The methodology that will be used for statistical analysis of the research data (and the timing(s) of the methodologies being applied) should be defined either within the protocol itself or (as is likely for the more complex clinical trials) within a separate document, a Statistical Analysis Plan (SAP). A statistical analysis plan should not be confused with the data management plan; there should be clear distinction between the activities associated with data management of a study data and statistical analysis even if these activities are performed by the same person (as delegated).
The pre-specified plan for statistical analysis should be finalised by appropriate personnel (for example, the trial statistician) before the study commences and before release of any randomisation codes for blinded studies. Noclor may request evidence of this review (either sign off of the protocol or SAP), if this is a requirement it will be clearly communicated to the Chief Investigator further to the initial /bespoke risk assessment (and documented in any subsequent study-specific monitor plans).
The finalised plan for analysis (whether it is within the protocol or a separate SAP) should be followed. The Chief Investigator is responsible for ensuring that the analysis plan (where this is a separate document) remains consistent with the version of the protocol, any changes should be version controlled and retained in the Study/Trial Master File (S/TMF).
The processes for provision of interim/final datasets for analysis and reporting as per the requirements of the statistical analysis plan should be detailed in the data management plan (for example, details of who will perform the database lock and how data will be transferred, to who and when etc). All outputs of statistical analysis must be appropriately version controlled (tables, figures etc) to ensure that the resulting reports are accurate.