Subgroup analyses are generally fraught.1 Although clinicians have always been concerned about subgroup effects—the possibility that a diagnostic, treatment, or management modality might work better, worse, or not at all in some subsets of patients—their eyes tend to glaze over when methodologists talk about subgroup analyses, viewing these discussions as tempests in a statistical teapot. This is unfortunate because in various quality, performance, and patient satisfaction measurement attempts, emergency departments (EDs) and physicians are increasingly evaluated as subgroups.2 and 3 Comparative performance assessments among EDs or physicians on door-to-balloon time, time to antibiotics, waiting time, patient satisfaction, ED throughput, or billings are now regularly reported. These are all fundamentally subgroup analyses, with the ED or the physician as the subgroup.
(June 2010)
Annals of Emergency Medicine
2010
http://www.sciencedirect.com.ezp-prod1.hul.harvard.edu/science/article/pii/S0196064410003768