Subgroup analysis in clinical trials deals with evaluation of treatment effects (both efficacy and safety) with respect to the primary and key secondary endpoints in subsets of patients defined based on baseline characteristics, including demographic, clinical, genetic and other variables.
Confirmatory subgroup analysis is performed in confirmatory Phase III trials with one or more prospectively defined subpopulations of patients that are more likely to benefit from the treatment than patients in the overall population. Analysis of the subgroups may result in a modified regulatory claim or additional regulatory claims.
Exploratory subgroup analysis is commonly utilized in Phase III trials and relies on a post-hoc subgroup search. Exploratory subgroup analysis can be performed in Phase III trials with a positive efficacy signal in the overall patient population to evaluate the benefit-risk profile of multiple subgroups. Based on the results of these evaluations, a subgroup may be excluded from a trial if the treatment mainly benefits patients in the complementary subgroup. Alternatively, if safety issues are detected in the overall population, a subgroup with an acceptable safety profile can be selected. Similar analyses can be also performed in failed Phase III trials to help identify one or more subgroups with a beneficial effect.
QSPI Subgroup Analysis Working Group was created in 2013 to provide general information on exploratory and confirmatory subgroup analysis. The working group is chaired by Ilya Lipkovich (Quintiles).
Survey on common practices for subgroup identification and analysis organized by the QSPI Subgroup Analysis Working Group in 2014. The survey was designed by Cristiana Mayer and Ilya Lipkovich.
Subgroup analysis bibliography (key publications on confirmatory and exploratory subgroup analysis). This page is maintained by Ilya Lipkovich and Alex Dmitrienko.
Subgroup analysis sofware (open-source software tools for subgroup identification and analysis). This page is maintained by Ilya Lipkovich.