How would you test for effect modification by age when studying a drug exposure and an outcome in a cohort?

• A. Include an interaction term between exposure and age in regression and report stratum-specific effects
• B. Stratify by age and report separate effects without an interaction term
• C. Exclude older participants
• D. Use only crude analysis

Answer: (A)

Effect modification by age means that the drug’s effect on the outcome changes across different ages. To test this in a cohort, you build a regression model that includes an interaction term between the drug exposure and age. This product term directly tests whether the association between exposure and the outcome depends on age. If the interaction term is statistically significant, or if the age-stratified effects show different magnitudes, you have evidence that age modifies the effect and you can present stratum-specific results to illustrate the pattern. You can model age as a continuous variable by including exposure×age, or use age categories and test the interaction between exposure and those categories, depending on your data and pre-specified plan.

This approach is preferred because it provides a formal test of modification while adjusting for confounders and keeps the full sample. Simply stratifying by age and reporting separate effects without an interaction term does not quantify or test whether the modification is real. Excluding older participants or relying only on crude analyses would remove generalizability and fail to address effect modification.