Matched Cohort Study
Also known as: matched follow-up study, paired cohort study, propensity-matched cohort, matched prospective study
A matched cohort study is an observational design in which each exposed participant is paired with one or more unexposed counterparts who share key characteristics — such as age, sex, or comorbidity status — before both groups are followed forward in time to compare incident outcomes. Matching controls for measured confounders at the design stage, reducing bias that would otherwise require statistical adjustment alone.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
+1 more
When to use it
Use a matched cohort study when you have longitudinal individual-level data on an exposed and a readily identifiable unexposed population, and when a small set of strong confounders can be measured and controlled by design rather than only in analysis. It is especially valuable when the exposed group is small, when random assignment is unethical or impractical, and when preserving temporal sequence to infer incidence rates is important. Do not use it when key confounders cannot be measured (matching cannot control for unmeasured variables), when the pool of potential controls is too small to find adequate matches, or when the primary goal is prevalence rather than incidence — a cross-sectional design is then more appropriate.
Strengths & limitations
- Controls for measured confounders at the design stage, before any outcome is observed, reducing selection bias.
- Increases statistical efficiency by ensuring comparison groups are balanced on major nuisance variables.
- Preserves a temporal exposure-outcome sequence, supporting causal inference more strongly than cross-sectional designs.
- Well-suited to rare exposures where recruiting a naturally balanced unexposed group is difficult.
- Compatible with time-to-event analysis (Kaplan-Meier, stratified Cox), enabling incidence and hazard-ratio estimation.
- Can only control for confounders that were measured and used as matching variables; unmeasured confounding remains a fundamental threat.
- Over-matching — matching on a variable that is on the causal pathway — can bias the exposure estimate toward the null.
- If many exposed subjects cannot be matched (poor match rate), the analysable sample shrinks and results may not generalise to the full exposed cohort.
- Analysis is more complex than for unmatched cohorts; standard software must be instructed to account for the matched structure.
- Matching on time (calendar period or age) can introduce immortal-time bias if the index date is not defined symmetrically for exposed and unexposed.
Frequently asked
How does a matched cohort study differ from a matched case-control study?
Both designs use matching to improve comparability, but they differ in sampling direction. A matched cohort study starts with exposure status and follows participants forward to observe who develops the outcome, yielding incidence rates and hazard ratios. A matched case-control study starts with outcome status — it identifies cases who have already developed the disease, then matches controls who have not — working backward to assess prior exposure. When the outcome is rare or follow-up is long, the case-control design is more efficient; when incidence rates are of primary interest, the cohort design is preferred.
What is propensity-score matching and when should I use it?
Propensity-score matching estimates each participant's probability of being exposed given observed covariates (via logistic regression), then pairs exposed and unexposed individuals with similar scores. It is especially useful when there are many potential confounders that cannot all be matched on simultaneously. However, it only balances measured covariates; if important confounders are unmeasured, residual bias remains.
Can I use a 1:2 or 1:many match ratio?
Yes. Matching each exposed participant to two or more unexposed controls (1:2, 1:3, etc.) increases statistical power at the cost of finding suitable controls. Beyond a 1:4 ratio the efficiency gains become marginal. When using more than one control per case, the matched-set structure must still be preserved in analysis — for example, by stratifying Cox models by matched set.
How do I handle matched participants who are lost to follow-up?
Loss to follow-up is handled by censoring the participant at the last known contact date, exactly as in an unmatched cohort. If an exposed participant is censored early, the matched unexposed control is not automatically censored — they contribute follow-up time until their own censoring or outcome event. The matched-set Cox model accommodates this asymmetric censoring correctly.
What sample size considerations apply?
Sample size calculations for matched cohort studies should account for the expected match rate (the proportion of exposed participants who can be successfully matched), the intraclass correlation within matched pairs, and the anticipated outcome incidence in each group. Software packages such as Stata's 'power cox' or R's 'powerSurvEpi' can incorporate matching when parameters are specified correctly.
Sources
- Rothman, K. J., Greenland, S., & Lash, T. L. (2008). Modern Epidemiology (3rd ed.). Lippincott Williams & Wilkins. ISBN: 978-0781755641
- Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1), 41–55. DOI: 10.1093/biomet/70.1.41 ↗
How to cite this page
ScholarGate. (2026, June 3). Matched Cohort Study. ScholarGate. https://scholargate.app/en/epidemiology/matched-cohort-study
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Case-control studyEpidemiology↔ compare
- Cohort StudyEpidemiology↔ compare
- Nested case-controlEpidemiology↔ compare
- Propensity Score MatchingResearch Statistics↔ compare
- Prospective Cohort StudyEpidemiology↔ compare