A review study on comparing treatment effects among subgroups

Abstract

It is interested in examining treatment effect on a particular category of subjects or in comparing treatment effects among different subgroups as well as overall treatment effect due to heterogeneity of study subjects. Subgroup analyses are exceedingly common, but they are also often misleading. Conclusions based on subgroup analyses can do harm both when a particular category of people is denied effective treatment (a "false-negative" conclusion), and when ineffective or even harmful treatment is given to a subgroup of people (a "false-positive" conclusion). Because of the frequency and the importance of clinicaL.application of subgroup analysis, researchers need to be cautious about doing subgroup analyses. This study presents guidelines to help conducting subgroup analyses correctly.