Checklist for Statistician Before Data Base Lock (DBL) of Blinded Trials

Once the clinical trials have finished recruiting patients, management will look forward to Data Base Lock (DBL) and get their hands on the Executive Summary of results. This builds up a pressure on sites, clinical research associates, data management teams, and Stats/Programming teams. This may also lead to overlooking a few less significant, but still important, activities such as internal documentation. Even after Data base lock, timelines allow a maximum 2 to 3 days for Executive Summary results on locked databases and somewhere around 12-15 days to generate a complete set of outputs. Therefore, it is necessary to have a checklist in place for activities to be done before DBL to ease out the analysis after DBL and there are no unexpected erroneous situations which may delay the delivery of results.

The following could be a part of a checklist for Statistician’s review before the DBL of a blinded trial:

Database queries are resolved and closed: Statistician shall check with Data management team that all the queries raised during Data Base review have been resolved and closed by sites. If there are any pending issues, raise attention to superiors to expedite the process.

Queries from dry runs: Any queries raised during dry run must be resolved before Data base lock. Although, purpose of dry runs shall be solely focused on preparation of outputs creation, if there are any queries raised during statistical review of outputs, these shall be resolved. During statistical review, SAP and mockshells can also be updated if it is observed that current version of SAP is not covering the whole purpose of endpoint analysis.

Data Coding: Medical coding of AE terms, medical history, concomitant medications or any other information in datasets shall be completed before DBL and as per requirements of analysis plan.

Protocol Deviations review: Statistician shall discuss with Clinicians to review all the deviations reported by sites and categorize them by criticality (if needed). If there are any deviations that can be verified programmatically then those deviations shall be verified. All the protocol deviations shall be discussed for impact on analysis and categorized as major or minor (as defined in Statistical Analysis Plan).

Programs for outputs creation is complete: Programs for creating TFL outputs shall be completed and tested on datasets of dry run before DBL. Programs shall also be validated as per the validation plan. It shall be ensured that outputs required for executive summary are at least generated and validated on final complete database before DBL.

Dummy Randomization testing: Metadata from final randomization shall be used to create Dummy randomization for Dry Run. Programs shall be in place to merge randomization data with Database.

Documents finalization: Protocol, SAP, and mock-shells shall be finalized before DBL.

Review of actual Randomization Data: Study Statistician shall be available immediately after DBL to review the actual randomization data before merging it with locked database. Statistician shall review that the treatment codes are as expected, stratification factors (if any) are captured correctly, list of subjects is matching with list of subjects in database, and if there is anything additional in randomization list as per study design.