Desain Studi Ekologi dan

Studi Observasi

MK. EPIDEMIOLOGI GIZI

DEPT. GIZI MASYARAKAT, FEMA, IPB
April, 2014
 Definitions of Epidemiology

Lilienfeld A: in Foundations of Epidemiology

THE STUDY OF THE DISTRIBUTION OF A DISEASE OR

A PHYSIOLOGICAL CONDITION IN HUMAN
POPULATIONS AND OF THE FACTORS THAT
INFLUENCE THIS DISTRIBUTION

Last JM: A Dictionary of Epidemiology

THE STUDY OF THE DISTRIBUTION AND

DETERMINANTS OF HEALTH RELATED STATES AND
EVENTS IN POPULATIONS AND THE APPLICATION OF
THIS STUDY TO CONTROL OF HEALTH PROBLEMS“
Types of studies in epidemiology

• Descriptive studies
– describe occurrence of outcome

• Analytic studies
– describe association between
exposure and outcome
 Design study Epidemiology
• Descriptive study
– Population:
(Correlation or Ecology study)
– Individuals:
Case report
Case series
Cross-sectional
• Analytical Study
– Observational study
Case-control study
Cohort study (retrospective and prospective)
– Experimental study
Clinical trial
Community trial
Ecology Study
 Ecological studies
a. Unit analysis: group of people
(Posyandu, Puskesmas, district, province,
country etc)
b. These may compare disease frequencies
among different groups during the same
period, or compare disease frequencies
in the same population at different
points in time as a function of some
exposure.
20/03/2023 MK Epidemiologi Gizi (DBR)
a. Ecological studies usually are quick and
easy to perform, and can be undertaken
with already available information, but
great care is needed to avoid reaching
conclusions based on spurious
associations.
b. Ecological studies cannot link exposure
to outcome in a given individual.

20/03/2023 MK Epidemiologi Gizi (DBR)

 Ecologic studies are the design
of choice in certain situations:
• When the level of inference of interest is at the
population level
– Food availability (e.g., Goldberger et al: Public Health Rep
1916;35:2673-714).
– SES inequality and health
– Effects of tax in cigarette sales

• When the variability of exposure within the population

is limited
– Salt intake and hypertension (Elliot, 1992)
– Fat intake and breast cancer (Wynder et al, 1997)
 CONTOH
Worldwide epidemiological studies have compared
sugar consumption and levels of dental caries
at the between-country level:
 Sreebny correlated the dental caries experience (DMFT) of
12-year-olds with data on sugar supplies of 47 countries and
found a significant correlation (+0.7)
 Woodward In countries with a consumption level of sugar
<18 kg per person per year caries experience was consistently
<DMFT.
Decayed, Missing, Filled Teeth (DMFT) Index
20/03/2023 MK Epidemiologi Gizi (DBR)
 References:
Sreebny LM. Sugar availability, sugar consumption and
dental caries. Community Dentistry and Oral
Epidemiology, 1982, 10:1--7.

Sreebny LM. Sugar and human dental caries. World

Review of Nutrition and Dietetics, 1982, 40:19--65.

Woodward M, Walker AR. Sugar consumption and dental

caries: evidence from 90 countries. British Dental Journal,
1994, 176:297--302.

20/03/2023 MK Epidemiologi Gizi (DBR)

 CONTOH
Miyazaki & Morimoto reported a significant correlation
(r=+0.91) between sugar availability in Japan and DMFT at
age 12 years between 1957 and 1987. Populations that had
experienced a reduced sugar availability during the Second
World War showed a reduction in dental caries which
subsequently increased again when the restriction was lifted
(Takeuchi, Sognnaes).
Weaver observed a reduction in dental caries between 1943
and 1949 in areas of northern England with both high and low
concentrations of fluoride in drinking-water.
20/03/2023 MK Epidemiologi Gizi (DBR)
 References:
Miyazaki H, Morimoto M. Changes in caries prevalence in
Japan. European Journal of Oral Sciences, 1996, 104:452--458.

Takeuchi M. Epidemiological study on dental caries in Japanese

children before, during and after World War II. International
Dental Journal, 1961, 11:443--457.

