Download PDF

Database locking

Database Locking or Closing the study database is fundamental in preventing inadvertent or unauthorized changes once the final analysis and reporting of the data have begun. Although important in open-label studies, it is even more critical in randomized trials in order to preserve the integrity of the randomization process once the blind has been broken. A well-defined process for closing a database and change control procedures in the event of the necessity of reopening the database are essential.

Minimum Standards

  • Ensure there is a procedure defining database closure methodology.
  • Document completion of all defined tasks or criteria prior to database closure.
  • Ensure that all team members are notified and edit access is removed and documented at final database closure.
  • Have written procedures with clear criterion for unlocking a database after closure.

Database Closure Process and Checklist

Database lock must be documented as a definitive point in time, where proof of the removal of edit access can be shown. In order to decrease the necessity to unlock the database after this point, a well defined and organized procedure must be followed to ensure that all data have been processed the quality level assessed and relevant study personnel are notified or approve the database lock.

Items to consider in the database closure preparation include the following:

  • All data have been received and processed.
  • All queries have been resolved.
  • External data (e.g. electronic laboratory data) are reconciled with the study database and are complete.
  • If a separate, serious adverse event database exists, it is reconciled with the main study database.
  • The coding list has been reviewed for completeness and consistency.
  • Final review of logic and consistency check output has taken place.
  • Final review for obvious anomalies has taken place.
  • Quality audit of the data and documentation of the error rate have occurred.
  • All documentation is updated and stored where required by Standard Operation Procedures.

Once all steps are complete, a documented approval process should take place, which includes sign-off by relevant study personnel (e.g. Data Management, Biostatistics, monitoring representative, clinical/scientific representative). Once approvals have been obtained, edit access to the database should be removed and the date documented.

Errors Found After Database Closure

If, after database lock, errors are found, careful consideration should be given as to how to handle and document these errors. Issues to consider are primarily the effect on the safety and efficacy analysis. It is important to remember that not all errors found must be corrected in the database itself. Errors may also be documented in the statistical or clinical report. Although some companies choose to change all errors found, others may only change those that have a major impact on the safety/efficacy analysis. What is of primary importance is that a company has in place a predefined process to determine how such errors will be processed and documented.

If the database is unlocked after initial lock, the process must be well controlled and, once again, documented. Procedures should include notification of the project team, a clear definition of the change(s) being made and the date of the change. Re-locking the database should follow the same process as the initial lock for notification/approval.


Recommended Standard Operating Procedures

  • Database Closure, to clearly define the steps to be taken to close and lock a database. A task checklist is recommended for this purpose as well as sign-off forms for approval and lock details.
  • Change Control/Errors after closure to define conditions under which the database will be re-opened and the necessary documentation required for re-lock.
  1. Changes can be made by user with Privilege update only.


  1. New data can be added.


  1. Can add new subjects.


  1. Can run batch validation.


  1. Can unlock.
      1.No changes can made for existing data.

  1. Can’t add new data for existing subjects.


  1. Can’t add new subjects.


  1. Can’t perform batch validation.


  1. Can’t unfreeze