Skip to main content

Table 6 Survey: comments

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

Category Example comments
Consensus/awareness amongst trialists/clinicians (n = 11): ranging from needing consensus on which methods to use when, understanding when multiplicity adjustments are required, and clinician awareness “Lack of consensus among statisticians leaves a lot of uncertainty and makes CIs uncomfortable”
“Many trialists don’t know the different methods that can be used (or haven’t got the time to investigate their correct implementation) so a state of the art type review and a course for the most useful/suitable methods would be great”
“Deciding when it is required and providing justification when the decision is not to adjust”
Informal hypothesis testing (n = 2): including repeated presentation of primary outcome data by arm to DMCs, and data dredging  
Confidence intervals (n = 2) “We are supposed to be concentrating on measures of effect and confidence intervals, and downplaying p-values. How does this factor into multiplicity testing procedures?
Multi-arm trials (n = 2): including multiple treatment arms and adaptive trials  
Multiple outcomes (n = 2): including classifying the purpose of secondary outcomes, and risk/benefit synthesis  
Subgroup analyses (n = 1) “Design for subgroup effects on basis of meta-analysis including previous results”
Interim analyses (n = 1) “Minimise interim analyses”
Miscellaneous (n = 2): including the increased importance of personalised medicine and the lack of consensus of what the denominator is for significance  
  1. Abbreviations: CI Chief Investigator, DMC Data Monitoring Committee