A subgroup analysis is done when a set of data is broken down into separate groups. For instance if I’m looking at data about lung cancer patients, I can do a subgroup analysis to only look at those who smoke <5 cigarettes, 5-10 cigarettes, etc.
According to Pocock et al. (2016), results from subgroup analyses in randomized controlled trials (RTCs) are “useful for generating hypotheses at best” and should only be used in designing future experiments, particularly when the primary outcome fails. Burke et al. (2015) outline some best practices for reporting subgroups to improve their credibility.
Synonyms:
sub-group analysis