Protocols and CRFs often change during a study. An ethics committee may request clarification. A protocol amendment may add a new endpoint. A safety review may require additional adverse event details. A data quality review may reveal that a field label is confusing. In such cases, CRFs and databases may need to be revised. Version control ensures that these changes are documented, approved, and implemented consistently.
Version control is necessary because data collection instruments are part of the study record. If different sites use different versions without documentation, the final dataset may combine variables collected under different definitions. For example, if one version of a CRF defines fever as temperature above 37.5 degrees Celsius and a later version defines fever as above 38.0 degrees Celsius, the outcome is not comparable unless the difference is documented and handled during analysis.
ACRF version control process should assign version numbers and dates to forms. It should document what changed, why it changed, who approved the change, when it became effective, which sites received the update, and whether previously collected data are affected. In electronic systems, database changes should also be recorded. REDCap includes project change review processes when moving from development to production, but the study team still needs its own governance and documentation.
Version control is especially important in multisite studies. A central team may revise a form, but one site may continue using old paper copies or may not receive updated guidance. This creates avoidable inconsistency. Implementation should therefore include communication, training, retrieval or replacement of old forms, confirmation of database updates, and monitoring to ensure the new version is being used.
Data managers should work closely with investigators and statisticians when CRF changes affect analysis variables. Some changes are minor, such as correcting a spelling error in a label.
Others are major, such as changing response options or adding a new outcome category. Major changes may require amendments, ethics notification, statistical review, or special handling in the analysis dataset.