Sognnaes RF. Analysis of wartime reduction of dental caries in

European children. American Journal of Diseases of Childhood,
1948, 75:792--821.

Weaver R. Fluorine and wartime diet. British Dental Journal,

1950, 88:231--239.

20/03/2023 MK Epidemiologi Gizi (DBR)

 ADVANTAGES
• CONDUCTED AT GROUP LEVEL, NOT AT INDIVIDUAL
LEVEL, HENCE RELATIVELY EASY TO DO AND QUICK
• USE EXISTING DATA
• GENERATE AND SUPPORT NEW HYPOTHESES
• ECOLOGICAL STUDIES CONDUCTED OVER TIME ON A
SPECIFIC GEOGRAPHICAL AREA ARE MORE
CONVENIENT TO PERFORM AND FORM HYPOTHESES
RATHER THAN STUDYING WHOLE POPULATIONS OR
SAMPLES

20/03/2023 MK Epidemiologi Gizi (DBR)

 Ecological fallacy
“The bias that may occur because an
association observed between variables
on aggregate levels does no necessarily
represent the association that exists at the
individual level.”

John M Last: Dictionary of Epidemiology, 1995

 Example of ecological bias*

Population A

$10.5K $34.5K $28.5K $12.2K $45.6K $17.5K $19.8K

Mean income: $23,940 Traffic injuries: 4/7=47%

Population B

$12.5K $32.5K $24.3K $10.0K $14.3K $38.0K $26.4K

Mean income: $22,430 Traffic injuries: 3/7=43%

Population C

$28.7K $30.2K $13.5K $23.5K $10.8K $22.7K $20.5K

Mean income: $21,410 Traffic injuries: 2/7=29%

*Based on: Diez-Roux, Am J Public Health 1998;88:216.
 60
Traffic injuries (%) Ecologic
50
analysis
40
30 Higher income is
20 associated with
10 higher injury rate
0
21 22 23 24 25

Mean income (US$, in 1000)

 Example of ecological bias*

Population A

$10.5K $34.5K $28.5K $12.2K $45.6K $17.5K $19.8K

Mean income: $23,940 Traffic injuries: 4/7=47%

Population B

$12.5K $32.5K $24.3K $10.0K $14.3K $38.0K $26.4K

Mean income: $22,430 Traffic injuries: 3/7=43%

Population C

$28.7K $30.2K $13.5K $23.5K $10.8K $22.7K $20.5K

Mean income: $21,410 Traffic injuries: 2/7=29%

*Based on: Diez-Roux, Am J Public Health 1998;88:216.
 60
Traffic injuries (%) Ecologic
50
analysis
40
30 Higher income is
20 associated with
10 higher injury rate
0
21 22 23 24 25

Mean income (US$, in 1000)

Individual-based
Non cases analysis

Injury cases have

Injury lower mean income
cases than non cases

0 10 20 30 40

Mean income (1000 US$)

• Which of the two levels of inference is wrong?
– Concluding that high income is a risk factor for injuries
(based on the ecologic data) is subject to ecologic fallacy.
– BUT … concluding that, because injury cases tend to have
lower income, communities with higher average income
should have lower injury rates is also wrong!

• The real problem is cross-level reference*

– Using ecologic data to make inference at the individual level
(ecologic fallacy).
– Or using the individual data to make inferences at the group
(population level).

• When used to make inferences at the proper

level, both approaches might be right.
*Morgenstern: Ann Rev Public Health 1995;16:61-81.
 PRAKTIKUM: STUDI EKOLOGI

1. Rumuskan tujuan/pertanyaan penelitian

2. Sebutkan peubah yang akan diteliti
3. Jelaskan sumber data dan cara
memperolehnya
4. Jelaskan analisis statistik yg digunakan
5. Dalam 1 lembar

20
 Types of Analytical Study:

1. Observational Studies

2. Intervention Studies
Analytical Study: Observational Studies
• non-experimental
• observational because there is no subjects
intervention/treatment
• exposures occur in a “non-controlled”
environment
• Subjects can be observed prospectively,
retrospectively, or currently
Case Control Study
 Case-Control Studies
Examples :
• The relationship between
thalidomide and unusual limb
defects in Germany
• Studies smoking and lung
cancer in 1950, establishing the
method in epidemiology.
 Case-Control Studies
• Type of analytic study
• Unit of observation and analysis: Individual
(not group)
• Case-control studies are the most frequently
undertaken analytical epidemiological studies
• They are practical approach for identifying risk
factors for rare diseases
 Design

