Randomized controlled trials (RCTs) are considered the paradigm to study effects of medical interventions because randomization (theoretically) results in comparability of study groups.1,2 Confounding is therefore considered a nonissue in RCTs. Nevertheless, effects that are estimated within subgroups of RCTs can be confounded. This argument might seem rather obvious because the issue is just elementary epidemiology, but it has not received adequate attention (ie, minimal or no discussion on this issue is found in various standard texts on clinical trials). Confounding of subgroup effects has mainly been described in relation to individual patient data (IPD) meta-analyses.3,4 However, proposed methods to control for such confounding (ie, the so-called 2-stage approach) fail, as illustrated by the example described herein, which exemplifies confounding of subgroup effects and demonstrates how to adjust for it.
Groenwold RHH, Donders ART, van der Heijden GJMG, Hoes AW, Rovers MM. Confounding of Subgroup Analyses in Randomized Data. Arch Intern Med. 2009;169(16):1532–1534. doi:10.1001/archinternmed.2009.250
Browse and subscribe to JAMA Network podcasts!
Customize your JAMA Network experience by selecting one or more topics from the list below.
Create a personal account or sign in to: