Skip to main content

Table 4 Approach to multiplicity due to subgroup analyses

From: Approaches to multiplicity in publicly funded pragmatic randomised controlled trials: a survey of clinical trials units and a rapid review of published trials

a) Review: multiplicity approach taken
  Formal adjustment Hierarchical testing Other approach None
Subgroup analyses 8/85a (9%) 0/85 (0%) 0/85 (0%) 77/85b (91%)
b) Survey: responses to posed scenarios
  Yes No Unsure
Would you consider adjusting for multiplicity arising from performing multiple subgroup analyses? 6/27 (22%) 17/27 (63%) 4/27 (15%)
Consider a parallel group trial with multiple subgroup analyses performed. Would you adjust for multiplicity in the following scenarios?    
Subgroup analyses pre-specified in the study protocol? 3/27 (11%) 22/27 (81%) 2/27 (7%)
Subgroup analyses determined post-hoc? 4/27 (15%) 22/27 (81%) 1/27 (4%)
Subgroup analyses specified for the following reasons: a) to confirm biological plausibility, b) to confirm existing hypotheses, AND c) to show subgroup effects for supporting decision making in target populations. 3/27 (11%) 19/27 (70%) 5/27 (19%)
Would you be more likely to adjust for multiplicity if the number of subgroup analyses was increased? 5/27 (19%) 21/27 (78%) 1/27 (4%)
  1. Notes: a One trial performed a Bonferroni correction, two a Holm correction and five studies used a threshold of 1% for significance
  2. b Of these, five studies stated that results from secondary outcomes were exploratory/hypothesis generating