• At baseline:

– Selection of cases (disease) and controls (no

disease) based on disease status

– Exposure status is unknown

 Case Control Study Design

Exposed
Diseased
(Cases)
Not Exposed
Target
Population
Exposed
Not Diseased
(Controls)
Not Exposed
 Selecting Cases

• Select cases after the diagnostic criteria and

definition of the disease is clearly established

• The study need not include all cases in the

population

• Cases may be located from hospitals, clinics,

disease registries, screenings, etc.
 Selecting Cases (cont.)

• Incident cases are preferable to prevalent cases for

reducing recall bias

• The most desirable way to obtain cases is to

include all incident cases in a defined population
over a specified period of time
 Selecting Controls

• Controls should come from the same

population at risk for the disease as the
cases

• Controls estimate the exposure rate to be

expected in cases if there were no
association between exposure and disease
 Selecting Controls (cont.)

• Multiple controls can be used to help add

statistical power when cases are difficult to obtain

• Using more than one control group lends

credibility to the results

• More than 3 controls for a case is usually not cost-

efficient
 Public Health, 127 (2013): 241-246

Body mass index, waist circumference, waist-

hip ratio, waist-height ratio and risk for type 2
diabetes in women: A case-control study

L. Radzevicien_e*, R. Ostrauskas
Institute of Endocrinology, Medical Academy,
Lithuanian University of Health Sciences, Eiveniu 2,
50009 Kaunas, Lithuania
Article history:
Received 27 September 2011
Received in revised form 6 July 2012
Accepted 3 December 2012
Available online 2 January 2013
 summary
Objective: To assess the relationship between various
anthropometric indexes and risk for type 2 diabetes in
women.

Study design and methods: A case-control study of 168

cases with newly diagnosed type 2 diabetes and 336
controls who were free of the disease. Cases and
controls (ratio 1:2) were matched by age (5 years). A
questionnaire was used to collect information on
possible risk factors for type 2 diabetes. Odds ratios (OR)
and 95% confidence intervals (CI) for type 2
diabetes were calculated by conditional logistic
regression.
Results: After adjustment for possible confounders,
increased risk for type 2 diabetes was associated with
body mass index (BMI) 30 kg/m2(OR 4.68, 95% CI 2.09-
10.49), waist circumference (WC) >88 cm (OR 6.99, 95%
CI 1.60-30.42)and waist-height ratio (WHtR) 0.5 (OR 3.15,
95% CI 1.91-15.81).

Conclusions: Both general and central obesity are

associated with type 2 diabetes. The results suggest that
high BMI, WC and WHtR are significant risk factors for
type 2 diabetes in women.
 Sources of cases and controls
CASES CONTROLS
All cases diagnosed in the Sample of general population
community
All cases diagnosed in a Non-cases in a sample of the
sample of the population population
Sample of patients in all
All cases diagnosed in all
hospitals hospitals who do not have the
disease
All cases diagnosed in a
Sample of patients in the same
single hospital
hospital who do not have the
disease
Any of the above methods Spouses, siblings or associates
of cases
 Assessing Exposure

• Exposure is usually an estimate unless past

measurements are available
– It has to be assumed that the exposure incurred at
the time the disease process began (this may not be
valid)
• Exposure estimates are subject to recall bias and
interviewer bias
• Potential confounders need to be accurately assessed
in order to be controlled in the analysis
 BASIC ANALYSIS:

Data is expressed in a four-fold table, and

an odds ratio is calculated  
Cases Controls

Exposed a b
Unexposed c d

OR = (a/c):(b/d)=ad/bc
 Odds Ratio (OR):

• A ratio that measures the odds of exposure for cases

compared to controls

• Odds of exposure = number exposed  number unexposed

• OR = ad/bc
 Calculating the Odds Ratio

Disease Status
CHD cases No CHD
(Cases) (Controls)
Exposure Smoker 112 176
Status Non- 88 224
smoker
Total 200 400
AD 112 x 224
Odds Ratio = = = 1.62
BC 176 x 88
 Interpreting the Odds Ratio

