Recently I have been reviewing a number of trial protocols. One reflection is the growing trend of noting “we will write a full statistical analysis plan” without giving any further details. The assumption is that this is sufficient ... it is not. 1/6

The purpose of a protocol is so that others can understand the logic of the trial, the justification for it (clinical and methodologically) with sufficient details that anyone can critically appraise & understand them and replicate it for themselves. 2/6

It is important to justify design choices and explain the statistical implications of them for the design, the power of the trial and the stats analysis eg use of unequal randomisation ratios, clustered observations, time to event vs comparison of proportions etc 3/6

One also needs to be reassured that the power calculations are appropriate for the proposed analysis approach. They are intrinsically linked and can’t be adequately assessed if both elements are not provided. The analysis needs to have been considered before the trial starts 4/6

There is clear guidance available to help with the writing of a comprehensive trial protocol (the SPIRIT guidance). We should use it: https://www.spirit-statement.org  5/6

Finally, as well as listing the clinical lead and the local clinical contacts, protocols should always, in my view, also identify who is the trial statistician/methodological lead. Both the clinical & methodological elements are intrinsic to the design and delivery of a trial 6/6