Introducing Results Validation for Clinical Trials

In clinical trials, the path from raw data to analysis-ready datasets is complex and requires a deep understanding of both the data and its transformations. Reviewers of statistical analysis results must ensure that every step—from the initial raw data to the finalized ADaM (Analysis Data Model) datasets— is correctly implemented. This process is laborious and error-prone, as any error can stem either from issues in the data or the code that processes it.

When a result appears incorrect, therefore, root cause analysis represents a significant challenge. Determining the root cause is a time-consuming process that requires expertise in the programming language and codebase used for the analysis. Complicating matters, the separation of data from code in statistical computing environments and data management software often forces programmers to switch between systems, re-run parts of the analyses to inspect intermediate results, or using additional tools to explore data at different levels of derivation. This manual, repetitive effort not only consumes resources but also introduces the potential for human error at every step.

The Verisian Results Validation Module was designed to address these challenges by streamlining the results validation process in clinical trials. It automatically extracts, displays, and enables validation of results directly from SAS logs. When reviewing the TLF, its source data is displayed and can be explored using intuitive plots and filters. This allows users to dive into subsections of a TLF and identify possible data issues (Figure 1).

Variables required for the result are identified and explorable through Code Traceability with a single click (Figure 2).

Code traceability allows for the immediate and transparent exploration of data transformations alongside the code that produced them, providing a clear view of how each data point is derived from raw to analysis-ready form (Figure 3).

With this seamless integration, users can avoid time-consuming re-runs and manual data inspections. The results validation module provides instantaneous and integrated access of TLFs and relevant data and code, offering reviewers a level of transparency and efficiency previously unavailable. According to feedback from our partners, implementing Verisian’s system has reduced the time spent on root cause analysis and resolution by approximately 95%. These immense efficiency gains allow for faster, more accurate reviews, accelerating the delivery of high-quality clinical trial results.

Quality control (QC) of statistical analysis results is essential for all stakeholders in clinical trials. Whether a contract research organization (CRO) is delivering results to a sponsor, a sponsor is preparing to submit them to regulatory authorities, or regulators are evaluating a submission: ensuring the accuracy and validity of results is absolutely critical. Especially as technology advances and AI enters the workforce both as a companion and sole contributor, the importance of rigorous human-led QC processes only increases in importance.

The Verisian Validator, of which Results Validation is a component, represents a comprehensive solution to this challenge, integrating four key components (Specification, Code, Log, and Results Validation). By leveraging the Verisian Validator, organizations can shift away from the slow, costly process of double programming to a more effective review and iteration framework. This approach will not only save time and resources but also enhance the overall quality of the clinical trial results.

If you’d like to be part of our journey, we look forward to seeing you on our online forums or at our next events!