The odds of exposure for cases are 1.62

times the odds of exposure for controls.

or
 Interpreting the Odds Ratio

Those with CHD are 1.62 times more likely to be

smokers than those without CHD

or
Those with CHD are 62% more likely to be
smokers than those without CHD
 OR<1 OR=1 OR>1

Odds of exposure Odds of Odds of exposure

Odds exposure are for cases are
for cases are less
comparison equal among greater than the
than the odds of
between cases cases and odds of exposure
exposure for
and controls controls for controls
controls

Exposure
Exposure
Exposure as a reduces Particular
increases
risk factor for disease risk exposure is not a
disease risk
the disease? (Protective risk factor
(Risk factors)
factors)
• statistical testing is by simple chi-square
(unmatched analysis)

• or by McNemar’s chi square (matched-pairs

analysis)

• Can be extended to multiple strata (Mantel-

Haenzel chi-square)
 Strength of Association

Relative Risk;(Prevalence); Odds Ratio Strength of Association

0.83-1.00 1.0-1.2 None
0.67-0.83 1.2-1.5 Weak
0.33-0.67 1.5-3.0 Moderate
0.10-0.33 3.0-10.00 Strong
<0.01 >10.0 Approaching Infinity

Source:Handler,A, Rosenberg,D., Monahan, C., Kennelly, J. (1998).

Analytic Methods in Maternal and Child Health. p. 69.
ORs, P-Values and 95% CIs for Case-Control
Study with 3 Different Sample Sizes
Sample Size

Parameter
n=20 n=50 n=500
Computed

OR 2.0 2.0 2.0

p-value 0.500 0.200 0.001

95% CIs 0.5, 7.7 0.9, 4.7 1.5, 2.6

 Possible Sources of Bias and Error

• Information on the potential risk factor

(exposure) may not be available either from
records or the study subjects’ memories
• Information on potentially important
confounding variables may not be available
either from records or the study subjects’
memories
 Possible Sources of Bias and Error (cont.)

• Cases may search for a cause for their disease and

thereby be more likely to report an exposure than
controls (recall bias)
• The investigator may be unable to determine with
certainty whether the suspected agent caused the
disease
 Possible Sources of Bias and Error (cont.)

• Identifying and assembling a case group

representative of all cases may be difficult
• Identifying and assembling an appropriate
control group may be difficult
Advantages of Case-Control Studies
• Quick and easy to complete, cost effective

• Most efficient design for rare diseases

• Usually requires a smaller study population

than a cohort study
 Disadvantages of Case-Control Studies
• Uncertainty of exposure-disease time
relationship

• Not efficient for studying rare exposures

• Subject to biases (recall & selection bias)

 Preventive Medicine 57 (2013): S31–S33

Habitual physical activity reduces risk of ovarian

cancer: A case–control study in southern China

Andy H. Leea, Dada Sua, Maria Pasalicha, Yut Lin

Wongb, Colin W. Binnsa
a
School of Public Health, Curtin University, Perth,
WA, Australia
b
Department of Social & Preventive Medicine,
University of Malaya, Kuala Lumpur, Malaysia
Objective. To ascertain the relationship between habitual
physical activity and the risk of ovarian cancer among
southern Chinese women.

Method. A case–control study was conducted in

Guangzhou, Guangdong Province, during 2006–2008.
Information on physical activity exposure and lifestyle
characteristics was obtained from 500 incident ovarian
cancer patients and 500 hospital-based controls (mean
age 59 years) using a validated and reliable questionnaire.
Logistic regression analyses were performed to assess the
association between physical activity levels and the
ovarian cancer risk.
Results. The control subjects reported significantly
longer duration of strenuous sports and moderate
activity in daily life than the ovarian cancer
patients. Increased engagements in such leisure
time activities were associated with educed cancer
risks after adjustment for confounding factors. A
significant inverse dose–response relationship
was also found for total physical activity exposure,
with adjusted odds ratio 0.49 (95% CI 0.35–0.68) for
women engaged in 23 or more metabolic
equivalent tasks (MET)-hours per week relative to
those less than 12 MET-hours per week.
Conclusion. The study provided evidence of an
inverse association between habitual physical
activity and the risk of ovarian cancer, which is
important for the promotion and encouragement
of leisure time exercise activities to prevent the
disease.
 PRAKTIKUM: STUDI KASUS-KONTROL

