Managerial Epidemiology: Assignment Week 3

Chapter 6

Study Designs: Ecologic,

Cross-Sectional, Case-

Control

Learning Objectives

• Define the basic differences between observational and experimental epidemiology

• Identify an epidemiologic study design by its description

• List the main characteristics, advantages, and disadvantages of ecologic, cross-sectional, and case-control studies

• Describe sample designs used in epidemiologic research

• Calculate and interpret an odds ratio

How Study Designs Differ

• Number of observations made

• Directionality of exposure

• Data collection methods

• Timing of data collection

• Unit of observation

• Availability of subjects

Observational vs. Experimental

Approaches

• Manipulation of study factor

– Was exposure of interest controlled by

investigator?

• Randomization of study subjects

– Was there use of a random process to

determine exposure of study subjects?

Typology of Epidemiologic

Research

Overview of Study Designs

• Experimental studies

• Quasi-experimental studies

• Observational studies

– Descriptive studies: cross-sectional

surveys

– Analytic studies: many ecologic studies,

case-control studies, cohort studies

The 2 by 2 Table Represents the

Association Between Exposure and

Disease Status

Ecologic Studies

• The unit of analysis is the group, not the

individual.

• They can be used for generating

hypotheses.

• The level of exposure for each individual

in the unit being studied is unknown.

• Generally makes use of secondary data.

• Advantageous with cost and duration.

Types of Ecologic Studies

• Ecologic comparison study—involves an

assessment of the correlation between

exposure rates and disease rates among

different groups over the same time period.

• Ecologic trend study—involves correlation of

changes in exposure with changes in disease

within the same community, country, or other

aggregate unit.

Example of an Ecologic

Correlation

• The association between breast

cancer and dietary fat for 39

countries.

• High intakes of dietary fats

associated with high rates of breast

cancer mortality.

Examples of Questions Investigated

by Ecologic Studies

• Is the ranking of cities by air pollution levels

associated with the ranking of cities by mortality

from cardiovascular disease, adjusting for

differences in average age, percent of the

population below poverty level, and

occupational structure?

• What are long-term trends (1950-1995) for

mortality from the major cancers in the US,

Canada, and Mexico?

Applications of Ecologic

Approach

• The effect of fluoridation of the water

supply on hip fractures

• The association of naturally occurring

fluoride levels and cancer incidence rates

• The relationship between neighborhood or

local area social characteristics and health

outcomes

The Ecologic Fallacy:

Definition

• Observations made at the group level may

not represent the exposure-disease

relationship at the individual level.

• The ecologic fallacy occurs when incorrect

inferences about the individual are made

from group level data.

Implications of the Ecologic

Fallacy

• The conclusions obtained from an

ecologic study may be the reverse of

those from a study that collects data on

individual subjects.

The Ecologic Study: Example

• An ecologic study examines 10 individuals who

go into the sun.

• The study finds that 7 persons (70%) have

sunburned foreheads although 6 persons (60%)

wore hats.

• The expected number of sunburned foreheads

is 4 (the number who did not wear hats).

• The media report that wearing hats will not

protect you from sunburn.

What the Individual Data Show

Individual Data (cont’d)

• From the individual data, one observes

that 100% of persons (4) who did not wear

hats were sunburned.

• Among persons who wore hats (6), only

50% were sunburned.

• This conclusion reverses the conclusion

from the ecologic data, i.e., that wearing

hats affords little protection from sunburn.

Ecologic Studies: Advantages

and Disadvantages

• Advantages

– Quick, simple, inexpensive

– Good approach for generating hypotheses

when a disease is of unknown etiology

• Disadvantages

– Ecological fallacy

– Imprecise measurement of exposure and

disease

Cross-Sectional Study

• Also termed prevalence study

• Exposure and disease measures obtained

at the individual level.

• Single period of observation

• Exposure and disease histories are

collected simultaneously.

• Both probability and non-probability

sampling is used.

