the exposure, a State Sovt.offici al has stated
that during the first year on an average 15 such

deaths were occuring

per month (Diamond,1985). Whatever the basis and validity of this

statement it raises the important question of the need to evolve a
reliable method to take count of these attributable excess deaths.

The number of excess abortions and stillbirths should also
be added to the death toll.
To summarise, therefore, the main features of mortality in the

data reviewed are :
1) Number of deaths :
estimates range from 1,700 - 2,500 - > 5,000.

2) Exposure related death rates :
varies in different studies and in different localities

3) Area distribution
severly affected areas - JP Nagar, Kazi Camp, Kenchi Choi a and

Railway Colony.
4) Sex distribution

apparently an excess in males.
5) Age distribution

excess in under 5 and over 60 age groups
(age / sex estimates based on crude rates)
6) Socio economic class distribution
lower socio economic classes Dost affected.
7) Excess abortions and still births occurred.

8) Continuing mortality attributable to exposure needs to be
consi dered.

32

I

CHAPTER 8

AUTOPSIES

Autopsies were conducted at the Medico Legal Institute based
at Mahatma Gandhi Medical College, Bhopal. Findings nave been reported
i n the

ICMR reports (Mar and May 1985) and are described in some

detail here as they are of relevance to understand the pathology

produced in the human system by acute, severe exposure.
8. 1

The first week
The usual postmortem lividity or cyanosis was not present, but

there was a pinkish discolouration. Conjunctiva were red. Hypostasis
was present all over the body and was not restricted to the dependant

parts. A common finding was the presence of thick, tenacious , foamy
froth covering the nose and mouth.

Pathological changes were present in the entire respiratory
tract. Lung weight was 2

to 3 times the normal. The lungs were

waterlogged and had a cherry red colour. They showed congestion,
haemorrage and consolidation . The vessels were filled with thick,
viscid, dark cherry red blood. The bronchi and trachea were red in
i

colour and the lumen was filled with white tenacious material.

Microscopic examination showed severe tracheitis and bronchitis with
denudation of the epithelium in some sections and necrotising
bronchiolitis in some. There was marked congestion and thickening of

the alveolar septa. The alveoli were filled with albuminous fluid.
There was very little evidence of secondary infection.

8.2

The second wetk

While grossly and microscopically the lungs continued to be
the seat of primary change, there was a gradual transition i n the

pathological changes. Characteristic cherry red colour of the blood,

33

t

heavy □edematous darkly reddish lungs and varying degrees ot oedema of
the brain continued. Acute desquamatous changes in the trachea and

main divisions of the bronchi persisted. There were varying degrees of
bronchioli 11s, bronchopneumonia and infiltration of the alveolar

spaces by polymorphonuclear cells.
In the acute phase there was oedema of the brain and congestion of the

1eptomeninges. In a few cases, the liver showed a mild degree of fatty
change which can be either incidental or secondary to anoxia

consequent to pulmonary changes.

8.3

The third week
The respiratory tract showed the same appearance, though the

lungs were relatively reduced in size and weight. However they were

still reddish and exuded a lot of fluid from cut surfaces.
A striking feature in some cases was that on opening the
thoracic or abdominal cavities, viscera which was normal in colour

rapidly acquired a reddish tinge on coming in contact with

the

atmosphere. There was variable involvement of other viscera. 8rains

which were uniformly heavy and oedematous, showed either uncal
grooving or tonsillar herniation with compression of the cerebellum by
the tentorial edge. In a few cases the liver showed severe congestion.
The spleen though markedly congested was shrunken in size and the

capsule was wrinkled. The kidney showed extreme congestion in the
cortex and medulla. The heart contained blood clots which were cherry
red with some chicken fat like material. , ae liver showed haemorrhage

all over. The capsule of the liver was found to be separated and could

be easily pulled off. The gallbladder was distended. The stomach and

I

intestines had haemorrhages in the submucosa of the wall. The spleen

was found to be softened. The kidney showed haemorrhages.

36

8.4

HiBtQP^thplPQY

The lungs showed congestion and oedema. The bronchial lumen
was full of exudate. The trachea showed superficial ulceration and

loss of cilia. Muscle fibres in the bronchial wall showed
fragmentation. The kidneys showed necrosis of the proximal tubules.
Microthrombi were present. The liver showed centrilobular congestion.

patchy necrosis and widening of the central veins. Heart showed

interstitial oedema and necrosis. Atrophy of the malpigian capsule was
a consistent finding in the spleen. The cornea showed denudation of
the epithelium. The thymus ,testis and ovaries showed no changes.

Multiole lesions were common. Sections of the lungs of stillborn
showed no abnormalities.

They quote the case of a woman who had manifested MIC effects
and subsequently recovered. She died of hyperpyrexia following

a

Caesarian section in the third week of March 1985. On autopsy (on

29.3.85) lungs showed congestion, oedema and haemorrhage, and the
small bronchioles showed obstruction. There was no pontine

haemorrhage.
Electron microscopic findings from autopsies

Lungs showed loss of lining membrane of the epithelial cells.

Type 11 pneumocytes were present indicating that capacity for
regeneration was not totally lost. Red cells looked different- they

had lost their electron opacity. This could be because haemoglobin is

lost or a change has occurred in the structure of the haemoglobin.
Several areas showed zones of activated fibroblasts. Some lungs showed
secondary bronchopneumonia. The brain showed presence of siderosomes.
Neurones showed flocculent opacity indicating necrosis. Examination of
one placenta showed loss of microvilli on the maternal side of the

syncitial trophoblast.

CHAPTER 9
nORPiPITY- CUINICAU FINDINGS

9. 1

General comments
A pro-file erf the morbidity caused by the exposure has been

compiled from the following sources of information :

a) the two reports by the Indian Council of Medical Research

(ICMR) in March and May 1985 describe the clinical picture and results

of laboratory investigations in the acute and subacute phases. These
observations were made by staff of the local Mahatma Gandhi Medical

College at its attached Hamidia Hospital as well as by specialists

sent in from other parts of the country. Several research projects
(22-25) were set up by the two organizations in the initial phase for

long term follow up of the victims. However further reports of their

findings or progress are not available.
b) studies done by non-governmental groups. These include
research teams from academic departments of universities, independent

professional groups and voluntary agencies.

The clinical picture of morbidity is described first followed

by the epi demiologial studies. Comments on the strengths and
limitations of the data are interspersed with the description. There

will be a further discussion of methodological issues later.
The pattern of morbidity varied over time. For the purpose of
this report the acute phase has been considered as the first 2 weeks

post exposure. The subacute phase is from 2 weeks to 4 months and

beyond that period is the chronic or long term phase. This
classification is arbitrary and partly artificial to help understand

the clinical picture over time.

The most striking symptoms and signs with which most people
presented immediately after the disaster or in the acute phase were

36

related to the eyes and respiratory tract. There were also a wide
variety ot

clinical symptoms related to different organs and systems.

A description o4 the clinical -findings in the acute and

subacute phases follows.

9.2

ACUTE AND SUBACUTE PHASES

9.2.1

The eyes
Mittal (ICMR, 1985) reported the following eye conditions in

patients from the hospital OF'D and wards of Hamidia Hospital.

8,000

patients were seen here in the first 24 hours, and 34,000 patients

were treated in the first few weeks.
Patients initially complained of a severe burning / foreign body

sensation in the eyes, blurring of vision, profuse lacrimation, and
difficulty in opening the eyelids. On examination there was lid

oedema. 60 - 70 7. had superficial keratitis and conjunctivitis.

Superficial ulceration of the cornea in the interpalpebral region was

observed in several cases. Many had punctate keratitis in the lower
sector. Corneal pathology was mainly confined to the epithelial

layers, rarely penetrating the stromal tissues. They observed that
children had fewer ocular problems. Eyes of nearly all the patients

returned to near normal in a weeks time with healing of the lesions.
Detailed investigations did not suggest involvement of the posterior

chamber. There was no evidence of blindness or deterioration in

vi si on.
Andersson et al (Dec?84) reported findings in 10 hospitalised

patients on the eighth day post exposure. All had discrete superficial
1esi one, usually in a band across the interpalpebral region with the

typical whorling pattern of new epithelial growth. No limbal necrosis
or abnormal endothelium was detected. Their findings were in keeping

37

with the report given above.
Andersson et al (Dec’84) also reported findings from a study

o-f community based clusters, 2 weeks post exposure. The exact location
and distance o-f the clusters +rom the -factory were not speci-fied. They

reported that over half the community demonstrated eye signs which
could be attributed to the exposure. These were mainly interpalpebral
i njection

and signs o-f healing epithelium. Fundal changes, mostly

venous dilatation were more common in the exposed. There was no
difference in the age standardised visual acuity between exposed and
unexposed groups.

A.Bang (Jan’85) reported that in the week after the disaster a

quick and crude community survey in JayaPrakash Nagar (100 yards from
the factory) revealed that about 507. of the population had eye

symptoms.
The Nagrik study (1985), found that 807. of people within

I

km

of the factory had ophthalmic symptoms, as did 607. of those at 2 km

and 40 7. of those at a distance of 8 km. This revealed a gradient of
effect as well as the fact that exposure occurred even upto 8 km away.

9.2.2

The respiratory system
The ICMR report’s (1985) state that people initially

complained of sudden onset of difficulty in breathing, coughing and in

some cases, pain in the chest. On auscultation, many had bilateral
crepi tations. Xrays revealed interstitial pulmonary oedema, alveolar
type oedema, pneumonitis, hyperinflation of lungs and collapse of
surrounding area. Rapid deaths following exposure probably resulted

from massive pulmonary oedema and associated hypoxia.
A.Bang (1985) in the survey mentioned above, found that about

257. of the population in JP Nagar had respiratory symptoms. A large

number, even those with minimal respiratory symptoms had coarse

38

crepitations and rhonchi. Many of the" mild" cases were either not
attending clinics or were not being given a thorough clinical
examination. He pointed out that with inadequately designed studies

and poor documentation the real epidemiology of morbidity may be
missed. He also observed that the expected tide of secondary infection
did not follow the initial period of chemical pneumonitis. Reasons for

I

this were not understood. At the community level. antibiotic cover was

either not given or was too inadequate for most of the affected

persons and hence could not explain the phenomenon. There are also
anecdotal observations that dead bodies of people / animals discovered



a day or two after the disaster were not decomposing. On experimental

studies (ICMR,1985) MIC was found to have a bactericidal effect.
Andersson et al (1985) observed that respiratory distress was
most marked in the community cluster "second in distance from the
factory", affecting 20 7. of the population. Many were too disabled by

breathlessness to move more than a few steps or even to talk.
I

S.R.Kamat et al (ICMR,1985) noted on 24 subjects. ( a self

selected group who had gone to Bombay for treatment shortly after the

I

disaster ), that lung function tests suggested

restriction,

reversible obstruction and defects in oxygen exchange. Blood gas

analysis revealed anoxia, compensatory respiratory alkalosis, raised
carboxyhaemoglobin and methaemoglobin. They reported evidence of

lymphoid granulations in the throat.
They later reported findings from an extensive evaluation of

pulmonary function of 82 patients (also self reporting to Bombay).
Xrays in 78 were suggestive of interstitial pneumonitis. This was

corroborated by blood gas and lung function studies. Methaemoglobin

levels were raised in 63 out of 80 and was stated to be suggestive of
interstitial alveolitis. Pulmonary function tests indicated central

airway obstructior

A large majority of the 82 also showed some

39

o-F respiratory symptoms had ventilatory impairment, 127. had

restrictive lung disease, 67. obstructive airway disease and 227.
obstructive cum restrictive defect. Ventilatory defects were not
observed in patients with mild exposure to toxic gas.
The ICMR report (1985) states that ot 35 patients in whom

blood gas analysis was done, 23 severely exposed patients had arterial
oxygen tension (Pa02) less than 60 mm Hg le. moderately low levels.

Moderately and mildly exposed patients had normal Fa02. Low arterial
carbon dioxide tension ( less than 35 mm Hg ) was observed in 12

cases. It was stated that these results indicate significant
alteration in blood gases and suggest alteration in oxygen carrying
capacity of the blood.

