GLOBAL MINISTERIAL CONFERENCE ON RESERCH FOR HEALTH , BAMAKO, MAIL, 17TH -19TH NOV. 2008
Item
- Title
-
GLOBAL MINISTERIAL CONFERENCE ON RESERCH FOR HEALTH , BAMAKO, MAIL, 17TH -19TH NOV. 2008
- extracted text
-
1/
RF_RES_1_SUDHA
RES' I
AN ANAL YSIS OF MORPHOLOGICAL AND
MORPHOMETLUCAL PARAMETERS IN
ENDOMETRIAL CARCINOMA
Rohini Gajraj M.B.,B.S.
Intern,
Kanma Rameshkumar M.D., Ph .D.
Associate Professor,
As. Mohamed M.Sc.,D.P.Sc,*
Assistant Professor,
Department of Pathology and
Department of Community Medicine*
St. John’s Medical College Hospital,
Bangalore
No. of pages : 8
No. of figures: 1
No. of tables : 1
Correspondence to:
Dr. V. Kanina M.D., Ph .D.
Assoc. Professor, Dept, of Pathology
St Johns Medical College
Sarjapur Road, Koramangala,
Bangalore - 560 034
ABSTRACT
Histopathologic evaluation plays a major role in the prognosis and treatment of
endometrial carcinoma.
This when supplemented with morphometry could provide
important information the advantages being objectivity and reproducibility
Methods:
47 cases of Stage 1 endometrial carcinoma were reviewed over a period of
twelve years to analyse tlie morphological and morphometric pai ameters and correlate diem
with prognosis. Of these, 17 patients were premenopausal and 30 were postmenopausal.
Only seven patients were below 40 years of age.
Results : Based on morphological analysis, 12 patients had well differentiated tumours,
22 had moderately differentiated tumours while 12 belonged to the poorly differentiated
group. Among the nuclear features, nuclear area and perimeter showed significant difference
between the different histological grades, the former being more significant than die latter.
Conclusions: Hence the application of nuclear morphometry to diagnosis and tumour
grading could lead to better evaluation of die tumour and aid in predicting the prognosis. In
women below 40 years, diere is a danger of well differentiated carcinomas being diagnosed
as hyperplasia as carcinoma is rare in this age group. The application of nuclear
morphometry, which is both objective and reproducible, could aid in proper diagnosis of such
tumours and thus lead to earlier detection and control of endometrial carcinoma.
In troduction:
Endometrial carcinoma is the most common gynaecological malignancy next to
cervical cancer in developing countries. Histopathological evaluation plays a major
role in subdividing endometrial carcinomas into treatment groups. WHO has
recommended endometrial carcinomas to be graded into well differentiated,
moderately differentiated and poorly differentiated based mainly on the architectural
pattern of glands and secondarily on cytological and nuclear appearances1. Though
architectural and nuclear grading generally correspond, sometimes they are at
variance in which case the nuclear grade is the more reliable indicator of tumour
prognosis. In well differentiated examples of adenocarcinoma, both architectural and
nuclear differentiation may be so good that the tumour appears to satisfy none of the
criteria for such a diagnosis and is difficult to distinguish as a carcinoma. Hence the
application of nuclear morphometry for diagnosis and tumour grading overcomes tlie
limitation of observer error in subjective assessment of the histological grade which
may not always be reproducible.
Among the prognostic determinants other than grading are age, race, symptoms,
stage, depth of myometrial invasion, DNA ploidy and receptor status. Various
studies have linked all these parameters with survival, ’ ’ but little data is available
relating nuclear morphometry and prognosis in an Indian set up.
The present study was designed to evaluate the diagnostic and prognostic value
of clinical, histopathological and morphometric features. These were assessed by
light microscopy and eyepiece micrometry in Stage 1 endometrial adenocarcinoma.
MATERIAL AND METHODS
Material :
88 cases of endometrial carcinoma over a period of 12 years (from 1984 -
1997) were seen in the department of Pathology. Among them 47 cases of Stage 1
carcinoma where the tumour was confined to the corpus including isthmus were
considered for the study which comprised 53.4% of the total cases. Staging was done
according to the jointly agreed scheme of FIGO, the International Union against
cancer (UICC) and the American Joint Committee for cancer staging and end result
reporting.5 The clinical details including treatment details were obtained from the
archives of the department and the medical records department and were tabulated
depending on age group and menopausal status.
