![IconGems-Print](images/IconGems-Print.png)
191 – Biomarkers and Endpoint Validation II
Estimating Adverse Event Rate Differences Using Data from Blinded Trials: The Canary in the Coal Mine
A. Lawrence Gould
Merck Research Laboratories
William W.B Wang
Merck Research Laboratories
The development of therapeutics whose mechanisms of action may extend beyond their target indications has led to a need to identify potential toxicities even in ongoing blinded clinical trials. Recently issued FDA rules regarding safety reporting requirements raise the possibility of breaking the blind for serious adverse events that are not the clinical endpoints of a blinded study. However, unblinding individual cases of frequently occurring adverse events could compromise the overall validity of the trial. The possibility of elevated risk can be addressed without unblinding the trial using external information about adverse event rates among patients not receiving the test product in populations similar to the trial population. This article describes a Bayesian approach to determining the likelihood of elevated risk suitable for binomial or Poisson likelihoods that applies regardless of the metric used to express the difference. The method appears to be particularly appropriate when the adverse events are not 'rare'.