Rdi As A Method For Reviewer Performance Monitoring In Bicr Setup For Improving Data Quality.

e18082 Background: Double read with adjudication is a preferred independent review model of regulatory authorities for blinded independent central review (BICR) in order to minimize reviewer bias in clinical trials. Adjudication rate (AR) and adjudication agreement rate (AAR) are commonly used indicators for monitoring independent reviewer performance quality. We will present data on Reader Disagreement Index (RDI), an innovative indicator for accurately monitoring reviewer performance and triggering timely intervention when applicable. Methods: A detailed review of BICR adjudication data was performed for 12 oncology clinical trials, with a total of 5,369 subjects (ranging from 119 to 894 per individual study) with 27,056 time points using RECIST, the Lugano classification or iwCLL assessment criteria. RDI for each reviewer was calculated as RDI = (# of cases where adjudicator disagreed with given reviewer ÷ Total # of all cases read) × 100, with high RDI indicating high % disagreement. RDI was used to identify the discordant reader (i.e. reviewer with the highest level of cases disagreed with by the adjudicator) when approximately 10% of the total reads were completed for each study. RDI was also calculated and compared with AR and AAR on an ongoing basis throughout the study. Mean RDI + standard deviation (SD) were used to identify outlier readers. Results: RDI reliably identified the most discordant reader consistently across all 12 studies, while AR & AAR did not. The results confirm the advantage of RDI as a lagging and leading indicator for independent reviewer performance across indications and criteria using double read with adjudication review model. RDI, when calculated as early as at the 10% of total reviewed cases benchmark, demonstrated a positive predictive value of 91% and negative predictive value of 93% (Sensitivity 71%; Specificity 98%). Conclusions: Early identification of an outlier reviewer as per RDI (i.e. after reviews completed for ~ 10% study visits), followed by detailed analysis and corrective measures, such as retraining of the reviewer can serve as timely intervention to improve review quality. Thus, RDI proves to be a better indicator for not just monitoring reviewer performance, but also as an excellent tool for triggering timely corrective intervention. [Table: see text]