Methods :
Haematoxylin and eosin stained sections were used to study the basic
morphology of the neoplasm and for grading. In the histopathological evaluation
three different steps were used. At low magnification (x250) tissue architecture
features were assessed. At higher magnification (x400) nuclear size and shape were
evaluated for grading. Grading was performed in compliance with the system of
WHO and FIGO. The nuclear features were measured using eyepiece micrometry
which can be fitted to a 1 ight microscope.
The long axis and short axis of the nucleus were measured in 50 nuclei after
selecting areas containing largest and atypical nuclei in 10 adjacent fields in high
power objective. The nuclei selected were clearly visible, not artificially deformed
and had intact nuclear membrane and chromatin. The nuclear area and perimeter
were calculated using the formula nab and 2n ^/((a2 +b2)/2) respectively using a
computer program. For each specimen, the arithmetic mean and standard deviation of
these parameters were calculated. Thus for nuclear morphometry, nuclear area,
perimeter, longest axis, shortest axis, nuclear axes ratio (longest -s- shortest axis)
were assessed in addition to shape.
To find the number of nuclei to be measured, these features were re-measured in
the same nuclei and the 95% limits based on inter-observer variation was assessed.
For the features mentioned, 25 to 40 nuclei were sufficient to have cumulative
average within the 95% limits. To be on the safe side, 50 nuclei were measured in
each case. Mitosis were observed, but
were not quantified and hence mitotic
activity index was not calculated.
Statistical analysis was done using analysis of variance metliod (ANOVA). Hie
results of nuclear morphometry were tabulated against histological grade to observe
whether any significant diflerence was present The treatment and survival data were
then correlated with histological features and nuclear morphometry.
RESULTS
The total number of patients diagnosed with endometrial carcinoma during the
twelve year period (January 1984 to January 1997) was 88. Among them, 47 were
diagnosed to have Stage 1 disease prior to treatment. The rest were in Stage 2 or had
only curettage performed for diagnosis and no staging or follow up was available.
The mean age of the patients was 53.70 + 11.46 years. Seven were below the age of
40 years with a mean age of 33.8 4- 4.6 years. Using the criteria of menstrual history,
17 patients were in the premenopausal group with a mean age of 42 4- 4.1 years.
(Refer figure 1)
A review of the presenting symptoms indicated that all post-menopausal patients
presented with post-menopausal bleeding. The remaining women presented with
irregular bleeding (23%), abdominal pain (8.5%) and other symptoms (4.25%). No
significant difference was observed between the two groups with regard to the
incidence of obesity, hypertension or diabetes mellitus. The seven patients who were
below forty years presented witli menorrhagia or irregular periods. One of tliem in
this group had associated diabetes.
Assessment of grading based on the architectural pattern resulted in 12 patients
with tumours in the well differentiated grade (Grade 1), 22 in the moderately
differentiated (Grade 2) and 13 in the poorly diflerentiated (Grade 3) group.
Nuclear grading performed independently generally showed good correlation, except
in two cases where the nuclear grading was lower than the architectural grading.
Calculation of nuclear area and perimeter showed a significant difference between
the three grades (Refer table 1), the difference being more significant with the
nuclear area than perimeter. The premenopausal group had a higher incidence of
well differentiated (6/17) and moderately differentiated (9/17) carcinoma than die
post-menopausal group (well differentiated - 7/30; moderately differentiated
14/30). None of the patients below the age of 40 years had Grade 3 carcinoma
(Refer figure 1).
Depth of myometrial invasion was measured independent of grade and was
classified as superficial or deep. 25 cases had myometrial invasion of which only
two could be classified as deep. Both these cases belonged to Hie poorly
differentiated group and had in addition vascular invasion. Necrosis and/or
lymphoplasmocytic response were observed only in well and moderately
differentiated adenocarcinoma (4.8%) Concomitant endometrial hyperplasia was
seen in four cases, of which two were seen in association with moderately
differentiated grade. Other associated conditions observed were leiomyoma, polyps
and adenomyosis.
Treatment consisted of total abdominal hysterectomy with bilateral salpingooophorectomy in all patients. Adjuvant radiation therapy was done when indicated.
Status of the patients was evaluated by repeated cytological examination or by
recurrence of symptoms. Of Hie 47 patients, twenty eight were alive and well and
these patients had tumours of histological grade 1 and 2. The five patients who died
of the disease had poorly differentiated or grade 3 tumours. Two patients with grade
2 tumours died due to unrelated causes and twelve patients were lost to follow up
after surgery7.
DISCUSSION
In an ideal situation, a tumour from a patient would be subjected to a wide range
of laboratory investigations including histopathological evaluation, a panel of
antibodies, morphometry and flow cytometry. But in a limited set up, a combination
of morphology and morphometry could provide important information relating to
prognosis, the advantages being objectivity and reproducibility.
Each histological subtype of endometrial carcinoma has distinct biological
behaviour. The two types that have favourable outcome are adenoacanthoma and
adenocarcinoma which fortunately comprise 83% of all endometrial carcinomas6.
Majority of patients present with Stage 1 disease. Better evaluation of prognostic
determinants could be done by restricting the study to patients with Stage 1 disease
who had a favourable subtype (adenocarcinoma) and near optimal treatment.
Age and menopausal status at the time of diagnosis were important risk
determinants. As all the post-menopausal
patients presented with bleeding,
presenting symptom served as an important preliminary' prognostic indicator in this
series. The clinical spectrum of the disease and the distribution of patients according
to age in the present study is similar’ to that reported by Western studies and some
Indian studies.7’8,9'10
As nuclear grading is included in the revised FIGO recommendation, a precise
definition of nuclear atypia will be of great relevance. Hie nuclei of malignant cells
are char acteristically larger, less regular in outline, more densely staining and have
larger and more numerous nucleoli, Ilian those in benign or normal cells. These
characteristics are quantifiable and have shown to be prognostically relevant in some
morphometric studies.11,12 Connelly et al proposed a nuclear grading system and
found that it predicted the subsequent clinical course better than the histological
grade12. They found more neoplasms with histological grades at higher levels than
the nuclear grade. However, Geissenger et al observed that nuclear grade was
greater than histological grade4. In the present study, observations similar to that of
Connelly et al were made; however, this study had in addition morphometry which
overcomes the limitations of subjective assessment. By selecting the area of section
containing the largest and most atypical nuclei, nuclear area was found to show
significant correlation with tumour grade. Further, parameters like the nuclear
perimeter and nuclear shape will help additionally, rather than isolated assessment
of nuclear area for correct morphometric grading.
The mitotic index and mitotic count are well established procedures in tumour
grading.13 Though they seem simple, there are major limitations. The recognition of
mitotic figures is subject to observer error and the histologically visible event of
mitosis occupies only a short period in the cell cycle, so a large number of nuclei
have to be counted to obtain a statistically reliable estimation of the proportion of
proliferating cells. Any delay in fixation results in underestimation of the number of
mitosis.14 In the present study which was a retrospective analysis, the fixation details
were not known. Hence the presence of mitotic figures was observed but not
quantified.
Among the other morphological features, depth of myometrial invasion
correlated well with the prognosis, probably because the deep myometrial invasion
was
observed
only
in
poorly
differentiated
carcinomas.
Necrosis
and
lymphoplasmocytic response, which were observed in well and moderately
differentiated adenocarcinoma, also showed correlation with better prognosis.
The presence of endometrial hypeiplasia may demonstrate a more favourable
prognosis. In a study which compared patients with carcinoma with and without
hypeiplasia, those with hypeiplasia were better differentiated, lacked deep
myometrial invasion, cervical involvement and lymphovascular space invasion.15 In
the present series, concomitant hyperplasia was observed in the non-neoplastic
endometrium in 5 of the patients. Though the number is too small for statistical
evaluation, it is noteworthy that tumour in these patients was of Grade 1 or well
differentiated.
There is a danger of well differentiated carcinoma being incorrectly diagnosed
as hyperplasia as carcinomas are rare in those below 40 years of age. In such cases,
nuclear morphometry which can be easily done by eyepiece micrometry can be
applied to aid in diagnosis and further evaluation.
In summary, the diagnosis and grading of endometrial carcinoma can be based
on few simple procedures in addition to light microscopy, which when carefully
performed will pennit adequate evaluation of the tumour. This will have impact on
the treatment which will lead ultimately to better cancer control.
REFERENCES
1.Poulsen IIE, Taylor CW, Sobin LIT
Histological typing of female genital tract tumours - International Histological Classification of
Tumours , Geneva, World Health Organisation, 1975; 13: 64-65
2. Anderson B.
Diagnosis and staging of Endometrial Carcinoma
Clin. Obstet. Gynaecol;1982; 25:75-80
3. Hanc W.H.M., Putten V, Baak J.P.A. ct al
Prognostic value of quantitative pathologic features and DNA content in individual patients
with stage 1 endometrial adenocarcinoma, Cancer ,1989;63:1378-1387
4. Gessinger K.R., Marshall R.B., Kute T.E. and Homesety H.D.
Correlation of female sex steroid hormone receptors with histologic and ultrastructural
differentiation in adenocarcinoma of the endometrium.