Narayanan (ICMR,1985) reported, from experience in a 30
bedded, hospital set up adjacent to the factory and hence more
accessible to the affected people. The exposed population were
presenting with recurring respiratory problems. They also complained

of inability to perform accustomed physical activity. They had

tachycardia and severe tachypnoea. The haemoglobin level was normal or

raised. On blood gas analysis, Pa02 and PaC02 were moderately low.
Pv02 was moderately low and PvC02 very high. 2,3 DPG

(diphosphoglycerate) levels were also raised. These findings suggested
a defect in oxygen transport and tissue anoxia.
The ICMR report (1985) states that even two months post

exposure, nearly 407. of patients attending the hospital presentea with
respiratory symptoms of breathlessness, cough ar.J in some cases fever.
Persistent tachypnoea was a characteristic feature. In some patients,
symptoms were out of proportion to clinical and radiological

observati ons.

It has been reported that many of the delayed deaths were
preceded by severe respiratory distress.

41

9.2.3

Haematology
Ram Singh (ICMR,1985) reported initial haematological

■findings, 15 days after the exposure. There was haemoconcentration and
1eucocytosis. There was no evidence of coagulation disorders.

Approximately 257. of severly exposed cases, reporting to hospital, had
haemoglobin levels above 14 gm7. and 337. had raised eosinophil counts (

above 207. ).
The ICMR report (1985) describes another series of 237 cases
investigated during the first two weeks. Polymorphonuclear cells were

increased in 35 % of cases, 527. had raised lymphocyte counts, 197. had
eosinophilia in excess of 207., and 157. had haemoglobin levels above 14

gm 7..
The ICMR report (1985) quotes a study finding of a 20 - 60 7.

reduction of the free amino groups in the haemoglobin of persons
exposed to the toxic gas.

9.2.4

The gastrointestinal system
The ICMR report stated that patients also presented with

nausea, vomiting and burning in the stomach. Endoscopic examination

revealed superficial gastritis and oesophagitis. A small proportion
had hepatomegaly.

/

9.2.5

The neurologic#! system

ICMR reported that immediately after the disaster severly
affected cases showed varying grades of loss of consciousness ranging
from mild to deep coma. The main presentation in children was coma.

They report that in a study of neurological disorders in the affected

population , 128 adult subjects were screened and revealed the
following :

: II
42

I

neuromuscular weakness- 1,

right hemiplegia - 1,

hearing loss - 2,

tremors and vertigo - 2.

The

method of selection and source of cases has not been mentioned.

nor whether a standardised method of examination was used.

Andersson et al (1984) reported that those who tell

unconscious had few or no eye symptoms or signs on recovery. They also
found that collapse and unconsciousness was noted in the cluster
second in distance “from the factory and not in the others. No

quantitative data has been mentioned, to see if the difference is

significant. This observation suggests that different patterns of
morbidity may occur in different clusters.
Bharucha (ICMR,1985) reported initial observations of coma.

tremors and paralysis in some cases soon after the gas leak. No
recognizable patterns of neurological disorders were present 5 months
post exposure, though many people complained of general weakness.

I

In the 113 affected people seen at the KEN Hospital Bombay

neurological conditions such as sensory motor loss, tremors, loss of

consciousness, irritability and depression were found in a significant
number of cases.

I
9.2.6

PEychploqical disorders
Sethi (ICMR,1985), reported that of 168 cases of mental

disorders treated in special clinics, the majority were neurotic
disorders viz. neurotic depression,anxiety neurosis and hysteria.
Psychotic disorders were rare. Women under 45 years were predominantly

affected.
S.R.Kamat et al (ICMR,1985) reported on psychometric
evaluations carried out on 68 self reporting cases : 22 showed
evidence of depression and 19 showed evidence of cognitive defects

with poor memory performance.
I
43

I

I

The team from the National Institute of Mental Health and

NeuroSciences (NIMHANS) found a large community load of mental ill
health following the disaster (ICMR,1985). They reported that

approximate! y 10 - 12 7. of those affected, who visited community based
general practice clinics. were presenting with psychiatric
mam f estati ons.

Symptoms of anxiety and depression were foremost.

Sleep disturbances, nightmares, gas phobia, feeling of hopelessness
and grief reaction were common. Families of the affected population

were finding it difficult to cope with the stressful situation. This
as the NIMHANS team noted that long

is an area needing furthur study,

term after effects have been reported in previous man made disasters.

9.2.7

Outcome of pregnancy
The ICMR report (Mar’1985) reported findings of 97 women who

had delivered. Among them there were -

sti11 births

5 (5.157.)

abortions

17 (17.57.) ie. total pregnancy wastage of 22.77.

congenital anomalies (minor) - 3

Most babies were full term but with low birth weight, 2kg. on
average. In terms of development the babies appeared normal. The

mothers weight was 40 - 45 kg.
I
I

I

In the ICMR report (May’85), Dabke described the results of
645 pregnancies : stillbirths - 8

abortions

- 67

congenital abnormalies (minor) -

9

low birth weight - 29.87

They stated that these were not in excess of those expected in a
normal population. Rates are calculated by relating the numerator to a

given population ie the denominator in a given time period (usually a
year). Hence no comments about normality /abnormality can be made from

I
AZ*
I

the above data. Comparison with control groups and it possible with

national and regional rates should also be made.

9.2.8

The reproductive yystfA
R.Bang and *!.Sadgopal <1985) studied the impact on women's

reproductive health 2 months post exposure. 55 women were examined i n
Ob/Gynae tield clinics in two of the affected slums - 947. of these had

leucorrhoea, 797. pelvic inflammatory disease and 467. had excessive
menstrual bleeding. Women also gave a history of suppression of

lactation,impotence in the husbands, abortions and stillbirths. This
provided a clue that there were adverse effects on reproductive
health, particularly womens health, which needed furthur
investigation. They admitted limitations of the lack of a control

group and small sample size. However there was also the problem of a

self selection in clinic based data.

9.3

THERAPEUTIC TRIALS

Medical management of patients with eye, lung, CNS,
gastrointestinal and other presentations

was symptomatic and followed

standard practice. This approach did prove life saving and offered

some degree of relief to many. However, with the passage of time, it
was observed that patients kept coming repeatedly either with
persistent symptoms or relapses following a remission.

While it was acknowleged that some of the long term.
multisystem!c symptoms could be explained as being due to the

aftereffects of severe lung damage,

ICMR scientists (ICMR,1985)

suspected the presence of systemic toxicity. Autopsy findings,
laboratory investigations and a rapid literature search lead to the

hypothesis of an enhanced cyanogen pool in the body resulting from the

45

exposure either by direct inhalation ot o-f cyanide or more likely by
the breakdown of MIC within the body. The detailed rationale for this
has not been reported.

Autopsy findings had shown arterialization of venous blood
giving a reddish tinge to internal organs and tissues.

Carboxyhaemog1 obin and methaemoglobin were not detected. However,
samples from all victims showed twin bands of oxyhaemoglobin. It was
demonstrated that MIC could produce a reddish colour when mixed with

blood.
Urinary thiocyanate levels were found to be higher in the

exposed population.
Smoking and/or exposure to smoke,and the eating of certain

foods eg cabbage, spinach, cassava etc,are known to enhance the
cyanogen pool and result in increased excretion of urinary thiocyanate

(which is used as an indicator of cyanide exposure). Rosling (1986)
summarised the mechanism of detoxification of cyanide in the body as

follows - cyanide is trapped in the erythrocyte fraction of the blood
and is converted to the less toxic thiocyanate in the presence of

sulfur. This conversion is normally attributed to a reaction with

thiosulfate catalysed by the enzyme rhodanase. Thiocyanate is then

excreted in the urine. Intravenous administration of sodium
thiosulfate is known to increase the capacity for detoxification of
cyani de.

The toxic effects of cyanide result from impairment of the

mi tocnondrial respiratory chain by inhibition of the mitochondrial

enzyme , cytochrome oxidase. Studies (ICMR,1985) have shown that pure
MIC had no effect on cytochrome oxidase, but its degradation products
did. This results in under utilisation or non utilisation of oxygen

at

the cellular level.

Based on the hypothesis of an enhanced cyanogen pool resulting

46

from the exposure to toxic gases, sodium thiosulfate was administered

to some patients as an antidote <ICMR,1985). It was stated that this
resulted in marked clinical improvement and a significant increase in
PvC02 in both central and periferal veins indicating better

utilisation of oxygen by the tissues.
A double blind study using sodium thiosulfate and glucose was
conducted. Of 30 patients, 15 each were given 2 injections of sodium

thiosulfate or glucose. Urinary thiocyanate levels were determined at

3 and 5 hourly intervals and compared to the baseline level before the
i njecti on. In patients given sodium thiosulfate there was an 8 to 10

fold increase in excretion of thiocyanate in the urine in a
significantly large number - in 10 out of 15 patients. Only one of the

15 receiving glucose injections showed such an increase. Criteria for
selection into the study have not been stated.
Subsequently, Narayanan (ICMR,1985) reported that of 230 cases

treated, complete records were available for 167 ( 87 men, 69 women
and 11 children). Symptomatology before commencing treatment was

breathlessness and / or general weakness / tiredness. In 29 patients

these symptoms were present at rest. While in 132 they were elicited

by moderate exercise. Following administration of sodium thiosulfate,
9 showed no improvement while there were varying degrees of
improvement in the rest over different periods of time. Details of how
the analysis was done has not been stated. There were 10 cases of

adverse reactions - 5 with feverishness and one each of skin rash,
transient venospasm, a sense of heat over the body, exaggerated

reflexes and loss of memory.
On the basis of these studies, a recommendation to use

thiosulfate as a therapeutic agent for the victims was made.
Indications and contraindications for use, dosages for different age

groups and details regarding administration were spelt out.

47

(

H.Chandra (ICMR,1985) reporting on the results of sodium

thiosulfate in over 2000 cases stated that it was" found to give
beneficial results." It is not possible to comment on this. Anecdotal
reports of dramatic cures have also been reported.

N.P.Mishra (ICMR,1985) reported on a trial of sodium
thiosulfate treatment with 120 cases and 100 suitably matched
controls. Only results of clinical observations were recorded as there

were no facilities for blood gas analysis. Urinary and serum
thiocyanate levels were studied. Urinary thiocyanate levels of

controls were in the range of 0.5 to 5.65 mg which is much more than

w-

0.6 to 0.9 mg which has been suggested as the normal range. He

therefore doubted the utility of determining urinary thiocyanate

levels. He studied clinical findings such as dyspnoea, chest pain.

I
I

I
F'

I

general aches and pains, fatigueability, pain in the abdomen,

appetite, pulse rate, lung signs and subjective feelings. All these

I
I
I
I
I
I

I
I

were given an arbitrary score so that in the worst cases the score
totalled to 100. Results showed that in the 100 cases given sodium

thiosulfate :

60 showed a decrease in score,
19 had an increase in score and
21 showed no change
The trial was carried out double blind. An important finding was that

the greater the initial score before therapy, the smaller was the

reduction of the score after therapy. There were very few cases with
side effects. These were feverishness, skin rash, sense of heat all

over the body - all of which were relieved by antihistaminics. Urinary
thiocyanate levels were estimated in 60 cases. The mean value before

therapy was 1.068 +/_ 1.03 and after therapy 1.46

1.113. The basis

for selection of cases and controls has not been specified. No

comparison between cases and contro1

I
I

48

has been given regarding

response to sodium thiosulfate administration or regarding urinary
thiocyanate levels.
The data reported regarding all the studies done is inadequate

for a thorough appraisal.
It is also known that levamisole was used in a tew patients as

an immunomodulator.
Controlled clinical trials should be performed for treatments

being tried, as only then will a scientific evaluation of their
efficacy be possible.

9.4

DISCUSSION OF CLINICAL FINDINGS

1) From the above account it is apparent that tremendous effort
has gone into the clinical management of the thousands of patients who

poured into the dispensaries and hospitals following the disaster.
Medical staff worked round the clock, many of them suffering from

effects of exposure themselves.

2) The description reveals the involvement of several organ

I

systems in the body. This occurred even in the acute phase though it
was then masked by the severity of the effects on the eyes and

respiratory system.
The Environmental Health Criteria 27 (1983) states that few of
the non-biologic agents have unique effects on health and conversely
the effects considered may often be related to a wide range of

factors. Therefore many aspects of the situation should be takn into
account in trying to understan etiology and mechanism of action.

3) Inspite of receiving medical treatment people kept returning to
the clinics with histories of persistent symptoms or with relapses

following remissions. Recent studies and clinical experience provide

evidence of chronic effects.

