Data and safety monitoring board issues raised in the VA Status Epilepticus Study

Abstract

The Department of Veteran Affairs Status Epilepticus Cooperative Study was a randomized, multicenter clinical trial testing four intravenous drug regimens (lorazepam, phenobarbital, phenytoin and diazepam followed by phenytoin) to treat generalized convulsive status epilepticus. During the course of the study, two problems emerged that the study's data and safety monitoring board (DSMB) was required to address: poor recruitment and an unexpected difference in 30-day mortality between treatment groups. By the first annual DSMB meeting, recruitment was only 25.6% of expected. The DSMB recommended placing the study on probation and replacing poorly performing sites. At their second annual meeting, the DSMB recommended approval of proposed changes to the study design contingent on the study leadership removing nonproductive sites. These changes were a 2-year increase in the recruitment period and a change in study design that decreased required sample size. Nonproductive centers were terminated and the approved changes allowed the study to be successfully completed. At the second annual DSMB meeting, an unexpected doubling of mortality rates between drug groups was observed. Although not statistically significant, the finding raised serious concerns for patient safety. The DSMB recommended instituting monthly reporting on mortality and suggested additional analyses for exploring why the differences could be occurring. These analyses indicated that, by chance, older and sicker patients were being randomized to the drugs with the higher mortality rates. By the end of the study, the observed differences in mortality between drug groups had evened out. The DSMB's thoughtful recommendations, support and monitoring ensured that the study was successfully completed without endangering the study patients.