Cancer,1986;58:1506-1517
5. American Joint committee for cancer staging and end result reporting.
Staging system for cancer at Gynaccologic sites.
Chicago: American Joint Committee,1979:89-100
6. Christopherson, Cannily P.J., Albcrhasky R.C.
Carcinoma of the Endometrium- An analysis of prognosticators in patients with favourable
subtypes and Stage 1 disease.
Cancer,1983;51:1705 - 1709
7. Yasumizu T, Ogawa K, Kato J.
Comparison of the dinicopathological characteristics of premenopausal and post-menopausal
endometrial carcinomas.
Jpn. J. Clin. Oncol. 1996; June 26 (3): 152-6
8. Alan B.P. and Janies W. Reagan
Incidence and prognosis of endometrial carcinoma by histological grade and extent.
Obstet. Gynaccol;1970; 35:437 -443
9. Krissi H, Chetrit A, Menczer J
Presenting symptoms of patients with endometrial carcinoma. Effect on prognosis.
Eur. J. Gynaecol. Oncol. 1996;17 , 1:25-28
10. Sandhu A S, Patel F„ Singh D P et al
Results of treatment in endometrial carcinoma - Ten years experience
Indian J. Cancer, 1977; June 34 (2): 77 - 83
ll.Ncdcrgaard L, Jacobsen M, Anderson J E
Interobserver agreement for tumour type, grade of differentation and stage in endometrial
carcinoma
APMIS 1995; July-Aug 103 (7): 511-518
12. Connelly PJ, Alberhasky RCt Christopherson
Carcinoma of endometrium - III
Analysis of 865 cases of adenocarcinoma and adenoacanthoma
Obstet. Gynaecol., 1982;59:569-574
13. Don huijsen K
Mitosis counts: reproducibility and significance in grading of malignancy
Human Pathol.,1986; 17:122-125
Id.Cross S.S., Start R D , Smith J IIF
Does delay in fixation affect the number of mitotic figures in processed tissue
J. Clin. Pathol.,1990;43: 597 - 599
15.Kaku T, Tsukarnoto N, Hachisuga T et al
Endometrial carcinoma associated with hyperplasia
Gynaecol. OncoL, 1996; June 60 (1): 22 - 25
CORRELATION BETWEEN HISTOLOGICAL GRADE AND
NUCLEAR FEATURES
IN
ENDOMETRIAL CARCINOMA
NUCLEAR
AREA (jim2)
MEAN ± S.D
NUCLEAR
PERIMETER (p.xn)
MEAN ± S.D
WELL
DIFFERENTIATED
51.2 ±2.41
36.11 ± 0.97
MODERATELY
DIFFERENTIATED
42.54 ± 3.69
33.47 ± 1.83
POORLY
DIFFERENTIATED
37.12 ± 3.91
30.72 ± 1.92
ANOVA
F VALUE = 28.79
p VALUE < 0.001
F VALUE =5.53
p VALUE < 0.01
GRADE
Sheet 1
DISTRIBUTION OF PATIENTS IN RELATION TO AGE, MENOPAUSAL STATUS AND
HISTOLOGICAL GRADE IN ENDOMETRIAL CARCINOMA
8
54
I
NO. OF CASES 3
21-
0
il
21-30
I
31-40
II
1
41-50
I I
□ Seriesl
O S8ri8s2
a Serlas3
I
i
i
1
s I H 1
51-60
AGE IN YEARS
61-70
71-80
i
STEPS IN CARRYING OUT ANY RESEARCH
t
1.
Describe and define study problem.
'
Background, arguments for relevance, aspects related to study subject.
2.
-
Formulate / define study question.
-
Start with broad definition, try to specify / narrow down the question(s) to be answered in
-
a series of subsequent steps.
Make sure that all the terms / concepts occurring in the study question are stated in
concrete / measurable term ('operationalization'). You should be able to explain all the
terms.
3.
Study the relevant literature. Make a review / overview / meta-analysis / research synthesis.
What is already known about the problem to be studied, from previous investigations (fre
quency of occurrence, related phenomena, potential confounders, etc.)? What are the white
spots'? What were the methodological weaknesses of previous research efforts?