49

4)

It was not possible to tit the apparantly unrelated symptoms

and signs with which people were presenting into detinate diagnosis.
In the toxic oil syndrome which occurred in Spain clinical
observations in the acute and chronic phase showed features resembling

those of wel1 known disease entities, Put the combined clinical
picture and pathology findings were unique and suggested that the

syndrome was new (Grandjean and Tarkowski, 1983). A similar situation

is seen in Bhopal. Here, unlike in Spain the exposure is much more
delineated in time and place. Though the exact composition of the
chemicals causing the morbidity and mortality is not full\ known,

there is certainty about it being MIC and its breakdown products or
metaboli tes.

It is the first instance of human exposure to such high

concentrationsof these chemicals. The event has resulted in a pattern
of morbidity which probably comprises a new disease complex.

I

5) It has been observed that maintenance of records for individual

patients regarding clinical presentation and treatment were a casualty
under the acute emergency conditions. This is a lacunae both for the
future treatment of the individual patient as well as for a thorough

understanding of the pattern of morbidity caused by the disaster.

6) A number of medical professionals and researchers have
documented the general clinical picture with which the affected people
presented to the clinics at varying periods of time after the

di saster. This does provide very valuable qualitative information
about the cases seen and investigated.

7) The approach in the documentation reviewed has been focussed

mainly on the individual" case" and on specific organs and systems.
Some of the limitations of this approach are

a)

Because only people who voluntarily attended clinics were

considered there is a self selection in the patients seen and

documented. This "selection bias" would result in the picture of

50

I

morbidity not being representative o4 the morbidity in the community.
However it would still provide descriptive information about the
morbidity produced in those individuals.

From the estimated 200,000 people exposed or at risk only a

proportion would have utilised the Govt, health services on which

most of the ICMR reports were based. Banerji et al (1985) have
documented his. Other possible health services that people may have

utilised are as follows :
- medical relief camps set up by a variety of volunatary agencies

- local private practitioners or registered medical practitioners -

special health schemes of which they were members eg ESI
hospitals, Rai 1 way health services etc.

- health services outside Bhopal

- other systems of medicine / healing prevalent in India eg ayurveda,

unani, siddha, homepathy etc
some may have utilised several systems of medicine

or some may not have utilised any service.
Factors affecting utilisation of services would include

accessibility in terms of finance and distance; severity of illness;
tolerance of symptoms; mental health status and the beliefs and
culture of the people.

b)

Cases reported only represent individual patients and cannot

be related to a population to derive rates. Morbidity rates besides
providing an estimate of the magnitude of the problem and its

distribution in the population, could also help one to identify
priority groups for treatment and care.

I

i

-r

■' -

OH-Uo - • i
O’

X.

A?.

'

I

CHAPTER IQ

REVIEW QF EPIPEMIQLPGJCAL STUDIES
INTRODUCTION

There have been very •few epidemiological studies about the
health impact of the disaster. Reports that are available have been of

studies conducted by non governmental groups. They provide important
information about the type and distribution of morbidity in the
community. The methodology and findings of these studies will now be
described in some detail. Findings from the population based, cross

sectional study by Banerji et al have been discussed earlier in the
section on mortality.

REVIEW OF STUDIES

10.1

Epidemiological study of eye and general morbidity
1)

Andersson et al (1984) conducted a survey in the first

fortnight (Dec 11-17,1984) "to assess possible long term visual

disability among survivors". The sample comprised of 8 clusters of
households, selected in different localities which had received
varying degrees of exposure. Two localities of similar socio economic

status, 15 and 17 km away from tha factory, were selected as the
control groups.Detai 1s of location of the exposed groups are not

known. The sample size consisted of 261 exposed and 91 unexposed
individuals. The sample size and distribution of the population in

each cluster is not known. It was stated that the sample size was
restricted because of shortage of time before the exodus from Bhopal

during Operation Faith, which disrupted daily life for some weeks.
Assumptions to determine sample size have not been mentioned. It was
observed that the worst afflicted families had left by the time of the

survey leading to an underestimation of effect. Method of sampling has

not been mentioned - it was probably not randomly done

52

Three

ophthalmologists (one with an interpretor) were the interviewers.

An attempt to maintain uniformity was made - standard questions.
method of examination and simplified nomenclature was used.

Findings :

The post exposure death rate (which was specified as the number of
deceased / the number exposed) in the worst affected cluster was 37..
There were differences in symptoms between the various clusters or

exposure groups :
- burning of eyes and throat and coughing were the most

frequently mentioned symptoms.
- vomiting was the third most frequently mentioned symptom in

clusters close to the factory.
- further away choking and shortness of breath was higher.
- collapse and unconsciousness was reported only in the cluster
second in distance from the factory. Among those unconscious.
there were few or no eye symptoms, upon recovery.

- signs of respiratory distress were most marked in this cluster
affecting about 207 of the community.

- over one half of this cluster demonstrated eye signs.
- fundal changes were more common in the exposed group
especially venous dilatation.

- there was no case of blindness, irreversible eye damage or
difference in age standardised visual acuity.

- there was a significantly higher proportion of people with

|

I

active eye infection in the unexposed communities ( 57. as
compared to 17 ). It was stated that this was possibly due to

widespread use of antibiotics in the week preceding the
survey. (Absence of secondary infection was observed by Bang
the respiratory system.)

53

in

1

- there was a similar incidence (this should be prevalence ) of

Bitot’s spots in the exposed and control groups reflecting a
similar nutritional status in the 2 groups.

- there was evidence of fairly widespread trachoma in all the
groups, chough very few active cases were seen.

Andersson et al (1985) reported on a 2 month follow up in the
clusters mentioned above. Among the exposed excluding one cluster to

which they had "no access M the follow up rate was 507.. In both the

II

exposed and nonexposed groups only 367. (131/360) were located and

reexamined.This is a-very high drop out rate. No information has been
given about the baseline or known characteristics or attempts to

follow up the dropouts. The clusters were enlarged and 490 people were
examined. No information is given about the

new examinees viz.

regarding their distribution according to localities, their
demographic structure, method used for their selection etc. Hence data
will have to be interpreted with caution.

I
Findings :

There were no cases of blindness, decrease in visual acuity or

defect in colour vision. There were no corneal scars in the original
group but 6 scars which could impair vision were detected in the new

examinees. It was not stated whether these were attributable to the
disaster. There was regression of the early healing seen in the first

I(

examination. There was one case of persistant unilateral corneal
oedema and 3 with complaints of persistent excess watering in an

otherwise quiet eye.

10.2

EoidfflUQjoqiciU ftudv of Q*nrr»l hfilth »tatu>
A.Patel et al (1985) conducted an epidemiological study of the

general health status of the exposed people.

54

It was a population based, cross sectional study, using an exposed and
a control group. The study was conducted 3 months after the disaster.

Post exposure mortality rates for the different localities, as given
in publications by the State Govt., were taken as indicators of the
degree of exposure. JayaPrakash (JP) Nagar, 100 yards from the factory

in the direction of the wind that fatal night, had an exposure

mortality of 2.34X and was chosen as the study population. Anna Nagar,
10 km from the factory with an exposure mortality of 0.327., was used

as the control group. Both areas were comparable with respect to

housing, sanitation and economic status of the population. Study
results showed that mortality rates were useful indicators of
exposure. However the crude mortality rates found in the study
population were much higher than those reported in the Govt.
publication. The study findings were:

r

i

J.P Nagar - 86.6/1000 population,
AnnaNagar -

7.9/1000 population.

Post disaster hospitalisation rates were also found to indicate
differences in exposure : J.P Nagar - 307.

AnnaNagar - 0.727.
Sample size determination was made on the assumption that morbidity

would be 157. in JP Nagar and 57. in AnnaNagar. With a 57. level of

significance and 90% power, a sample of 180 persons in each group

(exposed and control) was chosen. Persons of both sexes, more than 10
years of age were studied.
Numbering of all the households to provide a sampling frame was
already done by the ICMR and the same was utilised in this study. As

random selection of individuals was not possible, a random selection
of 50 household units was made to yield the required sample size.
A house to house survey was conducted. This consisted of the

following : a) a detailed history on a predesigned questionnaire.

55

Non standardisation or pretesting of the questionnaire has been

accepted as a limitation, and was reported not to have been done
because of shortage of time.
!

b) general clinical examination of all the systems, the
parameters for which had been defined.

c) pulmonary function tests

I

using Morgans electronic

spirometer set at BTPS. A trained investigator, with experience i n
field based studies carried out the tests.

d) estimation of haemoglobin percentage.

e) open ended questions on the people’s perception of the

health services available after the disaster.
Information about training of the interviewers has not been
given. They were not blind to the hypothesis as this was not possible
in any of the studies conducted in that situation. Group meetings were
conducted in the community to obtain consent. The people were informed

about the research

group - that they were not related to the Govt.,

nor were they providers of services, nor involved with the claims for

compensation. This would reduce the possibilities for "compensation
malingering" as claimed by some. It was found that members of the

particular sample chosen had not been included in any of the other
studies being conducted, thus ruling out the possibility of the
learning effect or Hawthorne effect.
The two populations were

sex structure, body surface area,

comparable with respect tc age and

history of chronic disease and

smoking. The exposed were slightly better off socio economically than
the controls.

There was a rather high non response rate of 297. in the

exposed group and 15% in the control group. However available
information about the non responders was collected. Their age and sex
structure was similar to the responders and 507. or more of them were
i

56

I
exposed. 607. and 50Z ot non responders in the exposed and control
groups respectively were out of town, while 257. were away for work.

There were no refusals. Repeated visits were made in the time
available to maximise the response rate (the investigators were a

group of people who had come from different parts of India and were
not resident in Bhopal). It has been argued that since the actual

difference in morbidity was much greator than the 10 /• assumed in

sample size calculations, a smaller sample size would have
demonstrated a difference and non response may not make such an

impact. Nevertheless the high non response would have altered the

process c

random selection

and it must be kept in mind that the

non-responders may differ from the responders with respect to the
outcome following the exposure in unknown and variable ways eg as

stated by Andersson et al if the more seriously ill were among the non
responders there would be an underestimation of effect.

I
Briefly the study findings are as follows :

Prevalence rates of 26 symptoms were measured in the exposed

and control group at the time of the study. Tests to see if the
differences were statistically significant were done.

The following 15 symptoms were found to be highly

significantly different, being higher in the exposed group : cough
with expectoration, breathlessness on usual exertion, chest pain/

tightness, blurred vision/photophobia, fatigueability, weakness in the
extremities, muscle ache, headache, tingling/numbness, loss of memory,
nausea, abdominal pain, flatulence and anxiety/depression.

The following 6 symptoms were significantly different : dry

l
I
I

cough, breathlessness at rest, watering of eyes, skin problems,

bleeding tendency and impotence.
The

ollowing 5 symptoms were not significantly different :

57

■k

•Fever, blood in sputum, jaundice, vomiting,blood in vomit and malaena.

As many as 637. reported all the important symptoms. Only 2.77.
reported exclusively pulmonary symptoms, while 35.147. did not report
any pulmonary symptoms. Every person in the exposed group reported at

least one serious symptom, but quite a -Few in the control group did
not report any.

There was a significant difference in the number of attacks of
respiratory infections in the month preceding the study. In the

I
I

exposed group it was often described as a continuous respiratory
problem. It was said that this could be a supportive finding to
indicate a state of lowered resistance or immunity.

Exposed women had a significantly higher rate of abnormalities
of menstrual flow, alteration in the length of the cycle,
dysmenorrhoea and leucorrhoea. The sample was too small to report on

abortions and stillbirths. 507. of exposed mothers in the exposed group

I

reported failure of lactation or a decrease in

exposure, compared to HZ in the controls.

milk output post

Impotence in men

was

reported by 8.17. in the exposed group and 0.7?Z in the controls.

f
';Y-£

1

On examination:

I
I

1

There was no difference in the resting pulse and respiratory
rates. The mean haemoglobin percent in both males and females was

significantly higher in the exposed group. There was no case of
cyanosis. This was stated to be a significant negative finding in view

of the findings of 877. with breathlessness on exertion, the raised
haemoglobin concentration and that extensive lung damage was expected

I
B

to have occurred. 9.47. of the exposed had crepitations and rhonchi in

the chest, as against 2.17. in the controls (P<0.025). This rate was

also stated to be too small to account for the much higher rate of

breathlessness on exertion.

