The Statistical Analysis Plan (SAP) is a document created by a statistician. It describes in technical detail what the statistical analysis will contain, and how any statistical testing is to be done. It is in fact a detailed elaboration and specification of the statistics section in the study protocol. When given to two independent programmers, they should end up with exactly the same statistical analysis. All planned statistical analyses are then set in stone, without any possible variations.
Per ICH-E9 guideline, the SAP has to be approved prior to database lock and study unblinding. This is very important, as it prevents data-driven bias. The greatest issue would be data-driven changes in statistical methodology, because that raises the suspicion that the “result then looks better”. Deviations from the planned analyses have to be mentioned in the study report. Deviations that have been introduced prior to database lock are often listed in the SAP, together with the reason for this deviation. Deviations that are introduced after database lock will obviously not be in the SAP, but still need to be clarified in the study report.
Often the SAP is accompanied by a set of Mock TLFs. The Mock TLFs are a sample book of empty shell tables, listings and figures (TLFs). They provide practical guidance to the programmers, showing them what type of statistical outputs are expected, also in terms of specific lay-out. This document should not contain any specifications on statistical methodology: all that belongs in the SAP. When reviewing the SAP, the study team can use the matching Mock TLFs as a help to better understand the planned analyses.
The Emtex statistician has over 25 years of experience in writing and reviewing SAPs and Mock TLFs.
Reviewing a document that is as technical as a SAP can be a daunting task for the study team. For that reason, a Statistical Analysis Plan Outline is sometimes created. This can be a PowerPoint slide deck containing a high level outline of how endpoints will be derived from the collected data, and how they will then be analyzed and presented. This is a very useful visual tool for in-depth team discussions on how the study data will be presented, without the team getting lost in the technical descriptions. This SAP Outline can then serve as a very useful starting point for the actual SAP. With the study team thus already aligned, the SAP review cycles will go smoothly. It also helps to identify up front which additional analyses might be of interest, and would allow these to be added to the SAP. The key for the SAP Outline is “open discussion”. The Emtex statistician can bring valuable experience into this discussion via the SAP Outline document.