4.
Reformulate the study question (based on the conclusions from the literature study).
Distinguish several subquestions. Can the be answered all at once, using the same study.^
Formulate research hypotheses (if possible; hypothesis testing research vs. exploative
research).
5.
Consider the nature of the study quesUon. Does it require a descriptive or an analytic (cause
effect) type of approach/research.
This distinction refers to the study objectives, so on the intended applications of the study
results. Does one just want to describe the frequency of occurrence of one or more relevant
phenomena (e.g., a particular disease, symptoms, immune status, a risk factor), including
differences in their distribution between subpopulations based on time-, place-, and person
characteristics. Or does one intend to end up with conclusions in terms of (causal) relations
hips between phenomena/variables measured.
6.
If the study problem is health/disease-related, consider which phase of the natural cause of
disease is involved (preclinical / clinical; asymptomatic / symptomatic;'etiology / diagnosis /
prognosis; prevention (primary/secondary / treatment (tertiary), - etc.). Is there yes/no an
intervention among the phenomena to be studied?
7.
Design options.
7.1.
Provisional choice of study design:
Level of aggregation of measurements / observations:
: ecological
study,. correlation study (geographical correlation study,
• Group level
w
time series analysis)
• Individual level : other types of study
Level of comparison:
• No
e.g., case study, case series
• Yes
other types of individual studies
Timing of study:
cross-sectional study, survey study
• One measurement time-point
• Several measurement time-points : longitudinal study:
• prospective: follow-up study, concurrent cohort
study (no random allocation), experimental
intervention study (random allocation)
• retrospective study: case-control study
For intervention studies in the social sciences the classification scheme introduced by
'I
•r
-ZbT--- ft
7.2.
Cook & Campbell is often used: three main groups of disigns are distinguished: preexperimental, quasi-experimental en true experimental designs, each containing several
designs (notation system: O = observation, X = Intervention, E = experimental group, C =
control group; types of bias: history, maturation, selection, attrition, statistical regression,
attrition, instrumentation).
General considerations in choosing a design:
• Does the underlying study question refer to a descriptive problem (occurrence, distribu
tion of relevant phenomena in the population; differences between subpopulations;
associations between phenomena / characteristics) or to an analytic problem (1. deter
minants / risk factors / etiologic factors. 2. effects of planned interventions).
When dealing with descriptive problems it is very important to work with a representati
ve sample (external validity I), for analytic problems one may choose a non-representative way of sampling that guarantees a high level of internal validity.
• To attain a high level of internal validity:
Prevent / control for selection bias, Information bias, confounding:
- Maximal contrast in status of central exposure / intervention factor
- Minimal contrast in status of other disease / outcome determinants:
restriction, randomisation, prestratification, matching, stratified analysis, etc.
- No measurement errors (systematic, random): measurement protocol, blinding, etc.
• Pursuit of a high degree of efficiency: statistical (precision), economical (money, per
sonnel, time needed)
Development of study design:
1. Choice of study population (who should be assessed, treated, etc. ?):
- Composition: inclusion- and exclusion criteria (eligibility criteria, admissibility criteria)
- Size of study population
- Selection procedure:
• Source population?
• Sampling frame available?
• Sampling procedure
Methods:
probability samples (known and often equal chance for each member of
the population to be sampled; e.g., simple random sampling, stratified
sampling, cluster sampling) versus non-probability samples (e.g. snow
ball sampling, haphzard sampling, etc.)
• Method of recruitment
2. Choice of measures (what should be measured?)
- Phenomena to be measured (conceptual level)
- Measurement tools: available or to be developed?
Attention for
• Validity
(face-, content-, construct-, criterion- (concurrent,
predictive): Se, Sp, LR, PV+, PV-, etc.)
• Reproducibility / Precision
(inter- and intra-rater reliability, test-retest reliability,
internal homogeneity / consistency: Kappa, r, etc.)
• Sensitivity to change / responsiveness (in case of effect measure)
• Practical applicability
- Methods of data collection needed (records, registers, interviews, questionnaires,
observation, physical examination, laboratory analysis)
- Other aspects of measurement procedures (e.g.: ’blinding’ of observers / raters)
3. Timing of measurements (when, how often, what sequence)
4. Way of allocation of study subjects to exposure factors / intervention factors:
- Random allocation vs. self-selection
5. Methods of data handling:
- Storage
- Statistical analysis: protocol
i
I
3
•
•
•
•
Univariate analysis: frequency distribution of each variable (mean + s.d., median +
inter-percentile values, counting of numbers in categories)
Inspecting and handling missing values, outliers, illegal values: data control
Bivariate analysis: cross tables, stratified analysis
Multivariate analysis: e.g. multiple regression, logistic regression, discriminant
analysis.
