How personalized are benefit and harm results of randomized trials? A systematic review

OBJECTIVES: This study aimed to review the degree of personalization of benefit and harm in the reporting of recent high-profile randomized controlled trials (RCTs) involving pharmacological interventions. STUDY DESIGN AND SETTING: This study is a systematic review of RCTs published between 2012 and 2017 with at least one intervention evaluating drug therapy and meeting the "high-profile" threshold in a premier academic literature abstraction service. Our primary outcome was the proportion of trials reporting subgroup analyses of a combined benefit-harm outcome. Secondary outcomes included the proportion of trials reporting subgroup analyses or clinical prediction guide for benefits or harms. We assessed the quality of the subgroup analyses using a modified version of previously published credibility criteria. RESULTS: Of 296 eligible RCTs, nine studies (3%) reported a combined benefit-harm endpoint. We found subgroup analyses of a combined benefit-harm endpoint in three studies (1%), a benefit endpoint in 167 studies (56.4%), and a harm endpoint in 18 studies (6.1%). The overall quality of the subgroup analyses was poor. Only one study reported a clinical prediction guide for an outcome. CONCLUSION: Despite great interest in the personalization of therapies, it is rarely reported in high-profile trials. Lack of rigorous and widely accepted methods may be the major barrier.