!

58

Comparison of Symptoms reported by individuals In J.P Nagar and Anna
Nagar. (Expressed in percentage. Numbers^of cases are shown in brackets.)
bi

Symptoms

No.

J.P. Nagar %

A. Nagar %

P Value*

P

0.01

1.

Dry Cougn

27.70 (41)

14 48 (20)

2.

Cough wrth Expectoration

47.29 (70)

23.91 (33)

< 0.001

3.

Breathlessness at rest

10J3 (15)

ZB9

(04)

< 0.025

4.

Breathlessness on
usual exertion

87.16 (129)

35 50 (49)

< < 0.001

5.

Chest pain/tightness

50.0

(74)

76.08 (36)

< < 0.001

6.

Weakness in Extremities

65.54 (97)

36’95 (51)

< < 0.001

7.

Fatigue

81.08 (T20)

39.85 (55)

< < 0.001

S.

Anorexia

^6 2.1 438)

-28-26 X39)

^0^001

9.

Nausea

58.r0 (86)

16.66 (23)

< < O.OQ1

10.

Abdominal pain

53.37 (79)

25.39 (35)

< < 0.001

11.

Flatulence

68.91

(102)

25.36 (35)

< < 0.001

12.

Lacrimation

58.78 (87)

42.62 (58)

< < 0. 01

13.

Blurred vision/photophobia 77.02 (141)

33.40 (53)

<< 0.001

14.

Loss of memory for
recent events

45.27 (67)

11.59 (16)

< < 0.001

■nSVTS T81)

70728 "(28)

<< 0.001

T5. Tingling/numbness

•(a) P Values were calculated by X2 method.

Source

:

Patel A and Patel A,

(1985)

The Bhopal disaster aftermath :
an epidemiological and sociomedical survey.

59

(a)

Comparison of Symptoms reported by individuals in J.P Nagar and Anna
Nagar
(Expressed in percentage. Numbers of esses are shown in brackets )
(Symptoms significantly different but not analysed further)
SI.
No.

I
(

Symptoms

J P. Nagar %

A Naga- %

P Value

’(a)

0 01

1.

Skin problems

29.05 (43)

11.59 (16)

<

2

Bleeding tendency

9.45 (14)

2.89 (04)

< 0 025

3.

Headache

66.89 (99)

42 02 (58)

<< 0.001

Muscle ache

72-97 (103)

36.23 (50)

< < 0.001

5.

Impotence

8.10 (12)

u 72 (01)

< .05

6.

Anxiety/De precsion

43^2 (65)

10.14 (14)

<<0.001

Comparison of Symptoms reported by individuals in J.P Nagar and Anna
Nagar
(Expressed m percentage. Members- of oases are sMmi in brackets.) —•
(Symptoms - Non-significant)

Si.
No.

Symotoms

J P. Nagar %

A. Nagar %

P. value*

1.

Blond in Sputum

10.13 (15)

7-24 (10)

N.S.

'2.

Fever

27 70 (41)

• 28.98 (40)

N.S.

3.

Jaundice

0.67 JOT)

00

N.S.

4.

Blood in vomit/stcol/malena

12.16 (18)

10 14 (11)

N.S.

■5

Vomiting

11 48 (17)

5.79 (08)

N.S

(a)

I

(a) P Values were calculated by X2methdd

Source

: Patel A and Patel A, (1985)
The Bhopal disaster aftermath :
an epidemiological and sociomedical survey.

60

u

There was a statistically significant difference in pulmonary
function tests in both sexes in the age groups of 15-45 and 45-60
years. The difference in other age/sex categories were not

significant. However there were only a small number of observations in
these categories. The mean values of FEV1 and FVC and the FEV1/FVC
ratio in all age/sex categories were diminished in JP Nagar compared

to AnnaNagar. The 15-45 and 46-60 age groups showed a restrictive
pattern while the over 61’5 had an obstructive pattern.

It was stated that the control population was also minimally

exposed, thereby diluting or masking the effect of the exposure.

I

10.3

Epidemiological study of wpmen’s reproductive health
R.Bang (19850 conducted a study of the status of women’s

reproductive health 3 months post exposure. This followed the earlier

survey of a small number of women in the 2 affected slums (refer to
section on morbidity). The sample consisted of 114 women in 2 severely

affected areas and 104 women in a control area. Reasons for selection
of sample size have not been given. Selection of the sample was from

community based Ob/Gynae clinics. This introduces the problem of self
selection as women with Ob/Gynae problems would be expected to attend

these clinics. These cases cannot be related to any population or
denominator. Hence epidemiological extrapolations from these case
studies cannot be made.

It is not known if standardised questionnaires

or examination schedules were used.

61

The -findings of the study were reported as -follows :

Exposed group

Control group

Total no.studied

114

104

Pelvic ex am.done

72 (637.)

52 (507.)

Leucorrhoea

65/72 (907.)

14/52 (277.)

51.67

P.I.D.

57/72 (797.)

14/52 (277.)

34.67

Cerv.erosion/
endocervi citis

54/72 (757.)

23/52 (447.)

11.39

Excess menstrual
bleeding since
exposure

27/87 (317.)

1/81 (1.27.)

26.19

Suppression of
1 actati on

16/27 (597.)

2/16 (127.)

10. 17

Chi square

The di-f-ferences are all highly significant (P < 0.001).

The results shown are -from a smaller subset o-f the original
sample, as pelvic examination could not be performed in some women due

to various reasons like pregnancy; not being married; and refusals ie
i

a selection at this stage has also occurred. However inspite of the
limitations mentioned and also because similar factors of self

select!oi. occurred for both the exposed and control groups the
difference between them is large enough to suggest real differences in

the two groups and point to the need to study this area. Other
studies, subsequently, too have reported similar findings (Patel et

,1985 and Sathyamala,1986).
In the exposed group there was a history of spontaneous

abortion in 7, stillbirth in 4, threatened abortion in 1 and
■ I

incomplete abortion in 1 after the gas leak. No women in the

control

group reported any of these adverse outcomes of pregnancy.

Severe pallor was found in 37 (36%) of the control group but
only in 3 (37.) of the exposed group. This corresponds to the finding

of an increase in haemoglobin percentage in the exposed population

62

I

f ■
■W'

■Found in other clinical and epidemiological studies.

F

10.4

Epidemiolpgic^J study pt outcome pf pregnancy
Sathyamala C (1986), conducted a community based study of

pregnancy outcome, 10 months post exposure. A large sample was needed
&

to detect significant differences in rates of abortion and stillbirth.

I
r

The sample size took into account a non response rate of 257. which had

been found in earlier studies. A total population of 8165 people in
1632 households were surveyed. Details regarding assumptions to

determine sample size, power of study etc were not reported. Three
exposed localities (bastis) were selected on the basis of post

exposure morbidity and or mortality rate. These were as follows :

I

JP Nagar- Mortality rate 65.3/1000, Morbidity rate 667.

S

Kazi camp -

I

Kenchi chola- "

H

H

ii

46.7/1000,
35.7/1000,

ii

’* 54-607.
" 91.97.

These figures were taken from later 9 unpublished analysis of the

study by Banerji et al. The sampling frame provided by the ICMR was
utilised and random sampling of households done. A" historic control"
was utilised ie history of pregnancy outcome in the year preceding the
disaster, in the same population was used as a comparison. This was

chosen on the basis of studies carried out elsewhere which

demonstrated an abortion recall of 827. accuracy even after a lapse of

I
I

si~

I

10 years. This may have lead to an under reporting in the controls and
an overestimation of the difference between the groups.
A pre-tested questionnaire was used. Methods used to train

interviewers and to avoid interviewer bias have not been mentioned.

I
1
I
I
?- I
'

The definition of abortion, missed periods and delayed periods used
for the purposes of the study have not been mentioned.



Tb
p
I

Misclassification between the three could possibly occur
Findings :
The non-response rate was 22 7., within the limits of what had

63

I

been considered in sample size determination.

There were 275 live births and 13 still births in the
population after the gas leak. The birth rate was stated to be
33.68/1000 population and was said to be comparable with the national

birth rate. However births for only 10 months were taken to calculate
the rate. Normally a period of 12 months is used and hence the rate

calculated would be an underestimation. It is also the crude birth

rate, not being standardised for the age and sex structure of the
population. The stillbirth rate post exposure was found to be
47.27/1000 live births. However live and still births together should

be taken in the denominator. The rate then is 45.25/1000 births. No

comparison with national, regional or study based stillbirth rates has
been made.

I

The overall spontaneous abortion rate after the gas leak was

370.96 which was statistically very significantly higher than the
spontaneous abortion rate of 32.178 before the gas leak.

Rate of spontaneous abortion before and after the gas leak (GL_)
Before GL

After GL

Total conceptions

404

310

No.of abortions

13

115

Abortion rate

32.178/1000

370.96/1000

conceptions

conceptions

A second important finding is that the rate of spontaneous

abortions in women who conceived after the gas leak is again
statistically highly significantly greater than the abortion rates

before the gae leak- The increase being about 5 times greater than
before the gas leak.

64

flbprtion r*tg in conception* btfort 5L which »t?prted befpre 6L

end »bortion r»te in conceptions efter 61
Concep.

Abortion

Abortion

Concep.

Abortion

Abortion

BGL

BGL

rate

AGL

AGL

rate

404

13

32.178/1000

310

45

145.16/1000

conceptions

conceptions

(BGL- before gas leak, AGL- after gas leak)
The overall foetal death ratio was statistically

significantly increased in the year following the gas leak in
comparison to the previous year.

Foetal death ratio before and after gas leak

1985

1984

I
Quarter

Number

Number

F D

Number

Number

F D

deli vered

aborted

ratio

deli vered

aborted

ratio

30

2

6.66

76

27

35.52

87

12

13.79

77

24

31.16

56

3

5.35

94

20

21.27

(LB+SB)

JanMarch
April-

June
Jul y-

I

Sept.

(LB - live births, SB - still births;

FD ratio - foetal death ratio )
The foetal death ratio has not been defined but appears to be the

number of abortions per 100 live and still births.
While past obstetric history, parity, period of gestation at

the time of abortion etc were measured they were not taken into

consideration in the analysis. These are important interactive and

confounding variables.

65

I

Changes in regularity ot the menstrual cycle, delayed and
missed periods, length of cycle and type of flow, were also found to
be statistically significantly different before and after the gas

leak.

10.5

C»se rgfsrent >tudy pf wttrino o* the tyf»
Andersson et al (1986) conducted a case referent study of

persistent eye watering. An eye hospital started in Bhopal in response
to the disaster was used as the source of cases and controls.

Two consecutive retrospective series of clinical records were

drawn, for outpatients on whom cynosure data were available. This

i

would be a source of selection bias as it is probable that exposure
status may not have been recorded equally in the exposed and

non-exposed groups. The method by which exposure was assessed and

recorded has not been mentioned.
Findings :
Gas exposed people were three times more likely to present with

watering eyes ( odds ratio -OR- 2.96, 95 7. confidence interval

CI

2.3 - 3.4) and nearly 4 times more likely to present with watering and

at least one other irritant symptom (burning, itching, redness)

( OR

3.8, 95 7. CI 3. 12 - 4.4). There was no association between exposure
and refractive errors ( OR 1.16, 95 7. CI 0.83 - 1.9).

There is no explanation for the symptom of persistent
watering of the eyes. The report suggests tear film instability due to

long term effect of exposure on epithelial maturation or abnormality
of the mucus component of the tear film which is derived from the

-I

epithelium itself and from conjunctival goblet cells.

66

I

In nummary, the epidemiological investigatons conducted have
studied different aspects of the health impact of the disaster at

different points in time. They vary in methodology used and critical
comments regarding this aspect have been given above. They were
conducted in difficult circumstances and despite some methodological
limitations they all record very serious effects on the health of
those exposed. They support clinical findings of multisystemic and

long term effects. However some of the important findings from these
studies, that may provide clues for etiology, if followed up are -

A) the varying pattern of morbidity in clusters at different distances

away from the factory in the acute phase. This was not just in
magnitude of effect, but there were qualitative differences of

differing symptomatology (Andersson et al) in different clusters. This

points to the possibility of the role played by different chemicals.
Follow up studies should look at different clusters over time.
B) the presence of a percentage or proportion of individuals with
multi systemic symptoms in the absence of lung disease (Patel et al) in
the sub-acute phase. This suggests that severe lung damage may not

account for all the chronic effects.
C)

significantly higher adverse outcome of pregnancy in exposed women

conceiving after the disaster, compared to controls. Congenital

abnormalities also need to be studied. This very serious observation

points to the presence of continuing toxicity.
Infants, preschool and school age children, a vulnerable group

have not been studied. Respiratory disability has not been

studied at

the population level. Natural history of the morbidity and the excess

mortality that continues to occur also remain

67

to be studied.

