What is Database Lock in a Clinical Trial?

What does database lock mean in clinical trials? DBL meaning

Database lock (also known as data lock, or DBL) is a critical step in clinical trial data management which occurs after trial data has been collected, verified, and cleaned. Database lock is the process of approving and finalizing the completed data set, and “locking it” to further changes. When the clinical database is locked, no further changes can be made to the data without authorization from a third party stakeholder.

Importance of database lock

Database lock is an important process that ensures data quality, accuracy, and completeness, and is a key factor in producing a final dataset for analysis that can serve as the basis for generating reliable and robust findings. It is a critical point that signifies the end of data collection for a clinical trial.

DB locks prevent unknown discrepancies between source documents and collected data, which could lead to errors in reports and conclusions drawn from the results. It also acts as a form of quality control by protecting against conflicting information, potential frauds and inaccurate observations from unreliable sources. Database locking thus enables researchers to have confidence that their evidence is reliable and trustworthy when sharing their findings with others.

Soft lock vs hard lock: Types of database lock in clinical trials

There are two stages or levels of database locks commonly used in clinical trials; soft lock and hard lock. Soft lock describes a state wherein all clinical data from case report forms (CRFs) has been entered into the database and known queries have been resolved. Permissions for editing the soft-locked database are generally limited to the clinical data manager from this point, although access rights may be maintained, as it is generally expected that minor corrections may still be made to a database in soft lock.[1] Database soft lock may also be called “database freeze.”

The database is then “hard locked” after a quality assurance review is completed and the data is considered to be finalized, at which point no further changes should be made to the data. In specific cases, the hard locked clinical trial databases can be unlocked, if required, with written authorization from the sponsor. Generally speaking, softlocked databases can be unlocked relatively easily to correct discrepancies, whereas a hard locked database should be considered ‘final’ and requires a more involved procedure for unlocking.

Requirements for locking database in clinical trials

Database lock can be initiated once the trial protocol has been completed and all clinical data has been collected and entered into the database. All data queries should be resolved and the data should be in a verified and validated state. There should be evidence of data reconciliation between finalized CRFs and other source documents such as patient medical records, and clinical monitors should certify that they have completed required safety reviews for each subject and reviewed any discrepancies in individual patients’ data. In other words, it should be possible to demonstrate specific steps that were carried out in order to prepare the database for locking and finalization.

The procedure for database lock, including preparatory steps, specific methodologies to be used, and tracking and auditing processes, should be outlined in an SOP, such as this example from the Cambridge Clinical Trials Unit.

The database locking process

Once all of the above conditions are met, the database can be soft locked. With the database locked, generally only the lead clinical data manager will have edit rights and any changes will require unlocking.

At this point, a quality assurance process will take place, wherein a random sample of the study data will be analyzed. The number of records audited is determined as the square root of the total number of records included in the database. For example, if a study involved 840 individual case report forms, then 29 records would be randomly selected and checked for errors by comparing to the original data source. The FDA establishes an acceptable error rate of 0.05%.

Once the QA step is completed and any remaining queries are resolved, database hardlock can be declared. In blinded studies, unblinding is performed after database lock in order to minimize the risk of bias. The locked database can then be used to generate final data tables and the statistical analysis can be carried out.

Database unlocking

Database unlocking is the process of unlocking a database that is in a locked state (either softlocked or hardlocked) to correct discrepancies or otherwise modify the data. Unlocking is generally avoided for hardlocked databases, unless a critical error is identified that requires correction within the data set. Since final databases have been reviewed and verified, errors might be detected later on by a statistician or medical writer; if the error is in a critical data field, the database will need to be unlocked to correct it.

The unlocking procedure for a hard locked or final database must be established in the study protocol and detailed in SOPs. During the change, an audit trail will document which data field was changed, by whom, why, and when. The database can then be re-locked following the same procedure for hardlocking used previously.

If amendments are only needed for non-critical fields, unlocking a hard locked database is generally not warranted. In such cases, a memo to the team or note to file can be generated instead of unlocking the database.

Unlocking should only be done under specific circumstances, and requires written authorization from the sponsor or an external stakeholder/auditor (as defined in the SOPs). All changes made, including those comprising the unlocking procedure itself, must be carefully tracked and documented.

Conclusion

Database lock in clinical trials is a crucial process that helps researchers ensure data accuracy and consistency, and which prevents the study data from unwanted or unauthorized changes after data has been verified and all corrections have been made. Database locking usually proceeds via two steps - soft lock and hard lock - and the hard locked database, considered “final,” is used in the analysis to generate the trial’s final results. Clinical trial databases can be unlocked under specific circumstances, namely when a critical error is identified which requires correction within the database. Both database lock and unlocking procedures should be clearly defined and outlined in SOPs in order to guarantee regulatory compliance.