Cross-Sectional Study:

Examples

• Surveys of smokeless tobacco use among high school students

• Prevalence surveys of the number of vasectomies performed

• Prevalence surveys of cigarette smoking among Cambodian Americans in Long Beach, California

Uses of Cross-Sectional

Studies

• Hypothesis generation

• Intervention planning

• Planning health services and

administering medical care facilities

• Estimation of the magnitude and

distribution of a health problem

• Examine trends in disease or risk factors

that can vary over time

Limitations of Cross-Sectional

Studies

• Limited usefulness for inferring disease etiology

• Do not provide incidence data

• Cannot study low prevalence diseases

• Cannot determine temporality of exposure and disease

Overview of Case-Control

Studies

• In a case-control study with two groups, one

group has the disease of interest (cases) and a

comparable group is free from the disease

(controls).

• The case-control study identifies possible

causes of disease by finding out how the two

groups differ with respect to exposure to some

factor.

Characteristics of the Case-

Control Study

• A single point of observation

• Unit of observation and the unit of analysis

are the individual

• Exposure is determined retrospectively

• Does not directly provide incidence data

• Data collection typically involves a

combination of both primary and

secondary sources.

Selection of Cases

• Two tasks are involved in case

selection:

–Defining a case conceptually

–Identifying a case operationally

Sources of Cases

• Need to define a case conceptually

• Ideally, identify and enroll all incident cases in a defined population in a specified time period

• A tumor registry or vital statistics bureau may provide a complete listing of all cases

• Medical facilities also may be a source of cases, but not always incident cases

Selection of Controls

• The ideal controls should have the same

characteristics as the cases (except for the

exposure of interest).

• If the controls were equal to the cases in all

respects other than disease and the

hypothesized risk factor, one would be in a

stronger position to ascribe differences in

disease status to the exposure of interest.

Sources of Controls • Population-based controls--Obtain a list that

contains names and addresses of most

residents in the same geographic area as the

cases.

– A driver’s license list would include most

people between the ages of 16 and 65.

– Tax lists, voting lists, and telephone

directories

– Patients from the same hospital as the cases

– Relatives of cases

Measures of Association Used

in Case-Control Studies

Disease Status

Yes (Cases)

No (Controls)

Yes

A

B

E x

p o

s u

re

S ta

tu s

No

C

D

A+C B+D

Odds A/C B/D

Odds Ratio AD/BC

Case-Control Studies

Sample Calculation

• On the association between chili pepper

consumption and gastric cancer risk: a

population-based case-control study

conducted in Mexico City

• Source: Lopez-Carillo, et al. Am J

Epidemiol. 1994;139:263-71.

Sample Calculation (cont’d)

Chili Pepper Consumption Cases of Gastric

Cancer

Controls

Yes A = 204 B = 552

No C = 9 D = 145

The OR (unadjusted for age and sex) is:

AD = (204)(145) = 5.95

BC (552)(9)

Interpretation of an Odds Ratio

(OR)

• OR = 1 implies no association.

• Assuming statistical significance:

– OR = 2 suggests cases were twice as

likely as controls to be exposed.

– OR<1 suggests a protective factor.

Odds Ratio (cont’d)

• An OR provides a good

approximation of risk when:

– Controls are representative of a target

population.

– Cases are representative of all cases.

– The frequency of disease in the

population is small.

Examples of Case-Control

Studies

• Young women’s cancers resulting from utero

exposure to diethylstilbestrol

• Green tea consumption and lung cancer

• Maternal anesthesia and development of fetal

birth defects

• Passive smoking at home and risk of acute

myocardial infarction

• Household antibiotic use and antibiotic resistant

pneumococcal infection

Advantages of Case-Control

Studies

• Tend to use smaller sample sizes than

surveys or prospective studies

• Quick and easy to complete

• Cost effective

• Useful for studies of rare diseases

Limitations of Case-Control

Studies

• Unclear temporal relationships between

exposures and diseases

• Use of indirect estimate of risk

• Representativeness of cases and controls

often unknown

Key Points to Remember

• Descriptive studies: cross-sectional

surveys (hypothesis generation)

• Analytic studies: ecologic, case-control,

and cohort (hypothesis testing)

Conclusion

• Study designs differ in a number of key

respects, including the unit of observation;

the unit of analysis; the timing of exposure

data in relation to occurrence of disease

endpoint; complexity; rigor; and amount of

resources required.