SUfllfPirY itblf of eDidgmialoaicil studiei conducted in Bhopal

Investigator

Focus
of study

Type of
study

Time

1.Andersson
et al

Eyes, gen
morbidi ty

Population
based,
clusters +
controls,*
2 month
follow up

2.D Banerji
et al

3.R Bang

Place

Person

Sample
size

Jj^tiodological iiotctB
Sampling
method

Non
response

Study
instrument

December Severely ♦ General .261
1984,and mcderately popula- exposed,
Feb.1985 + mildly
tion
91 un
exposed
exposed
areas
persons

opportunistic
sample ie as
many as could
be examined

64 7. at
fol 1owup

3 ophthalmologist
intervi ewers,
attempts to maintain
uniformity in history
taking and
examination

Mortality, Population
general
based,
features. cross sectional

January
1985

Severely * General .700
Moderately popu)a- households
exposed
tion
areas

Random
sampling

?

pre-designed
questionnaire,
trained investigator

Womens
Clinic
reproduct­ based case
ive he th series in
exposed *
control
areas

Feb-Mar
1985

Severely
exposed ♦
control
areas

Womenreprod
uctive
age
group

.114
exposed,
104 un exposed
persons

self selected
sample,women
attending
field based
Ob/Gyn clinics

pelvic
exam.
not done
in 43.27.

1 gynaecologist
investigator,
routine history
taking and
clinical exam.

4.A Patel
et al

General
health

Population
based,cross
sectional,
exposed +
control
areas

March
1985

Severely
exposed *
control
areas

General
populati on
>10 yrs
of age

180
persons
in each
group

Random
sampling

29 % in
exposed,
15 X in
control
group

Pre-designed
questionnaire
with defined
parameters,
? training of
investiqators

5.Sathyamala

Outcome
of
pregnancy

Population
based,
cross sectional,
hi stori c
control

Sept
1985

3severely pregnant
exposed
women
areas
out of
general
popula­
tion

8165
persons
in 1632
househol ds

Random
sampli ng

22 7.

Pre-designed,
pre-tested
questionnaire,
? training
of investigators

wateri ng
of eyes

case
control,
record
based

Nov’85
-Jan86

eye
hospital

989

those with
recorded
exposure
status



6.Andersson
et al

eye
patients
from
gen. pop
|-ulation

hospital case
records

l

I

CHAPTER H
EXPERIMENTAL STUDIES
11.1

Pr>- difittr

Data on the toxicology of MIC was scarce at the time of the
disaster. Median lethal doses in animals were available eg. it was 5
ppm for 4 hours by inhalation in the rat. In another experiment a dose

of 62.5 ppm for 4 hours killed all the exposed rats. Corneal injury
has been recorded in rabbits.
A dose ranging study in human volunteers has been referred to

by the ACGIH ( American Conference of Governmental Industrial

Hygienists). There were no effects at 0.4 pprr but exposure to 21 ppm
was unbearable.

Mention has been made of the intense irritation caused to
eyes, nose and the throat. Kimmerle and Eben (1964), studying MIC

toxicity by inhalation exposure, observed that it was highly

I

irritating to skin and mucosa and that it produced pulmonary oedema.
There was little published material on the effects of

sublethal doses, dose response and metabolic/chemical breakdown
products of MIC.

11.2

Post- disaster

Several toxicological studies on different animal models have

been conducted after the disaster. Because of the short life span of
the animals used, each animal year being equivalent to several years

of human life, an estimate of long term effects of exposure can be
made relatively early. Experiments and pathological investigations not

ethical or permissable to be conducted on humans can also be
performed. The main limitation of animal studies however, in general
is that extrapolation of results to humans has to be made with caution
!•

I



because of the differences in the biological systems. Another limiting

68

I
■factor to be borne in mind in this particular case is that in all the
animal experiments conducted so far, only pure MIC has been used as
the agent of exposure. In Bhopal under the prevalent conditions of

high pressure and temperature and in the presence of catalysts other
chemical reactions could have occurred with the formation of other
I

chemicals. However the advantage is that these experiments can

I
I

indicate lesions attributable to MIC. They can be used to support /
explain epidemiological observations and similarly epidemiological
data can provide clues for experimental work. The objective of both

endeavours together, being to explain mechanisms / pathogenesis to the
extent necessary for rational interventions in the treatment

and /or

rehabilitation of victims and in the prognosis of their condition.
I

The method of MIC exposure used in animal studies has been by

inhalation, with doses varying between experiments. They all tried to
simulate the possible dose range that could have existed during the

Bhopal disaster
Harding et al (1985) reported the development of lens
opacities or cataracts when rat lenses were incubated with MIC.

Salmon et al (1985) reported that at low concentrations in

rats MIC caused severe sensory irritation with slow, irregular
breathing and the production of a sedative effect. At higher

concentrations this was masked by arousal resulting from respiratory

distress. Eye damage was always confined to the epithelial layer with
most severity at intermediate exposures suggesting that at high doses

some protective response was evoked. Urinary thiocyanate levels in the
exposed were lower than in the controls. They observed a dose

I

dependant response and supported the use of death rates and incidence
of pulmonary damage as a crude index of exposure in epidemiological
studies.

Nemery et al (1985) reported that at

I
i

I

69

ery

high concentrations

(10 mg /L for 15 mins) 50 7. of the rats died. The lungs were enlarged

with air. Gross oedema or haemorrhage

was present only in 2 rats

killed after exposure. The main effects of low concentrations of MIC

on the respiratory tract was to injure the proximal airways with

little alveolar injury. At high concentrations lung parenchyma was

also damaged with resulting interstitial and alveolar oedema,
inflammation and haemorrhage. Though there was complete destruction of

bronchiolar epithelium, repair took place. However despite rapid

resolution, they -Found isolated foci of more recent injury in animals
killed 2-3 weeks after exposure. They found MIC to be a respiratory

I

irritant ie both a sensory (stimulation of nerve endings in the nasal

1

mucosa) and pulmonary irritant (impact on lower respiratory tract ).

Ferguson et al (1986) in mice experiments also found

MIC to be a potent sensory and pulmonary irritant. They have
considerable experience in working with isocyanates and have found MIC

to be the most potent pulmonary irritant they have tested in the

isocyanate series. They found that the RD 50 (the concentration

evoking a 507. decrease in the respiratory rate) and the RD 50 TC (the
RD 50 in tracheally cannulated mice) was seperated only by a factor of
1.5. Thus a concentration capable of evoking intense

sensory

irritation of the eyes ,nose and throat is close to that capable of
inducing pulmonary irritation. MIC is thus classified as a respiratory

i rritant. They found it to be 7 times more potent than chlorine.
Luster et al (1986) found a steep dose response for toxicity.

During 90 day recovery studies epithelial injury generally resolved,
but prominent fibrosis developed in the walls of the major bronchi.

They reported no injury to the spleen, liver, kidney, thymus or brain.

Haematological values except for slightly increased haematocrit were

within the normal range. They found humoral immunity to be unaffected.
Inspite of a 30% suppression in T cell lympho-proliferative response

L

70

they found host response resistance not affected.

Fowler and Dodd (1986) studied rats, mice and guinea pigs.

Gassert (1986) observed that this study was the most comprehensive
inalation study of MIC to date. It was produced some years before the
Bhopal disaster under private contract 48 with Union Carbide but was
not published until 1986. It provided evidence of bronchiolitis

I

obliterans in guinea pigs (only) exposed to 10.5 and 5.4 ppm MIC for 6
hours. They also noted dose related lesions in the respiratory tract.
No deaths occurred in animals exposed to 1 or 2.4 ppm MIC. The

majority of deaths for 10.5 and 20.4 ppm occurred through post
exposure day 3 ; at 5.4 ppm deaths occurred throughout the 14 days.

Deaths were attributed to pulmonary vascular alterations.
ICMR studies (1985) found that the cherry red appearance of
the blood could be due to the direct action of MIC (by carbamylatuon)
and need not necessarily be due to cyanide or carbon monoxide. Carbon

monoxide poisoning was ruled out. Analysis of human tissue by gas
chromatography indicated the presence of monomethylamine. On animal

studies they found that MIC had an LD 50 dose of 85 mg in mice, but

with thiosulfate therapy it shifted to 195 gms. For rats the figures
were 270 and 344 respectively. Normal rabbit lungs weighed 6 gms.
following MIC exposure they weighed 29 gms and had a large number of

haemorrhagic patches. When given sodium thiosulfate immediately after
i

MIC exposure the lungs weighed 24 gms but the appearance was normal.

/

With pure MIC they also found a dose dependant response in the
respiratory tract. They found that MIC had bactericidal activity.

Salmon (1986) also reported that MIC could produce a reddish
tinge to blood. However differences could be detected on spectrometric

analysi s.
Varma et al reported adverse effects on the oestrus cycle and
fertility in male and female mice

71
t

F
i

Gassert et al (1986) reported on a 14 month follow up of rats

exposed to MIC. Two exposed rats died at 6 and 8 months following
i

sudden onset of respiratory distress. 6 rats killed at 14 months
revealed a history of mild respiratory infections. Mild interstitial
fibrosis in the peribronchiolar region was present in all exposed

I

H

rats. A notable finding was that MIC exposed animals had 4 times the
amount of lymphoid aggregates found in control animals - adjacent to

the bronchiolar airways. A mild infiltrate of eosinophils was present

5

5-

in the bronchiolar mucosa. Eosinophil and lymphoid infiltrates were
found in the mucosa of the conjunctiva of the eyelids and perilimbal

regions. They state that long term changes in the eyes and lungs may
result from a single 2 hour exposure to acute sublethal doses of MIC

i1

vapours and that the immune system is most probably directly involved.

They suggest that lymphoid hyperplasia may be due to persisting
exposure related antigens or to an increased susceptibility to other

immunostimulating agents following MIC exposure.

H

Thus animal experiments reveal that MIC is extremely toxic on
inhalation - being a potent respiratory irritanaft. Chronic morbidity
and a continuing increase in mortality has been reported in the

exposed animals. The studies suggest 3 possible mechanisms by which
this may occur :
a) due to long term sequelae of severe lung damage caused by the

I
•3

i

direct toxic or irritant effects of the chemicals.
b) due to damage to the immunological system.
c) due to systemic toxicity caused by mechanisms as yet unknown.

j

1


i
4

72

I
CHAPTER 12

DISCUSSION

The main points arising from the description so far are
summarised. This will be followed by a discussion of the main

methodological points pertinent to an epidemiological understanding of

the health impact of the disaster. This will also be relevant to the

conduct of epidemiological studies in the future. Discussion of

methodological issues that have been interspersed in the report will
not be repeated here.

12.1

I

SUMMARY QF FINDINGS
1) A population of about 200,000 persons were exposed at a

point in time and at a localised place to an airborne toxic chemical
agent.
2) The composition and characteristics of the chemical agent/s

are not fully known. Studies are being conducted to elucidate this.
These include analytical chemical studies, toxicological studies on
animals etc.

3) The exposure resulted in immediate and continuing mortality
and morbidity in the population at risk.

4) The mortality was substantial, though the exact numbers are

not precisely known.
5) Clinical investigations show that the morbidity that has

occurred is complex and affects several systems. It does not fit
easily into previously known disease entities. The combined clinical
and pathological findings probably comprise a new syndrome.

6) Epidemiological studies indicate a large community load of

morbidity. This appears to vary with factors like locality, degree of
exposur

and time- Besides manifestations in the acute and subacute

[
73

I

I

phases, long term effects have also been observed. Disease occurrence
and distribution according to characteristics of person need furthur

study.

7) There have been several theories proposed to explain the
mechanisms causing the morbidity viz.
- sequelae of direct damage to lung and eyes which followed exposure

/

to the toxic chemicals,
- alteration of the immune system following the exposure,

- systemic toxicity caused by the toxic chemicals or their

metabolites in the body,’
- psychological trauma caused by the disaster,
- a combination of all the above.

