Case Study: Confirmation of Fraud - Diaries Completed by Staff

In a Phase III trial in vascular disease, over 4,500 patients were randomized across 160 sites. The CRF included 480 variables for 32 domains. The sponsor had a specific concern and the CluePoints solution was deployed after the last patient last visit to prove how significant the issue might be.

More than 45,000 p-values were generated by the SMART™ engine as part of the statistical analysis. The scoring algorithm summarized the information from the tests into a summary indicator or “score” that pinpoints atypical centers. In this analysis, center R38 was among the centers with the highest scores. Individual tests revealed that most of the signals were focused on patient diary data. The patients’ answers in these diaries showed a distribution of values that were completely different compared to the same data from other centers. The scores are given in Table below for the 5 top sites of interest (i.e. sites with the lowest score). It is interesting to note that in addition to site R38 which was known to have potential issues, other centers also exhibited atypical patterns.

In an ongoing trial, it would have been beneficial to trigger on-site monitoring visits in all sites identified as outliers in the CluePoints analysis.

This post originally appeared in some form on CluePoints.