Analysing the Data
Extracting datasets from the database
Once the trial data has been cleaned and validated and all data issues resolved, the data should be locked with files protected from editing and deleting. These procedures should be in accordance with the sponsoring institution’s data management standard operating procedures (SOPs) and the trial-specific Data Management Plan.
Following database lock, datasets should be extracted from the trial database for analysis
Undertaking the statistical analysis
The Sponsor-Investigator should ensure that a Biostatistician is included in the clinical trial team. The Biostatistician plays a crucial role throughout the trial, from designing the trial to the analysis and interpretation of the trial data. The statistical analysis should be undertaken, or at least overseen, by the trial Biostatistician. Statistical analysis should be conducted in a statistical package such as Stata or R and it should be reproducible.
If the trial is contributing to a higher degree, the Biostatistician will be involved in an advisory capacity since students should do their own statistical analysis.
The analysis should be conducted according to methodology that was outlined and agreed by the trial team before the database was locked. This methodology should be documented in either the protocol or a formal Statistical Analysis Plan (SAP). Sponsor-Investigators should use the SAP template provided by the sponsoring institution, see JAMA Network: Guidelines for the Content of Statistical Analysis Plans in Clinical Trials.
Any deviation from the SAP should be explained in the reporting of the trial results, ie in any manuscript publishing the results and in the Clinical Study Report (applicable only to trials included as part of a regulatory submission).
Once the statistical analysis is complete, the trial Biostatistician should generate a report for the Sponsor-Investigator which is then used to support any publication of results or write the Clinical Study Report (if applicable). The Biostatistician should be involved in writing publications of the trial data to assist with interpretation of the results.