Title: A Brief Introduction to Epidemiology IX Epidemiologic Research Designs: CaseControl Studies
1A Brief Introduction to Epidemiology -
IX(Epidemiologic Research Designs Case-Control
Studies)
- Betty C. Jung, RN, MPH, CHES
2Learning/Performance Objectives
- To develop an understanding of
- What case-control studies are
- The value of such studies
- The basic methodology
- Pros and Cons of such studies
3Introduction
- Epidemiology studies the distribution of disease
in a number of ways. - The two major categories of epidemiological
studies are Observational and experimental
studies. - Most epidemiological studies are observational.
4Epidemiological Study Designs
- Observational Studies - examine associations
between risk factors and outcomes (Analytical -
determinants and risk of disease, and descriptive
- patterns and frequency of disease) - Intervention Studies - explore the association
between interventions and outcomes. (Experimental
studies or clinical trials)
5Research Designs in Analytic Epidemiology
- Ecologic Designs Cross-Sectional Study
- Case-Control Study
- Cohort Study
6Case-Control Studies
- Flashback Studies (Paffenbarger, 1988)
- Retrospective - compare cases and controls for
presence of disease - Includes passage of time.
- Historical - assess past characteristics or
exposures in two groups of people- cases and
controls
7Examples
- The relationship between thalidomide and unusual
limb defects in Germany - The relationship between meat consumption and
enteritis necroticans in Papua New Guinea
8Value
- Simple to conduct
- Cost-effective way to study a rare disease
9Case-Control Studies Methodology
First Select the Cases and Controls
Cases
Control (without disease)
(with
disease)
Then measure, post-exposure
Were Exposed
A
B
D
C
Not Exposed
AC
BD
Population Exposed
A/AC
B./BD
10Case Control Design
- Time
- Direction of Inquiry
Cases with the Disease
Exposed
Population
Not Exposed
Controls without the disease
Exposed
Not Exposed
11Case-Control Studies
- Cases - Has condition or health outcome of
interest. Has higher frequency or greater degree
of exposure than non-cases. - Controls (non-cases) - Does not have the health
condition. Serves as the comparison group - Ask about history of contact with or exposure to
supposed causes
12Case-Control Studies
- If controls are well chosen, the only antecedent
difference will be in the level of a
characteristic that is related causally to the
development of a disease (I.e, exposure to a
chemical resulted in cancer). - Quantify with odds ratios
13Strength 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 - lt0.01 gt10.0 Approaching
Infinity
14Methodology Issue Matching
- Matching - control for confounding variables. If
you do not match then control by subject
selection (study only males to eliminate gender
as a confounding variable) - Matching
- Subject selection
- Statistical control during data analysis
- The more variables that need to be matched the
greater the universe we need. - Problem - match age, sex and SES - Control must
be the same.
15Methodological Weaknesses
- Biased reporting of the antecedent (having lung
CA -gt patients over reporting smoking (from
guilt, knowledge or selective memory) - Subject selection (decreases with cases and
controls in the same facility) - Limited to only cases, who have survived at the
same time. Selective survival
16Pros
- Cases easily available
- Good for less common or rare cases
- Quick, inexpensive
- Can be conducted by clinicians in clinical
facilities - Tend to support, not prove causal hypothesis by
establishing associations - Historical data available in clinical records
- Number of subjects needed is small
17Cons
- Info about antecedents depends on memory, which
could lead to bias - Clinical data may be inadequate or incomplete
- Case group may not be homogenous - criteria
for diagnosis may differ. - Clinical cases are selective survivors
18Cons
- Non-representativeness of cases. Those coming in
for treatment may differ from those not seeking
treatment and those going somewhere else. - Antecedent is not obtained from universe of all
antecedents. - Berksons fallacy - making generalizations from
hospital or clinical samples to the general
population. - Cannot know what association would be for all or
for a representative sample of all people having
the antecedent.
19References
- For Internet Resources on the topics covered in
this lecture, check out my Web site - http//www.bettycjung.net/