I

12.2

OUTLINE PF DISCUSSION
The discussion on methodological points will cover the

following areas : a) the exposure variable,

b) the populatic

at risk,

c) the health outcomes,
d) confounding variables and

e) sources of bias.

12.3

THE EXPOSURE VARIABLE

By now, results from the few early studies conducted

together with experience of physicians and social workers in Bhopal
and toxicological studies in animals indicate that the exposure has

resulted in long term adverse effects on health. These findings point

to the need for long term follow up of the victims. As a first step
valid measurements of exposure need to be evolved.
a) Defining exposure

It is necessary in the conduct of epidemiological studies in

74

I

Bhopal to have a wording definition of the exposure variable.

Indicators or measures of the degree of exposure are also needed to

estimate possible dose dependant responses in the outcome variables of
mortality, morbidity and disability.
Previous studies have used the following as indicators of
exposure ;

- post exposure mortality rates in defined localities as reported by

the State Govt. Study findings have shown that these did provide a
rough estimate of exposure in different localities. The rates found in

the studies were however much higher than the rates reported by the

Govt. I sues concerning mortality rates have been discussed earlier.

I

- a combination of death in the family or exposure related mortality
rates along with grades of morbidity as a measure of exposure.
- one study found that immediate post exposure hospitalisation rates

were also related to the degree of exposure.

b) Variability in exposure
Epidemiological studies reveal that control areas 10 kms away
from the factory have been mildly exposed. Studies have also shown a

variabilitv in the picture of morbidity in different localities as
well as a variability in individuals or groups of people living in the
same locality. Besides differences in individual susceptibility

accounting for some of the variability, both the above observations
su^Test that the factor of "exposure" needs to be considered more
carefully. The two important issues to be considered are :

- the area and hence the population exposed may be larger than
the accepted 200,000,

- several variables determine the exposure level for an

individual. Results from the various studies have indicate
that these are :

1) distan

I

from the factory at which the individual was at the

75

time of the disaster, taking into consideration the direction

of the wind,
2) type of housing - pucca (well built)

- kutcha (without brick and cement)
- presence of gaps/holes letting in air,

3) action taken at the time of the disaster viz.

a)

measures of exposure to the atmosphere :
- kept all doors and windows closed and remained indoors

- opened doors and windows, stayed in the house
- went out, remained in the area

b)

measurts of exertion :

- left area, walked
- left area , ran
- left area, cycled

- left area , used motorised transport
c)
(

use of neutralising / protective measures :

- used a wet cloth over the face
- covered face with a blanket

- went in a direction opposite to that of the wind.

Thus a single parameter by itself eg distance away from the

factory, may not reflect the true exposure status of the individual
which would also depend on other actions that the person took at ^he
time of the disaster. This could be one of the reasons to explain the
variability in mortality and in the pattern and degree of morbidity in
different individuals even in the same locality. Other factors like

age, level of nutrition and general resistance, presence of other
diseases etc would also play a role. All the above will have to be

considered in studies of morbidity as well as in determining priority
groups of people who would need greater care and follow up.

I
■-

I

76

I

c) Exposure at individual and population level

Mortality rates could be a measure of exposure to classify

localities and areas ie. they could be used as indicators of degree of
exposure at the population level. While the other factors outlined
above could be used as measures of the exposure status of individual s.

d) Measurement (assessment) of exposure

History taking is the traditional medical method of
determining the

exposure status of an individual. However in Bhopal a

large population has been affected. Several studies into the health
effects will need to be conducted over a long period of time and
several interviewers will be involved. To ensure comparability between
studies and consistency over a period of time, a standard, repeatable
and valid method of determination of exposure should be used.A

standardised questionnaire, using the factors discussed earlier would
provide a simple, inexpensive, noninvasive tool of investigation.

There have been attempts to develop biological markers of

exposure eg antibodies or enzyme related markers. They are still in
the experimental stage and will have to be field tested. However any

invasive method - in this case blood samples will be needed - have the
drawback of increased nonresponse. Besides this, increased costs, the

need for investigators who have requisite skills, the availability of
laboratory facilities etc

will have to be considered. Studies

carried out so far have shown that the use of crude morbidity and

as markers of degree of exposure. Salmon

mortality rates have served

et al (1985) have confirmed this on the basis of experimental studies.
With a little refining as suggested above, standardisation and
pretesting, questionnaires could continue to be used to measure the

degree of Pvposure.

77

I

12.4

THE POPULATION ftT RISK
The population at risk would comprise all those who were

exposed to the agent and who could potentially aanifest adverse health
outcomes as a result of the exposure. It would 4or«i the denominator in
calculating exposure related rates of morbidity and mortality for the
population. Various subgroups of this population could also be studied

eg according to age, sex , socio-economic status, degree of exposure
etc. Epidemiological profiles for groups broadly classified as

severly, moderately and mildly exposed could be built up. Factors
discussed under exposure variable will have to be considered.

Numbering of all the households to create a sampling

:|

I

framework was done shortly after the disaster. Since a relatively
small population has been affected and there is a need for long term

follow up, a population register or case registers could be maintained

on computer after a census of the exposed population.
This would provide a good base for fol1ow up studies.

12.5

THE HEALTH OUTCOME ■ MORTALITY

a) Mortality rates / standardization
The number of deaths following the disaster would have to be

related to the exposed population to derive crude rates. These could
be standardised for age and sex by comparison with a standard
population of similar socioeconomic status, and Standardised Mortality

Ratios (SMR’s) could be calculated. The time period during which

I

deaths are enumerated, would have to be considered in the calculation

of exposure related mortality rates. As with morbidity this could be
calculated for the acute, subacute and long term phases. Rates for
different localities should also be calculated.

These rates could be calculated using routine sources of data.

L
f

78

u

However in the longitudinal study, lite table analysis could be done.

There should be a good reporting system for deaths in the exposed and
control populations. Staff and investigators should be trained in the
use of the International Classification of Diseases and if necessary
suitable, standardised criteria could be evolved for the

classification of deaths. Autopsies should be performed in a sample of
deaths among the exposed group, as is the reouirement in any medico
legal case.

12.6

THE HEALTH OUTCOME - MORBIDITY
a) Assessment criteria- This has been the first time that a

whole population has been exposed to high concentrations of these

chemical agents. The exposure has therefore resulted in a group of
symptoms and signs which together do not fit easily into established
disease entities. This new disease complex would have to be named

appropriately eg the "Bhopal Toxic Gas Syndrome". For the purpose of

I

epidemiological studies working case definitions of this disease
complex would have to be developed. This would have to be done based

I

on the clinical experience of medical professionals treating the

exposed population together with the help of epidemiologists to ensure

simple, standard criteria which can be applied in the field. It would
basically comprise of groupings of character!stic symptoms and signs.

b)

There may be a lag period between the exposure and

some

pathological conditions which have not as yet manifested. Rothman
(1985) states that one must allow for the following
- a biologically appropriate induction time during which a sufficient
cause becomes complete. This may be quicker for heavy exposures and

slower due to interaction with other factors for lower doses of

exposure.
- latent period

which is the period after causation before the

79

u
I

disease is detected.

Early studies may thus miss still evolving disease conditions which
could be picked up by prospective longitudinal studies or

I

epidemiological monitoring systems.
c)

Complementary causes or predisposing factors would play a

role in the development o-f the disease outcome by increasing the
susceptibility ot individuals. People with a larger set of

complementary causes would need a smaller dose of exposure to complete
a sufficient cause and result in a diseased condition (Rothman 1986).
Exposure to the toxic chemicals may unmask or exacerbate existing
disease eg chronic bronchitis, asthma, TB etc. These would be

considered confounding factors in the analysis of studies. But, from
the point of view of the health condition of the people and for the
provision of health care services, their presence would cause the

individual to be placed in a priority group.

[

12.7

CONFOUNDING AND INTERACTIVE VARIABLES

As discussed previously socioeconomic status is closely

related to exposure and to outcome and would be a confounding factor.
Stratification in design or group matching could be used to account

for this. Age and sex would also have to be considered.

In the Bhopal situation, smoking, exposure to smoke or air

pollution in the home (cooking on smoky fires) or at work, nutritional

status, presence of chronic diseases eg TB, trachoma, asthma ,chronic
bronchitis would be interactive factors which would have to be

measured and allowed for in the analysis.

12.8

SOURCES PF BIAS
Several sources of bias have to be considered :

a) Stewart (1985) has raised the issue of^iurvivor bia«Hin follow up

80

studies of survivors of the atomic bomb explosion in Hiroshima and

Nagasaki. This could occur in any cohort of people surviving a major
catastrophe. The parent population loses a high proportion of

vulnerable individuals - the very young, the old and the sick. Thus

when comparisons of mortality are made with a control group in follow
up studies there will be an underestimation of the effect. One may get
a normal death rate in the survivors, though it may actually be

slightly raised. This is similar to the bias caused by the "healthy
worker effect" in studies of occupational groups. This factor would

have to be kept in mind in long terr. studies in Bhopal.

b) As discussed previously there would be a aelection bias in hospital

or clinic based studies due to self selection of people attending

these services. With the plurality of services and factors of

accessibility this would be important in Bhopal.
The utilization of health services in the subacute phase as reported

I

by Banerji et al (1985) shows that this occurs.

c ) Bias due to migration of people into and out of the populati on, new
births and deaths, all of which would affect the baseline population
have been considered earlier in the report.

d) Misclassification of exposure status or of outcome (if the diseased

condition is undiagnosed or misdiagnosed) will enhance or decrease the
association depending on the direction of the misclassification. In

Bhopal this is very likely when using routine sources of data, as many
medical professionals dealing with a previously unknown situation,

have tended to use the nearest known diagnosis to fit the presenting
symptoms and signs. This re-emphasises the need to have a working
defination of the outc -ne for documentation and study.

81

e) The non-respons« rate has been found to be quite high (20-297.) in

I
all the studies conducted in Bhopal. Besides altering the sample size
and the power of the studies, this would also affect the composition

of the sample, depending on the characteristics of the non responders.

l

Allowance for non response should be made in determination of sample

size and also in budgeting for time and finances to allow for more

intensive follow up of a percentage of the non-responders.

f) Observer bias leading to a bias in history taking, recording,

interpretation of finding

or in diagnosis may occur. The factors that

play a role specifically in Bhopal are
- Those who believe that all is well in Bhopal try to underplay or
explain away the symptoms of the people. This is evident in the

attitude of many who attribute every symptom to the presence of
chronic diseases or as psychosomatic symptoms or as compensation

malingering. On the other hand those who believe that a conscious anti
people crime has been committed in Bhopal may let tiieir beliefs affect
reporting or interpretation of what the people say.
The above factors could be reduced by the training of

interviewers and in the use of blind techniques when possible in
certain investigations eg in reading xray films etc. Keeping

investigators blind to exposure status is not possible.

g) Measurement bias would be important to keep in mind especially when

using instrumentation eg spirometers for lung function tests.

Standardised instruments and techniques are available. The instruments
should be calibrated and maintained to give accurate and reliable
readings over a period of time.

82

I

u
I

12.9

8U6GESTIQN6 FOR EPIDEMIOLOGICAL STUDIES IN BHOPAL
Several research projects, involving different specialities,

I
are being undertaken in Bhopal and elsewhere, on various aspects of

the disaster. As outlined earlier there is a need for supportive

epidemiological studies, especially those that are population based. I
have not attempted to design a specific study while based here in

London. More recent information would be available in India,

especially regarding the studies set up by ICMR. Also, more details
about the affected population, the numbers in different localities,

their demographic structure etc would be available in Bhopal. By
meeting local medical practitioners and research workers, an
assessment of the present status of morbidity could be made, which
would provide important clues for study. It is hence more appropriate

as well as practical to plan a detailed study design based locally in
India, at Bhopal. However, possible study designs and

methodologi cal

issues concerning longitudinal studies have been delineated.

In Bhopal a cohort of people have been exposed at a point of

time to chemical agents. There is a need to study :
1) the range of health effects stemming from the exposure,
2) the natural history of these health effects.
An epi demiolrm cal study is basically an exercise in quantifying
disease occurence and using a logical method in deriving inferences /

explanations to account for variations in disease distribution by
relating them to putative causes. In this particular situation, where
the exposure has defined time and place characteristics, though the
composition may be uncertain, observational follow - up or

longitudinal studies seem logical. Here the study population are
selected with refere

e to their exposure status.

