Risk-adjusted Case-Control Study — Covariate-controlled Retrospective Design
Risk-adjusted Case-Control Study · Also known as: adjusted case-control study, covariate-adjusted case-control, risk-stratified case-control study, matched and adjusted case-control study
A risk-adjusted case-control study is an observational design that identifies individuals with a disease outcome (cases) and comparable individuals without it (controls), then uses statistical adjustment — most commonly multivariable logistic regression — to estimate the association between an exposure and the outcome while controlling for confounding risk factors. The adjustment step is what distinguishes this variant from a simple case-control study, producing odds ratios that better reflect the independent contribution of the exposure of interest.
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.
When to use it
Use a risk-adjusted case-control study when the outcome is rare or has a long latency, making prospective follow-up impractical or unethical; when multiple exposures need to be examined simultaneously; and when administrative databases or registries already capture exposure and covariate data retrospectively. Risk adjustment is mandatory whenever cases and controls differ on known confounders — which is almost always in clinical and epidemiological settings. Do not use this design when the exposure is rare (the reference design then shifts to a cohort study), when temporality between exposure and outcome cannot be established from records, when reliable covariate data are unavailable (adjustment cannot be performed without measured confounders), or when the outcome prevalence exceeds roughly 10% and an odds ratio would seriously overestimate the relative risk.
Strengths & limitations
- Efficient for rare diseases or outcomes: cases already exist, so no waiting for events to accumulate.
- Multivariable adjustment removes the distorting effect of measured confounders, yielding more valid exposure-outcome estimates than unadjusted analyses.
- Can examine multiple exposures in a single study with a relatively modest sample size.
- Faster and less costly than prospective cohort studies for the same research question.
- Nested case-control variant within an established cohort further controls for time-varying confounding.
- Cannot directly estimate incidence rates or absolute risks — only odds ratios, which approximate relative risk only when outcome prevalence is low.
- Susceptible to recall bias: cases may remember exposures differently from controls, especially for distant or stigmatised exposures.
- Selection bias is a persistent threat if the control group is not truly representative of the source population that gave rise to the cases.
- Adjustment is limited to measured confounders; unmeasured or unknown confounders remain a source of residual confounding.
- Retrospective data quality depends on record completeness and accuracy, which is outside the researcher's control.
Frequently asked
What is the difference between matching and statistical adjustment in a case-control study?
Matching controls to cases on specific variables (e.g., age, sex) during the design phase reduces confounding by making cases and controls more comparable before analysis. Statistical adjustment via logistic regression accounts for confounders in the analysis phase, regardless of whether matching was done. When individual matching is used in the design, conditional logistic regression must be used in analysis to preserve the matched structure; multivariate adjustment alone without conditional regression is insufficient in that setting.
How many confounders can I adjust for?
A widely cited rule of thumb is at least 10 outcome events (cases) per covariate included in the logistic regression model. Fewer events per variable lead to overfitting, unstable coefficient estimates, and wide confidence intervals. If you have more candidate confounders than this threshold allows, consider a priori selection guided by a directed acyclic graph (DAG), or use penalised regression methods such as LASSO.
When does the odds ratio approximate the relative risk?
The odds ratio closely approximates the relative risk when the outcome is rare in the source population — conventionally below 10% prevalence. When the outcome is common, the OR exaggerates how large (or small) the relative risk is. In those situations, report the OR with explicit acknowledgement of this limitation, or use alternative estimators such as the prevalence ratio via log-binomial or Poisson regression.
What is a nested case-control study and how does it improve risk adjustment?
A nested case-control study is embedded within a defined prospective cohort. Cases are individuals who develop the outcome during follow-up; controls are sampled from cohort members who have not yet developed the outcome at the time each case occurs (risk-set sampling). Because both cases and controls come from the same well-characterised cohort, baseline covariate data are often more complete and measured before the outcome occurs, reducing recall bias and improving the quality of risk adjustment compared with a standalone case-control study.
Can I use propensity score methods instead of multivariable logistic regression for risk adjustment?
Yes. Propensity score matching, stratification, or inverse probability weighting are valid alternatives for risk adjustment in case-control studies, particularly when there are many confounders relative to the sample size. However, unlike regression adjustment, propensity score methods in case-control studies require careful handling of the sampling fraction and are most naturally applied in cohort or cross-sectional designs; specialist guidance is advisable before applying them in a case-control context.
Sources
- Schlesselman, J. J. (1982). Case-Control Studies: Design, Conduct, Analysis. Oxford University Press. ISBN: 978-0195029697
- Rothman, K. J., Greenland, S., & Lash, T. L. (2008). Modern Epidemiology (3rd ed.). Lippincott Williams & Wilkins. ISBN: 978-0781755641
How to cite this page
ScholarGate. (2026, June 3). Risk-adjusted Case-Control Study. ScholarGate. https://scholargate.app/en/epidemiology/risk-adjusted-case-control-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
- Logistic RegressionResearch Statistics↔ compare
- Matched case-control studyEpidemiology↔ compare
- Propensity Score MatchingResearch Statistics↔ compare