7.3.
Write down study design in protocol, including time schedule.
Discuss with collegues, funding agents, research committees, etc.
8.
Preparation / organization of the study.
- Appointments with ’the field'
- Development of questionnaires, forms, etc.
- Development of measurement tools
- Attention for legal requirements
- Fainancial aspects
- Housing + other facilities
- Etc.
9.
Conduct of the study.
-
10.
11.
Collecting data, control of data, additional information, enter data in computer
Etc.
Data analyis.
Reporting:
i
- Reports
-
Journal articals:
Structure:
Introduction: motivation, study question
Background information
Study design (Material and methods): population, measurements, analysis
Discussion
Conclusions + Recommendations
STEPS IN AN EVALUATIVE INTERVENTION STUDY
1.
6.
Implement intervention programme:
- Standardized protocol <—> Individualized intervention
- Control group: Placebo intervention
7.
Control + measure compliance + attrition:
- Contamination of subgroups: result of randomization procedure will be overruled
- Stimulate + measure (questionnaires, biological markers)
- Placebo-intervention will improve the level of compliance
- Attrition: selective or non-selective.
8.
Conduct follow-up measurements (how often? when?)
- ’Blinded’ measurement
9.
Be alert for (unwanted, adverse) side-effects (observations, questionnaires, register)
10.
Analysis.
Define the intended intervention effect:
Nature of outcome measure:
- Relevance
- Measurability
- Conceptual (e.g.: ’functional disability’) —> Operational, concrete (e.g.: a specific
disability questionnaire or rating scale (e.g.: ’Roland Disability Scale’)
- Measurement tool:
Existing one <—> newly developed one
Clinimetric qualities:
• validity (face-, content-, construct-, criterion- (concurrent, pre
dictive): Se, Sp, PV+, PV-, LR, etc.
• reproducibility / precision (inter- and intra-rater reproducibility,
test-retest reliability, internal consistency / homogeneity): Kap
pa, r, etc.
• sensitivity of change / responsiveness
• practical applicability (costs, time needed, hazards, etc).
• Size + direction of effect
• Time reference: How much time will it take before a possible effect will be visible, will fade
away? When and how often should outcome status be measured?
•
2.
Define intervention(s):
a. Experimental intervention:
- New programma <—> Existing programme
Or new elements added to existing programma
- Broadly defined, complex, multi-focal, national level <—> specific, local level
b. Reference / comparison intervention:
- Usual care
- Placebo intervention
- No intervention at all
3.
Define base population + study population
•
•
•
Composition
- Inclusion- / Exclusion criteria (Eligibility criteria, Admissibility criteria)
- Restriction (homogeneous population), matching, etc. in order to prevent bias (confoun
ding, selection bias) = to improve comparability / validity, to improve efficiency.
- Sampling procedure
Size(N)
- Depends of size of effect / amount of change in reference group, extra effect / amount
of change that one wants to detect in the experimental group, accepted chances of
type I- and type ll-errors
- Anticipate of non-response (%), attrition during follow up.
Procedure of acquisition (source (e.g.: advertisement, hospital, screening programme);
informed consent procedure (—> non-response)
4.
Allocate eligible and available candidates to intervention alternatives:
• Self-selection
• Randomization (+ pre-stratification)
5.
Measure base-line values (outcome variables, potential confounders / effect modifiers)
■y
Use of evaluation results:
effective presentation
k
/
\
Martine Collumbien
\
A
Objectives
o the moral responsibility to use data
o presenting and using process
indicators to improve intervention
o feeding back to the community
o presenting evaluation results
i
Moral responsibility
\
o if you cannot think of how a piece of
information might inform you, do not
collect it
o what decisions will be made on the basis
of your monitoring
o think ahead of how you will present your
findings
• use dummy tables (graphs) based on your BDI
and your indicators
• how will you feedback to the community?
I
h An example from
the EC/UNFPA
/
Reproductive
Health Initiative
for Asia:
/ \
Routine monitoring
■ft
& process
evaluation
Vietnam
/i
jA.
I
1
J
Sit,
■
t*'
2
Some challenges in collecting and
using data
I
\
)
. May be common sense to collect data by
sex, but many NGOs did not do it when we
started.
