Stefan Schandelmaier

Mar 28, 2024 | jamanetwork.com | Stefan Schandelmaier |Gordon Guyatt

Sherry et al1 considered the methodological quality of subgroup analyses reported in 379 oncology trials. The authors explored a number of fundamental problems with these analyses: in forest plots, use of linear rather than logarithmic scales and failure to present overall pooled estimates; failure to conduct and highlight tests of interaction; and readiness to make unwarranted inferences regarding the credibility of postulated subgroup effects.