83

I

I

The application of case control studies, where the study

population are selected with reference to their disease status, would
be limited. The disease outcome in Bhopal, is not a well defined

entity and is still evolving. The prevalence of what has occurred, is
not rare, but affects 30 - 60 7. of the severely and moderately exposed

population. Also, a large proportion of the local population of

similar socioeconomic background have been exposed to the agent, to
some degree.

12.9.1

Cross sectional >tudiey in the sub acute phase have provided

prevalence rates of various symptoms and have also indicated areas of

importance. A repeat cross sectional study
i

could give prevalence

rates of symptoms and signs 3 years post exposure. It could provide
age, sex and area specific distribution of the" Bhopal toxic gas

syndrome". Relationships with aspects of the exposure variable could
also be tested.

Cross-sectional studies using exposed and control

groups, a variant of case control studies, could be used for

analytical purposes, to study the relationship between symptoms or
groups of symptoms and exposure.

12.9.2

Longitudinal study
A cross sectional study should form the baseline for a

prospective, longitudinal study. Important points in the conduct of a

longitudinal study are now

1)

considered .

Objectives
The hypothesis should be explicitly stated. There is a need to

define the time period of the study. This would be selected based on
biologic assumptions of the disease outcome and its relationship with

the exposure.
The broad objectives could be :

&
r

84

u

a) to study the prevalence / incidence o-f the "Bhopal toxic gas
syndrome".

b) to relate symptoms / signs observed at the start or appearing

during the course erf the study, to various aspects of the exposure.
c) to study the natural history of the condition - its severity,

fatality, the impact of therapy etc.

Other specific areas to be studied are
- a) the percentage of exposed individuals with multisystemic
symptoms and signs in the absence of lung findings.

- b) levels of urinary thiocyanate in the exposed and control
population.
c)

prevalence of psychiatric disorders in the two groups.

d)

prevalence of respiratory disabilities

- e) outcome of pregnancy in the years following the disaster.

The specific parameters of these outcomes would have to be evolved

locally.

2)

Sample

Small clusters in different localities could be selected to be

able to study the variation in outcome in the different localities.

Other aspects of the exposure variable would also have to be measured
in the individuals in these clusters.

Sample size determination would have to be done locally, with
details of the baseline population. Differences in the prevalence of

<

symptomatology between the exposed and control groups, as found in
previous studies, should be used for the calculation. The level of
statistical significance and power required for the study should be

decided. The high non response rate as found in previous studies and

possible drop out rates

should be considered .

Method of sampling or rampling procedure - to ensure

85

u
I

representativeness and to avoid bias, population based, random

sampling should be used. With this method the probability o-f selection
into the sample is the same for all individual units.

Though the

sampling framework has been set up, it could be rechecked keeping i n

mind the discussion of the population at risk. The method used in the

I

cross-sectional and longitudinal studies would have to be a house to

house survey.

Community meetings as in the study by Patel et al should be

conducted with the people to inform them of the study, discuss with
them the need for continued study, reasons for random sampling etc.

3)

Ethical aspects
Obtaining consent from study participants and maintenance of

confidentiality of patients records should be planned for.

4)

Study population

Besides the general adult population, infants, preschool and
school age children should also be considered. This group has not been

studied in the studies reviewed. The advantages of this group are that
they would have had no serious exposure to smoke (though passive

smoking would have to be considered) and occupational pollutants.
Chronic disease would also be minimal.

Their respiratory systems are also more sensitive to insults which

makes it easier to detect adverse effects. It has been found that
children can carry out spirometric lung function tests from about 7

years and can manage a single measurement of PEER at 5 years (Florey
and Leeder 1982).

5)

Control population

An unexposed or minimally exposed population is needed for
comparison. It should be comparable in terms of broad socio- economic
characteristics. It would provide an estimate of disease rates

expected to occur in the absence of exposure.

86

I
6)

Measuring the expoeure and outcome variable*
The main issues regarding these variables have been discussed.

Working criteria / case definitions for the assessment of exposure and

outcome need to be defined. Several types of outcome can be observed
eg post exposure mortality, specified decrease in lung function, onset
and frequency of respiratory infections in addition to those mentioned

under specific objectives. Sub-classification into definate

probable

and ppossible" cases" could be made.

Criteria and methods of assessing the exposure and outcome

variables should be the same in both the exposed and control groups.
7)

Examination techniques
Simple, valid. repeatable, field tested instruments will have

to be used. These would include :
questionnaires

clinical examination
lung function tests etc
The parameters to be measured at entry and follow up should be
specified. Numerous studies of diseases of the respiratory system and
its risk, factors have been conducted. Instruments which are valid and

reliable are available. Three standard questionnaires have been
developed for the study of respiratory epidemiology

by the British

Medical Research Council, U S National Heart and Lung Institute (NHLI)
and the American Thoracic Society. A suitable one could be combined

with general health questionnaires. Standardised methods for
spirometric lung function tests are also available. Random and
systematic sources of error in measurment must be minimised.

8)

Other factors to be considered are

- the training of investigators,
- pilot testing and
- planning for the follow uj

of

87

a percentage of non-responders.

)

9)

Analysis
In a longitudinal study an unbiased estimate of the relation

between exposure and outcome is obtained. The relative risk (incidence

rate in the exposed/incidence rate in the unexposed) and absolute risk
(incidence rate in the exposed - incidence rate in the unexposed) can

be calculated. It would be more useful to work out person years of
and calculate the force of mortality / morbidity or the

instantaneous mortality / morbidity rate. The risk of developing a
particular outcome (death / disease) can be estimated for a variety of

f

initial characteristics eg distance from factory, action taken at the

time of disaster, main presenting symptom in the acute phase etc.
10) Difficulties
Drop outs causing attrition of the sample are to be expected.
Every effort to get a good follow up should be made.
Substantial loss to follow up may raise doubts the validity of the
results as bias would be introduced if the loss is correlated with

I

both exposure and disease.
It is important also to maintain consistent criteria and
techniques for measurement throughout the study period. This is in

view of the fact that turnover in staff and availability of newer
instrumentation and techniques will occur over time.
A longitudinal study is also a major undertaking in terms of
resources - personnel, facilities, finances etc. The seriousness of

the situation however demand this effort, which would be best

conducted under the auspices of the State health authorities and the

ICMR.

12.9.3

Multjplf or syrj^X qrpsf-srction*! ftudjci
Difficulties inherent in the conduct of cohort studies have

led to the use of multiple cross sectional studies. This would be

88

I

J

I

carried out on random samples of the population at different points in

time. In Bhopal they would be able to show if there are changes in
prevalence from one survey to another. However, since the same

individuals would not be followed up, the naturqal history of the

disease will not be studied. Changes in population structure in the

intervening period could cause a change in the measure of outcome. The
sampling method, method of data collection and analysis and response

rate should be comparable at each examination. Comparison of mean
values or frequencies of variables such as age and sex could give an
idea of changes occuring ig the population structure. Sample sizes

will be larger than for cohort studies because the greater power of
tests of difference between paired observations in the same individual
cannot be exploited (Florey and Leeder, 1985). Independent non

governmental groups could probably undertake this study design.

12.9.4

Epidemiological Monitoring Unit

A detailed longitudinal study with intensive efforts to obtain

a good response rate and follow up need be done only for a small
sample of the exposed cohort. The setting up of an epidemiological

monitoring unit should be considered for the entire exposed

population. This would be based on routine records from hospitals and
health centres regarding admissions and deaths. It would necessitate
the building up of an efficient system of recording, reporting and

analysis. A special census of the exposed population could be

conducted and a method of identification of exposed individuals

evolved. The system would be able to pick up important changes in
morbidity or mortality on which appropriate action could be taken.

89

)

I

CHAPTER 13

CONCLUSION
The Bhopal disaster has been a human tragedy ot immense

dimensions. The suffering caused is incalculable. Important tasks

rema?

ahead

for the provision of the best possible care for the

victims and for the prevention of such events in the future.

There is a need, first, for the measurement, understanding and
documentation of the impact of the disaster on the health of those

exposed, so as to be able to provide rational care. It is necessary

also to document the seriousness of the effects so as to prevent an
easy erasure from human memory of the event. Epidemiologic skills
could help in this effort as described in this report.

At the present time it is known that similar small scale

"technological disasters" occur frequently. Larger scale disasters
could also occur. Hence, along with the deeper causes of these
disasters being tackled, there is a need to have a strategy to deal

with such events.
Outlines for this are as follows :

- it is necessary to have epidemiological data for an adequate

understanding of the effects on human health. This would
include data regarding the numbers and demographic structure of

the population at risk, the age / sex / area distribution of
the fatalities if they occur, and similar data regarding

morbi di ty.

- through col 1aboration between clinicians and epidemiologists,
i

it would be necessary to evolve simple, standard criteria for
assessment and documentation of morbidity.

- similarly, a method to assess exposure needs to be evolved.

90

<>

- collaboration and communication between administrators, service
providers and researchers is important.

- close contact and communication with the affected people is the

/

most important factor. In the absence of this, one could easily
slip into esoteric, theoretical exercises, which are

meaningless to the problem at hand.

These efforts have to be seen in the context of the broader
issues raised by such events. In Bhopal, these would include :

- the economic relationship between multinationals and countries

of the Third World which determine factors like technologies
and safety systems used,

- the exploitative relationship with the work force and
the local community to maintain high profit margins.

- the siting and safety systems of hazardous chemical plants.

- legislation regarding and implementation of safety controls.

- the workers and communities right to information,

I

- the role of pesticides and
- the acceptable limits to the chemicalization of our world.
The true causes of the disaster and the scope for preventing such

events in the future, lie in the matrix of these issues.

91

I

REFERENCES
(
I
(

Adler J and Eldlar R, (1985)
Disaster management - the state of the art, 1984.
Views of an international conference.
Disasters, 9(3),187-189.
Agarwal A and Narain S,1986 (Ed’s)
The State of India’s Environment 1984-85. The second citizen’s report.
"The Bhopal disaster" in the Health chapter.
pp 205-233, Centre for Science and Environment, N Delhi.

American Thoracic Society, (1979)
ATS statement - Snowbird workshop on standardization of spirometry.
Am.Rev.Resp.Dis.,119,831-838.

American Thoracic Society (1985)
(Andrew C et al)
Guidelines as to what constitutes an adverse respiratory health
effect, with special reference to epidemiologic studies of air
pollution.
Am.Rev.Resp. Dis.,131,666-668.
Andersson N, KerrMuir M, Mehra V,(1984)
Bhopal eye.
Lancet,Dec 22-29,1984, 2:1481
Andersson N et al,(1985)
Bhopal disasterzeye -follow up and analytical chemistry.
Lancet,Mar 30,1985,761-762
Andersson N, KerrMuir M, Ajwani M K< Mahashabde S, Salmon A,
Vaidyanathan K, (1986)
Persistent eye watering among Bhopal survivors.
Lancet, Nov 15, 1986,1152 (letter)
Anon,(1985)
Recent events point to need for health preparedness for technological
disasters.
Disaster preparedness in the America’s, Issue No.22,Jan’85

Anon, (abstract),(1985)
Bhopal methyl isocyanate incident - investigation team report.
Union Carbide Corporation, Danbury,Connecticut, 25 pp.
APPEN (Asia Pacific Peoples Environment Network),1985
APPEN Action alert :Bhopal Anniversary, 23,Nov 1985,
Sahabat Alam, Malaysia.
Axford A T et al,(1976)
Accidental exposure to isocyanate fumes in a group of firemen.

Br.J.Ind.Med. 33,65-71,1976.

Aydelotte C L,(1986)
Stepped up inspections, laws mark Bhopals legacy in the U.S.
Occup.Hlth.and Safety, Jan 1986,18-20.

Banerji D and Lakhan Singh,(1985)
Bhopal gas tragedy- an epidemiological and sociological study.
J NU News, April 1985,N Delhi.
92

I

Bang A, (1985)
Learning from the relief work- a report from Bhopal.
Medico Friends Circle Bulletin, 109,Jan 1985.
Bang R,(1985)
Effects of the Bhopal disaster on women’s health - an epidemic of
gynaecological diseases. Part 1 and 2.
Mi meograph.

Baram S M,(1986)
Chemical industry accidents, liability ana community right to know.
Am.J.Pub.Hl th. Vol 76(5),568-572.
Baram S M, 1984
Charting the future course for corporate management of health risks.
Am.J.Pub.Hlth. 74(10),1163-1166.

Barker D J P and Rose G, (1984)
Epidemiology in medical practice,
pp 158,Third edition, Churchill Livingston.

Berardinelli P S and Moyer S E,(1987)
Methyl isocyanate liquid and vapour permeation through selected
respirator diaphragmsand chemical protective clothing.
Am. Ind.Hyg.Assoc.J. 48(4),324-329,Apri1 1987
Bhargava M P,1985
The Bhopal tragedy - a middle word.
Mi meograph

I

Bhopal disaster Monitoring Group,(1985)
Bhopal never again.
B.D.M.G. vopan, No. 1, Jan 15,1985.
Bhopal Group for information and Action,(1987)
BHOPAL :Issue’s 1,2,3,10,11,Jun’86,Jul’86,Aug’86,April ’ 87, May’ 87
B G I A, Bhopal, India.
Bhopal Working Group,1987
The public health implications of the Bhopal disaster.
Report to the Program Development Board, Amer.Pub. Hlth.Assoc.
(APHA technical report)
Am.J.Pub.Hl th. 77(2), 230-236, Feb 1987

Bronstein M J,(1984)
The effect of public controversy on occupational health problems;
Byssinosi s.
Am.J.Pub.Hl th. 74(10/,1133 -1137.

I
Canas R and Kilbourne E M,1987
Oil ingestion and the toxic oil syndrome iresults of a survey of
residents of the Orcasur neighbourhood in Madrid,Spain.
Int.J.Epid. 16,3-6.
Castleman I B, (1979)
The export of hazardous factories to developing nations.
Int.J.Hlth.Services. 9(4),1979.

93

J
I

Ibid.,(1983)
The double standard in industrial hazards.
Int.J.Hl th.Services. 13(1),1983
Christoffel T and Swartzman D,(1986)
I

Nuzlear Weapons Testing Fallout : Provinc causation for exposure
injury.
Am.J.Pub.Hl th. 76(3), 290-292.

Diamond S, (1986)
India Carbide trade charges.
New York Times, 20 June 1985, D5.
Diller W, (1986)
Isocyanate effects.
J.Occup.Med. 28(4),265-266, letter.

Documentation for Action Groups in Asia (DAGA)
Asia’s struggle to affirm wholeness of life. Report of a consultation
on TNC’S in Asia :The aftermath of Bhopal.
DAGA,Hong Kong,Nov 1985
Dodd D E et al,(1986)
Acute inhalation studies with methyl

isocyanate vapour - 1.

1) Methodology and LC 50 determinations in guinea pigs, rats and mice.
Fund.Appl.Toxicol. 6(4),747-755.
Ibid. (1986)
2) Respiratory tract changes in guinea pigs, rats and mice
Fund.Appl.Toxicol. 6(4),756-771.

Editorial (1966)
Hazards of disocyanates
Lancet,Jan 1,1966, 32-33.
Editorial (1984).
Calamity at Bhopal.
Lancet, Dec 15

Enright P,(1986)
Spirometer use helps detect lung disease in time for treatment.
Occup.Hlth and Safety. 74-77,April 1986.
Environmental Health Criteria, 27 (1983).
Guidelines on studies in Environmental Epidemiology.
IPCS International Programme on Chemical Safety.
UNEP, Geneva,1983.
Ferguson J S et al (1986)
Sensory and pulmonary irritation with exposure to methyl isocyanate
Toxicol.and Appl.Pharmacol. 82(2), 329-335.

Fienberg S E et al, (1985)
Longitudinal analysis of categorical epidemiological data :
A study of Three Mile Island.
Envir.Hlth.Persp. 63,241-148.
Florey C du V and Leeder S R (1982)
Methods for cohort studies of chronic airflow limitation,
pp 134, WHO Regional Publications, European Series No. 12.

94

Gassert T, (1986)
Toxicology of methyl isocyanate - a lesson from Bhopal for
prevention.
Dissertation for MSc in Occupational Hygiene, LSHTM,
University of London, Sept. 1986.

!

isaster

Gassert T, Mackenzie C, KerrMuir M, Andersson N, Salmon AG,(1986)
Long term pathology of lung, eye, and other organs following acute
exposure of rats to methyl isocyanate.
Lancet, Dec 13, 1986, 1403 (letter)
Goldsmith J R et al, (1986).
Pulmonary function screening and monitoring in occupational health.
J.Occup.Med. 28(8), 656-663.
Grandjean P and Tarkowski S (eds), (1983)
Toxic oil syndrome : Mass food poisoning in Spain.
Report on a WHO meeting, pp82,
WHO Regional office for Europe, Copenhagen.

Haastrup P, (1984)
Indoor fatal effects of outdoor toxic gas clouds.
J.Occup.Accidents. 5,279-290.
Harding J J and Rixon C K, (1985)
Lens opacities induced in rat lenses by methyl isocyanate.
Lancet, 30 Mar’85,762.

I

Higgins M W and Keller J B, (1^70)
Predictors of mortality in the adult population of Tecumsehrespiratory symptoms, chronic respiratory disease and ventilatory lung
function.
Arch. Environ. uealth Vol 21,418-424.

Indian Council of Medical Research, (1985)
Health effects of exposure to toxic gas at Bhopal.
An update on ICMR sponsored researches. 10,Mar,1985.

Ibid (1985)
Minutes of the third meeting of the working group on thiosulfate
therapy of the MIC exposed population. 4,April, 1985.
Ibid (1985)
Minutes of the review meeting of ICMR sponsored researches on health
effects of exposure to toxic gas at Bhopa1
May 3-4,1985, Bhopal.
Joffe M,(1982)
Biases in research on reproduction and women's work
Int.J.Epid. 14(1),118-123.

Kapila M,(1986)
An epidemiological report on the health consequences of the Lake Nyos
disaster.(Volcano)
Mimeograph, Yaounde and Cambridge, Min. of Pub.Hlth. and ODA,UK.
Kimmel C A and Kimmel G L (1986)
Interagency regulatory liaison group workshop on reproductive toxicity
risk assessment.
Envir.Hlth.Persp.66, 193-221.
95

I

Kimmerle G and Eben, (1964) Abstract
Toxicity of methyl isocyanate and its quantitative determination in
the air.
Archiv.Fue-.Toxicologie. 20, 235-241,(German)

I

Kreiss K, (1986)
Approaches to assessing pulmonary dysfunction and susceptiblity in
workers.
J.Occup.Med. 28(8),664-669.
Luster M I et al, (1986)
Immunotoxicity studies o' mice exposed to methyl isocyanate.
T ox i col.App1.Pharmacol.86,140-144.

MacMahon B and Pugh T F,(1970)
Epidemiology - principles and methods,
pp 376, Little Brown and Company.
Mac Sheoin T, (1986)
Bhopal :a bibliography.
Int.J.Hlth.Services. 16(3), 441-468.

f

I

I

Mcdonald A D et al, (1986)
Spontaneous abortion and occupation.
J.Occup.Med. 28(12), 1232-1238.
Me Farlane A C et al, (1985)
Australian research into the psychological aspects of disasters.
Disasters, 9(1),32-34.
Medico Friend Circle, (1985)
Medical research in Bhopal : the realities and recommendations.

Ibid. (1985)
Medical research in Bhopal - are we forgetting the people?
MFC Bulletin, 109,April 1985.

Merchant J A,(1986)
Preparing for disaster. (editorial)
Am.J.Pub.Hl th. 233-235, Mar786
Mileti S D et al, (1975)
Human systems in extreme environments :a sociological perspective.
Monograph 21, Institute of Behavioral Science, University of Colorado.
Morehouse W and Subramaniam M A, (1986)
The Bhopal Tragedy : What really happened and what it means for
American workers and communities at risk.
A report for the Citizens Commission on Bhopal.
Council on International and Public Affairs, USA

Moyer S E and Berardinelli P S, (1987)
Penetration of methyl isocyanate through organic vapour and acid gas
respirator cartridges.
Am.Ind.Hyg.Assoc.J. 48(4), 315-323.

96

I

I

Nagnk Rabat aur Punarvas Committee (Nagrik study) (1985)
Medical survey on Bhopal gas victims.
In association with the Voluntary Health Association of India, Delhi
and The Bhopal Reliet Trust, Bombay, Mar 1985.

Nemery B et al, (1985)
Effects of methyl isocyanate on the respiratory tract of rats.
Br.J.Ind.Med. 42,799-805.

NIOSH (abstract), (1977)
Organoisocyanates in "Information profiles on potential occupational
hazards.", NIOSH contract No. 210-77-0120,
265-275, Oct 1977.
Parnell H A, (1986)
Perspectives in the development of asbestos litigation and its future
implications.
Am.Ind.Hyg.Assoc.J. 47(11), 708-711.

Patel A and Patel A,(1985)
The Bhopal disaster aftermath san epidemiological and socio- medical
survey.
Medico Friend Circle,
PP? 1985
Peto R et al (1983)
The relevance in adults of air flow obstruction, but not of mucus
hypersecretion to mortality from chronic lung disease.- Results from
20 years of prospective observation.
Am.Rev.Resp.Dis., 128, 491-500.

Phoon W 0 et al (WHO task force), (1986)
Carbamate pesticides : a general introduction.
Environmental Health Criteria 64, pp 137.
International Programme un Chemical Safety (IPCS).
WHO, Geneva.
Rinsky et al, (1986)
A mortality evaluation of employees with potential exposure to MIC.
J.Occup.Med. (letter), 28(3), 243.
Rosling H, (1986)
Cassava, cyanide and epidemic spastic paraparesis- a study in
Mozambique on dietary cyanide exposure.
pp 52, Uppsala University, Sweden.

Rothman K J, (1986)
Modern Epidemiology,
pp 358, Little Brown and Company.
Rye A W, (1973)
Human responses to isocyanate exposure.

J.Occup.Med. 15(3), 306-307.

Salmon A G et al, (1985)
Acute toxicity of methyl isocyanate : a preliminary study of the dose
response for eye and other effects.
Br.J.Ind.Med. 42, 795-798.

97

u

I

i

i

Salmon A G ,(1985)
Does acute toxicity testing tell us anything useful?
Methyl isocyanate as a test case.
Br.J.Ind.Med. 42,577-578

Gangster 8 and Cohen H, (1985)
Medical aspects of environmental pollution - environmental incidents
in the Netherlands 1980- 84.
Clinical. Toxicol. 23 (4-6), 365-380.

Sango C R L and Skarping G, (1986)
Determination of isocyanate and aromatic amine emissions from
thermally degraded polyurethanes in foundries.
Am.Ind.Hyg.Ass.J. 47(10), 621-628.

Siesjo B K, (1985)
Oxygen deficiency and brain damage: localization, evolution in time
and mechanisms of damage.
Clin.Toxicol. 23(#_$), 267-280.

Simmonds J, (1987)
Europe calculates the health risk.
Nev* Scientist, 23 April, 1987.
Special Supplement (1986)
Health effects of volcanoes : an approach to evaluating the health
effects of an environmental hazard.
Am. J. Pub. Hl th., suppl ement, 76, pp 90.

Stewart A,

(1985)

Detection of late effects of ionizing radiation : why deaths of A-bomb
survivors are so misleading.

Int.J Epid. K

(1), 52-56.

Tamashiro H et a^, (1985)
Mortality and survival for Minamata disease.
Int.J.Epidem. 14(#0, 582-588.
Teague H J (1985)
Cyanide poisoning of Bhopal victims (?) - a hypothesis.
Unpublished article.

Traver G A, Cline MG, and Burrows B, (1979)
Predictors of mortality in chronic obstructive pulmonary diseasea 15 year follow up study
Am.Rev.Resp.Dis., 119, 895-902.

Varadarajar et al (1985)
Report on scientific studies on the factors related to the Bhopal
toxic gas leakage.
Council for Scientifi c and Industrial Research (CSIR), India.
81 pp, Dec 1985

I

I

Vincent W J. Ketcham N H, (1980)
A new fluorescence procedure for the determination of MIC in the
occupational environment.
Analy.Tech.in Occup.Hl th.Chemistry. ACS symposium series No. 120,
121-147.

98
4

1

Position: 4641 (1 views)