Computing End-of-Trial Estimates of Risk for Trials that Use Interim Safety Monitoring
View Presentation View Presentation
*Michael J Dallas, Merck Research Laboratories 

Keywords: Interim monitoring of adverse events; end-of-trial p-value, point estimate, confidence interval.

Proper end-of-trial statistical analyses must account for interim evaluations where stopping the trial is a possibility. While it is standard practice to adjust terminal statistical analyses due to opportunities to stop only for “positive” findings, e.g., conventional group-sequential efficacy trials, adjusting due to opportunities to stop only for “negative” findings are also important, especially in the context of a safety outcome. Here, a method to account for such designs and an illustration of its application in the context of a safety trial is presented. The method is demonstrated to have advantages over naive (i.e., unadjusted) methodology that is commonly employed with respect to estimates of risk.