Study Designs

Case-control, cohort, and cross-sectional studies all compare outcomes between exposed and unexposed groups, but they sample differently and estimate different measures: odds ratios, risk ratios, or prevalence ratios. Choosing the wrong design can make causal inference impossible.

The idea

A case-control study samples on outcome: find cases (disease) and controls (no disease), then look back at exposure. It estimates the odds ratio: (odds of exposure | case) / (odds of exposure | control). You cannot estimate risk because you fixed the case:control ratio.

A cohort study samples on exposure: find exposed and unexposed groups, then follow forward to measure incidence. It estimates the risk ratio directly: P(disease | exposed) / P(disease | unexposed). A cross-sectional study samples once and measures both at the same time. It estimates prevalence, not incidence, and cannot establish temporal order.

The maths

Try it

Exposed: 100

With disease: 80

Odds ratio

0.00

Risk ratio

1.00

OR = 0.00, RR = 1.00

Interactive demonstration of epidemiology concepts

Where I used it

Easy to get wrong

Sources

From self-study of epidemiology texts, focused on observational study design.