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

Table 1 Review: study characteristics

Characteristic		n/N	%
Journal	Annals of Internal Medicine	4/138	3%
	BMJ	11/138	8%
	JAMA	28/138	20%
	Lancet	30/138	22%
	NEJM	36/138	26%
	NIHR HTA journal library	17/138	12%
	PlosMED	12/138	9%
Trial design^a	Parallel group: 2 treatment groups	94/138	68%
	Parallel group: > 2 treatment groups	18/138	13%
	Cluster randomised	23/138	17%
	Crossover	2/138	1%
	Factorial	5/138	4%
	Stepped wedge	2/138	1%
	Non-inferiority	18/138	13%
	Equivalence	1/138	1%
Total number of randomised participants - median (IQR)		574	(312, 2043)
MULTIPLICITY
Primary outcome^b	More than one outcome stated	21/138	15%
	Two outcomes	17
	Three outcomes	4
	More than one comparison made	28/138	20%
	Two comparisons	16
	Three comparisons	4
	Four comparisons	2
	Five comparisons	1
	>Five comparisons (maximum 20)	5
Secondary outcome	More than one outcome stated	134/138	97%
	Median (IQR) outcomes stated	8 (5, 13)
	More than one comparison made	132/138	96%
	Median (IQR) comparisons made	14.5 (7, 26)
More than two treatment groups^c		23/138	17%
Number of treatment comparisons made	One^d	1
	Two	13
	Three	6
	Four	1
	Five	1
	Eight	1
Any subgroup analyses performed		85/138	62%
	Median (IQR) subgroup analyses	4 (2, 7)
Any interim analyses performed		38/135	28%
	One	22
	Two	9
	Three	3
	Four	3
	Five	1

Notes: ^a Trials could be classified in more than one design category, e.g. a cluster randomised, factorial, non-inferiority trial.
^bDiscrepancies between the numbers of outcomes stated and comparisons made were either due to multiple time points being analysed or multiple analysis approaches taken, with none stated as primary.
^cThis includes parallel group trials with > 2 treatment groups and factorial trials.
^dIn this trial one treatment arm was dropped due to futility at an interim analysis, so the final analysis comprised just two treatment groups and therefore one comparison.
Abbreviations: BMJ British Medical Journal, IQR interquartile range, JAMA Journal of the American Medical Association, NEJM New England Journal of Medicine, NIHR HTA National Institute of Health Research Health Technology Assessment, PlosMED Public Library of Science Medicine

ISSN: 1471-2288