Yesterday, we started a series of posts on protocol deviations by looking at how they are handled in GCP and regulations. Today, we ask our favorite question:  Why do we collect, classify, categorize, trend, track, and respond to protocol deviations? The regulations imply the following reasons:

• Compliance.  Investigators are responsible for protocol compliance, and sponsors must bring non-compliant investigators back into compliance or discontinue study drug shipments. Collection, classification, categorization, trending, and tracking are required to assess compliance; response is required to bring investigators back into compliance.
• Reporting. The investigator is required to document and explain deviations from the approved protocol, report changes made to eliminate "immediate hazards to subjects," and report "unanticipated problems involving risks to human subjects."  Sponsors are also required to report important protocol deviations in the Clinical Study Report.  To distinguish between these reporting requirements, we must collect, classify, and categorize.
• Analysis. Only subjects with the required minimal exposure to the treatment regimen, sufficient data to analyze, and "the absence of any major protocol violations, including the violation of entry criteria" are included in the per-protocol analysis. To identify subjects that do not meet criteria, collection and classification are required.

Note these three objectives are the responsibility of three different parties - investigators, clinical operations, and biostatistics.  Compliance and reporting to IRBs require real-time collection and assessment, whereas analysis and reporting in the CSR are post-hoc activities. Each objective requires stakeholders to distinguish a different subset of protocol deviations:  compliance is concerned with trends and serious breaches; reporting concerns "immediate hazards," "unanticipated problems," and "important" deviations; analysis looks at "major" deviations. These competing objectives and their inherent requirements for categorization and classification can result in a complex process that attempt to serve too many purposes.  In the next post, we'll look at how protocol deviations are typically collected, and how we can adjust the process to better achieve all three objectives.