Data management in clinical investigations – Part 3: Data traceability
Data traceability is a foundational principle for clinical investigations aiming to demonstrate the performance and safety of medical devices. As several information systems (CTMS, TMF, ISF, eCRF, ePRO, SAE database, Analysis software etc.) are used in collecting and processing data, systematic planning is needed to ensure consistent implementation of traceability. Labquality offers data management services as part of the broader CRO clinical investigations offering.
Clinical investigation objectives and procedures are described in the clinical investigation plan (CIP), and approved by ethics committees and competent authorities. Based on and traceable to CIP and its possible amendments during the investigation, a technical variable listing is composed and maintained containing the unambiguous technical specifications of data items collected. This includes how the questions in the eCRF are formulated and the value list options for list-type questions. Question formulations should be traceable to a data collection standard and value list items to a controlled vocabulary to support regulators and others reviewing the data.
Data is collected in eCRF, and each collected piece of data must be traceable to a specific version of CIP and eCRF, to a person who entered, modified or signed the data and to an audit trail of previous values. A value in the eCRF must be traceable to the unalterable source data outside of the eCRF. Similarly, protocol deviations during the investigation are traced back to a data subject enabling transparent identification and analysis of subjects included in the as-per-protocol analysis set.
A specific issue of clinical investigations for medical devices, especially software as a medical device, is the traceability of the treatment itself (release-version-revision-build at any given time during the study) to source code and frozen design documentation.
Statistical analysis is specified in the statistical analysis plan (SAP) and is based on a locked and reviewed study database including traceable post-lock corrections. Statistical analysis traceability covers both statistical analysis code including code correctness quality reviews and preprocessing steps from raw data, and logging of the code execution including a version of the statistical analysis software, libraries, IDE, OS, errors and warnings etc.
This post was written by Labquality‘s Data Manager Markus Vattulainen. You can contact him for more information.