{"entity":{"id":"subgroup-analysis","kind":"term","name":"Subgroup analysis (forest plots)","aka":["subgroup","subgroups","subgroup analysis","subgroup analyses","pre-specified subgroup","prespecified subgroup","subgroup effect","subgroup benefit","forest plot","forest plots","interaction test","test for interaction","heterogeneity of effect","consistent across subgroups","benefit was consistent","effect modification","biomarker subgroup","PD-L1 subgroup","high-volume subgroup","regional subgroup","Asian subgroup","Western subgroup","subgroup effects"],"tldr":"Splitting a trial's patients into groups (by age, sex, biomarker, region) to see whether the treatment worked the same in each. Shown as a forest plot. Genuine differences are rare and most striking subgroup results are noise, so they need confirmation.","summary":"Trials are powered for the whole population, so subgroups have wide confidence intervals and, with a dozen subgroups, one will look different by chance; the proper test is for interaction, not for significance within each subgroup. Regulators have nonetheless used subgroups to restrict labels (PD-L1 cut-offs in gastro-oesophageal cancer, IMpassion130's PD-L1-positive population) and to question generalisability (regional differences in STARGLO, ex-US benefit in IMbrave050). Biologically plausible, pre-specified subgroups with a stratified randomisation carry more weight; a post-hoc TP53 wild-type subgroup with no interaction test (selinexor in endometrial cancer) needed its own trial.","asOf":"2026-09-09","wikipedia":"https://en.wikipedia.org/wiki/Subgroup_analysis","links":[{"label":"Wikipedia","url":"https://en.wikipedia.org/wiki/Subgroup_analysis"}],"tags":[],"related":["prespecified-vs-post-hoc","statistical-significance","pd-l1-testing","trial-failure-modes","biomarker-stratified-design","stratified-randomisation"],"cancers":[],"sections":["drug-discovery"],"technologies":[],"targets":[],"drugs":[],"companies":[],"institutions":[],"pathways":[],"terms":[],"trials":["rxponder","progesterone-preop-tmh","magnitude"],"people":[],"bottlenecks":[],"keyPapers":[],"journals":[],"dependsOn":[],"notes":[],"category":"Trials"},"route":"/terms/subgroup-analysis/","neighbours":{"term":[{"id":"biomarker-stratified-design","kind":"term","name":"Enrichment and biomarker-stratified designs","route":"/terms/biomarker-stratified-design/"},{"id":"pd-l1-testing","kind":"term","name":"PD-L1 expression testing (22C3, SP142, SP263)","route":"/terms/pd-l1-testing/"},{"id":"prespecified-vs-post-hoc","kind":"term","name":"Pre-specified vs post-hoc analysis","route":"/terms/prespecified-vs-post-hoc/"},{"id":"statistical-significance","kind":"term","name":"Statistical significance (P values, alpha, multiplicity)","route":"/terms/statistical-significance/"},{"id":"stratified-randomisation","kind":"term","name":"Stratified randomisation, allocation concealment and minimisation","route":"/terms/stratified-randomisation/"},{"id":"trial-failure-modes","kind":"term","name":"Why trials fail: underpowered, wrong endpoint, control arm drift, subgroup fishing, crossover","route":"/terms/trial-failure-modes/"}],"section":[{"id":"drug-discovery","kind":"section","name":"Drug Discovery Platforms","route":"/fronts/drug-discovery/"}],"trial":[{"id":"magnitude","kind":"trial","name":"MAGNITUDE","route":"/trials/magnitude/"},{"id":"rxponder","kind":"trial","name":"RxPONDER (SWOG S1007)","route":"/trials/rxponder/"},{"id":"progesterone-preop-tmh","kind":"trial","name":"Single-injection depot progesterone before breast surgery (Tata Memorial)","route":"/trials/progesterone-preop-tmh/"}]}}