. Even when they did, data is often
aggregated, and few analyse it by sex.
. Similar with age
. Data often not integrated into project
management, or fed back to project staff, so
little time for reflection and adjustment.
Exercise in using data
/ \ Imagine that you are a project manager of a reproductive
health project offering outreach services through peer
education in 7 areas. You work mainly with youth (11-25
years), and have a mandate to reach both sexes.
You have just been given the complete set of routine data
from your 7 project sites on counselling visits for ARM,
family planning and RTI/STD/HIV to review with your
staff.
What are your main observations and what questions does
the data raise that might improve your programme?
3
Reporting period
\ Client visits to
Jan-Jun
\
•00
/ specific services
/ //
i
Jul-Dec
'00
Jan-Jun
*01
Jul-Dec
'01
Family Planning
10,490
18,273
20,604
25,908
RTI/STD/HIV/AID
S
2,135
8,149
4,879
6,234
n/a
n/a
12,043
8,808
12,625
26,422
50,108
63,701
Youth centres/
libraries
Total
70,000
Number of client visits to selected
services/activities
60,000
50,000
40,000
30,000
20,000
10,000
0
Jan-Jun '00
Jul-Dec '00
Jan-Jun '01
Jul-Dec '01
4
Damaam4 Jir+nikn+iflM i
100%
80%
rJ
lllpBWO
//x/
/ // /
i.
■ < 15 years 15-19 years 20-24 years □> 25 years ■ age not specified
----------------------------------------------------------------------------------
-
■
—.......................................................... ....................
~
"
___________________________________________________________________________
Sex distribution of clients 15-19 years attending
group counselling (2000-2001)
||g|
32%
'I
68%
(n=200,296)
H total male ■ total female
5
Sex distribution of 15-19 year olds
attending group counselling
100%
80%
60%
40%
20%
o%
Jan-Jun '00 (n-28602)
Jul-Dec'OO (n=90473)
■■
Jan-Jun '01(n=25638)
Jul-Dec‘01(n=55464)
s total male ■ totaljemale
\
Observations from routine data on
counselling
/ \
. More females than males
. More males seen for RTIs/HIV than for
family planning
. Some evidence that contact with males
declines over time
. Are we reaching our intended target
group? What is the age profile?
6
An example from CINI - LCA Cell
\
7
The Cell works to promote
development of a holistic life cycle
based approach addressing critical
life-stages through
• evidence-based programming
• technical support, quality assurance and
capacity building of agencies in
implementation of LCA programmes
• documentation, information
dissemination and advocacy
MIS Information Flow
Information
Sheet
Cohort Register
(health worker
wise)
Mother &
Child
Protection Card
_____
\
Village
Information
Supervisor’s Visit
]____
Household
Information
..
Vital Events
■•■J Database 4
One time
Monthly
--------- ►
Annually
Referral Slip
I
Reports:
I
1.
4-
I2
3.
Follow up list
Monthly / Annual Report
Supervisor’s Report
4.
Ranking of villages (Quarterly)
5.
Sponsor Report
7
Case Management Activities
October 2001 to September 2003
/
/
Low Birth
Weight
554 (20.9%)
Abortions
117
Live Born
2645 (97.5%)
Normal Birth
Weight
1966 (74.3%)
Still Alive
2563
Pregnancies
3819
, Hz
Deliveries
2714
Stillborn
69 (25/1000)
Died
0-7 days
Died
8-28 days
Died
29 days -1 yr
46
19
17
(17.4/1000)
(7.2/1000)
(6.4/1000)
TT Coverage (NFHS-2)
\
\
1
K
100.0 r
(/)
0)
(O
(V
o
**o
§2-2
80.0 r70.0
62
2 I
50.0 •
40.0
30.0
20.0 -10.0
0.0
* NFHS - 2
8
Taken 100 or more IFA tablets
tfi
o
tn
cu
o
o
1§8:8
?8:8
38:8
18:8
A<o>
.N^
1
s
&
>
z
1 cl
Feedback for community audit:
village performance in field unit one
Name of Village
1.1
T.T
In in N in Regjhj
I „
yr 1 yr 2 weeks
03
02
02 03 02 |03 02103 02 03 02 03 02 03 02 |03 02 03
37
57^B
i__ 44
goK |a|
|__ 23
35|
j
Fakirpara
Ramkrishnapur
Chandi
1
.£!*!.
7
‘IB
1Q
|a
lA IA
_22
68 B
_40
731A
95IB Mb
55g |A B
5941
_40
31
300
;B|B
A rl6JA
56
Enayatnagar
« IperfonnI
ii
nuu.
ii vviun
IFA
Attd. LBW Neonatl
j
arade
Deliv
al death
nee
m^MB gj|B jip
BJ B B |b A ]a; IB B
IIb’b AA B B 1
b Bp. 1
rF
B JeHI
0 ® |j b
B
B 01
„3'5B
I
3
3|HS 50.0
3^ 50.0
i^u6 B
3|il 50.0
7^1^ Zia
3
11
a a
3
50.0
i
50.0
itH
J X
J B |A IB
4
Ril
“
3|^|I 50.0
3[59| 63.3
9
Presenting evaluation results
o Consider group
• Donors
• Other NGOs (resource mapping!)
• Research community
o Be honest and comprehensive, show all
sides of the picture
O Compare your results with others
O Show what has not changed
• your targeted determinants
• determinants beyond your control
Full or selective data presentation?
\
k\
| Table 2. Results of surveys on sexual behaviour among high school students, 1998-2002
Year
Had sex in the past year Had multiple partners in the
|% of all)________ past year (% of all)
Always used condoms
(% ofthose with multiple partners)
1998
1999
2000
2001
2002
73___________ 9______________
50_________________
70___________ 11_____________
48_________________
68___________ 15_____________
41_________________
58
20_____________
j8______________
48
24
35
UNAIDS/WHO Guidelines for effective use
of data from HIV surveillance systems
10
Figure 1. Percentage of students at a high
\
school in which an abstinence-only
sexual health education curriculum
was introduced in late 1998 who report
being sexually active, 1998—2002
X
Abstinence programmes work!
Sexual activity among young people has fallen
by over a third
”
O
G>
Ol
80
70
60
50
40
30
20
10
0
73
70
68
58
1998
1999
2000
2001
2002
UNAIDS/WHO Guidelines for effective use
of data from HIV surveillance systems
Figure 2.
Percentage of high school students
who reported having sex in the past
year who had multiple sex partners
and percentage who always used
condoms
Abstinence programmes don’t work!
Sexual activity is rising and condom use is falling
60 1
50
50
50
48
41
38
40
8 30.
20
34
22
G>
Q.
12
15
^Multiple sex partners*
Always uses condoms**
10.
0 .
1998
1999
2000
2001
2002
UNAIDS/WHO Guidelines for effective use
of data from HIV surveillance systems
11
Figure 3.
k
/ \\
j
Proportion of high school students
having no sex In the past 12 months,
sex with one partner or consistent
condom use and unprotected sex with
multiple partners, 1998—2002
I
More young people are abstaining from sex,
but unprotected sex with multiple partners is rising too
69
1998
65
B
59
Kfl
46
KOI
1999
2000
V;
■
2001
32
M
2002
0%
10%
20%
40%
30%
El
60%
50%
70%
16
80%
90%
100%
■ No sex In the past 12 months
® Sex with one partner or consistent condom use
■ Unprotected sex with multipe partners
UNAIDS/WHO Guidelines for effective use
of data from HIV surveillance systems
Figure 13. Needle sharing among participants and
nonparticipants In the needle exchange
programme In Thansa
■s
100
0?
JS
90
80
J:
70
.£
<A
0>
1
S
£
82
84
84
82
79
60
50
40
30
■5
90
86
20
10
0
!8
Participants
Non-participants
1998
1999
2000
2001
2002
UNAIDS/WHO Guidelines for effective use
of data from HIV surveillance systems
12
Data for advocacy - packaging
/
o Define your goals and audience
o Find out what influences their
thinking, what their biases may and
use the data
o Use the right language to
communicate ...
Same information, different
language
o The HIV incidence in the 15- to 19-yearold cohort is high, and the prevalence
among 19-year-old women is 33%
o New HIV infections are common in the
late teens; a third of 19-year-old girls are
already infected with the virus
o Hundreds of teenagers get infected with
HIV every week. If there are 30 girls in
your daughter's class, about 10 of them
will have HIV by the time they
13
\
! \\
Data for advocacy - packaging
o ...
o get the length right
o choose the messenger
o strategic timing
• avoid election time or festivals
• around World AIDS day, Women's day,
Global conferences
14
- Media
RF_RES_1_SUDHA.pdf
Position: 3267 (2 views)