In a case-cohort study, which measures of association may be used?

• A. Odds ratio or hazards ratio
• B. Hazard ratio only
• C. Relative risk only
• D. Rate ratio only

Answer: (B)

Case-cohort designs deal with time-to-event data and censoring, so the analysis centers on how quickly events occur over follow-up. With the subcohort sampling, the standard approach uses a Cox proportional hazards model with appropriate weighting to account for the design. The parameter that comes from this model is the hazard ratio, which expresses the relative instantaneous risk of the event at any given time between groups. This makes the hazard ratio the natural and valid measure of association in a case-cohort study.

Other measures rely on different concepts (such as cumulative incidence for risk, or simple rate comparisons) and would require additional modeling or assumptions to be unbiased under the case-cohort sampling. The odds ratio is tied more closely to case-control logic and time-to-event data with proper weighting isn’t as straightforward in this design, and relative risk or rate ratio aren’t the primary targets without extra modeling steps. Therefore, the hazard ratio is the appropriate measure of association in a case-cohort study.