abstract
- In randomized controlled trials, investigators often explore the possibility that the treatment effects differ between subgroups (eg, women vs men, old vs young, more versus less severe disease). Investigators often inappropriately claim subgroup effects (also called "effect modification" or "interaction") when the likelihood of a true effect modification is low. Criteria for assessing the credibility of subgroup analyses, nicely summarized in a formal Instrument for Assessing the Credibility of Effect Modification Analyses (ICEMAN), include investigator postulation of a priori hypotheses with a specified direction; support from prior evidence; a low likelihood that chance explains the apparent subgroup effect; and only testing a small number of subgroup hypotheses. PATIENT SUMMARY: Randomized clinical trials often use subgroup analyses to explore whether a treatment is more or less effective in a particular patient subgroup (eg, women vs men, old vs young). In this mini-review, we explore the common pitfalls of subgroup analyses.