Abstract
This is an accepted article with a DOI pre-assigned that is not yet published.
Risk-based quality management (RBQM) is used to help ensure that clinical trial data are fit-for-purpose and that trial quality is protected. RBQM begins by identifying critical-to-quality (CtQ) factors and risks related to such CtQ factors, defining controls such as key risk indicators and data checks, and reviewing data to detect signals. Many organizations can identify signals through centralized monitoring and analytics, but a persistent gap remains in determining whether a signal represents a true issue and what action, if any, is appropriate given study data and site context.
The authors met as a cross-functional group with experience in clinical data management, statistics, quality, clinical operations, and clinical trial technology. Participants represented both sponsor and service provider organizations and brought practical experience from active trials. The group met to discuss how RBQM is applied today and how organizations might better move from identified signals to appropriate, data-supported actions.
This paper presents a consensus view of the capabilities needed to close this gap. The group concluded that agentic artificial intelligence (AI) is unlikely to be useful in this setting unless it is grounded in structured study context, well-understood data, and an issue management process. Neither statistical analysis nor AI can establish causality; what they can do is focus attention and support structured investigation so that study teams can determine the underlying reason for a signal and decide on an appropriate action. The capabilities described here are offered as design considerations for the field rather than as an endorsement of any product or implementation.
Keywords: Manage Clinical Research Data, Measure Data Quality, Identify Discrepancy, Store Data, Metadata, Track Data