1. Rumuskan tujuan/pertanyaan penelitian

2. Tentukan populasi dan cara penetapan subjek
penelitian (kasus dan kontrol)
3. Jelaskan pengukuran outcome dan exposure
4. Jelaskan analisis statistik yg digunakan
5. Jelaskan hasil penelitian tsb
6. 1 lembar

55
Cohort Study Design
Cohort study
 Cohort Study Design

Concurrent Retrospect
Define Population
1995 1975

Non-randomizing

2005 1985
Exposed Non-Exposed

2015 Disease No Disease Disease No Disease 1995

 Cohort Studies

• In a cohort study, subjects with an exposure

are identified and the incidence of a disease
over time is compared with that of controls
(persons who do not have the exposure).
• In a longitudinal study, subjects are followed
over time with continuous or repeated
monitoring of risk factors or health outcomes,
or both.
 Cohort Studies
– an “observational” design comparing
individuals with a known risk factor
(exposure) with others without exposure
– looking for a difference in the risk (incidence)
of a disease over time
– best observational design
– data usually collected prospectively (some
retrospective)
 DAFTAR PUSTAKA
1. Hennekens, C and Buring JE. 1987. Epidemiology in Medicine . Little Brown
Company. Boston.
2. Bea Skhole R, R. Bonita & T. Kjellstrom. 1996. Basic Epidemiology. WHO.
Geneva.
3. Murti, B. 1997. Prinsip dan Metode Riset Epidemiologi. Gadjah Mada
University Press.
4. Murti, B. 2010. Desain dan ukuran sampel untuk penelitian kuantitatif dan
kualitatif di bidang kesehatan. Gadjah Mada University Press.
5. Sastroasmoro S dan Ismael S. 2008. Dasar-dasar metodologi penelitian klinis.
CV Sagung Seto.
 Journal on nutrition and epidemiology
1. British Journal of Nutrition (http://journals.cambridge.org
2. Journal of Nutrition http://www.nutrition.org
3. American Journal of Clinical Nutrition http://www.ajcn.org
4. Asia Pacific Journal of Clinical Nutrition
http://apjcn.nhri.org.tw
5. Journal Critical Reviews in Food Sciences and Nutrition
http://www.tandf.co.uk
6. Journal of Early Adolescence
http://jea.sagepub.com/
7. Journal of Clinical Epidemiology
http://www.jclinepi.com
8. Journal of Annals Epidemiology
http://www.annalsofepidemiology.org/
9. International Journal of Behavioral Nutrition and Physical Activity
http://www.ijbnpa.org/
 Cont’d
10. Journal of the International Society of Sports Nutrition
http://www.jissn.com/
11. International Journal of Sport Nutrition and Exercise Metabolism
http://www.humankinetics.com
12. Journal of the American College of Nutrition
http://www.jacn.org/
13. Journal of Nutrition Education and Behaviour
http://www.jneb.org/
14. Journal of Psychology and Psychological Therapy
http://omicsonline.org
15. American Journal of Epidemiology
http://aje.oxfordjournals.org/
16. Bulletin of the World Health Organization
http://www.who.int/bulletin/en/
 Cont’d
17.Epidemiological Review
http://epirev.oxfordjournals.org/
18.European Journal of Epidemiology
http://www.springer.com/public+health/journal/10654
19.International Journal of Epidemiology
http://ije.oxfordjournals.org/
20.Journal of Clinical Epidemiology
http://www.jclinepi.com/
21.UNICEF (Infant & Children health)
http://www.unicef.org/index.php
22.SCC-SCN (Nutrition issues)
http://www.unscn.org/
23.IDPAS (Iron Deficiency)
http://www.idpas.org/
24.US-CDC (public healths)
http://www.cdc.gov/
25.INACG (Nutritional Anemia)http://www.globalhealth.org/