Case-Control Study
Learning Objectives
“A primary goal is to reach the same conclusions in a retrospective (casecontrol) study as would have obtained from a forward (cohort) study, if one had been done.”Source: Mantel, N, Haenszel,W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst. 1959; 22:719-48
Example of the Ideal Design
But we cannot achieve the counterfactual ideal
Population of HBV infected
Follow-up: 2000-2015
Measure Liver Ca Incidence
Different subjects Measure Liver Ca Incidence
Population of HBV non-infected
Follow-up: 2000-2015
Desirable situations for a Case-Control Study
SampleControls =
Sample of the denominator
Representative with
regard to exposure
2 by 2 table
illNot ill
a
ExposedExposed
(unprotected)
illbExposed(unprotected)healthy
c
Not exposedNot exposed
(protected)
illdNot exposed(protected)healthy
Case-control study
ill
exposedhealthy
Not
exposed
Time
Case-control StudyCase-control Studiekrank exponiert gesund
illExposed Not exposed
Healthy 20
a c
b d
nicht exponiert
10
Zeit
10
60
a/c 20 x 60 OR= b/d= ad= 10 x 10 cb
= 12
95%-confidence Interval (CI)= 3,93 - 38,02 2
ad bc T2
N1 N 0 M1 M 0
p= 0,0000002
Example: Passive smoking and Breast Cancer
Fundamental conditions for the validity of this
case-control design
Solution:
Case Definition
Control definition
Why not use hospital controls?
Have you ever been exposed?
Unexposed= never active, never passive
Case-control study design
Results of the study
Interpretation of the Odds Ratio Alternatively:
(in what situation?)