Abstracts

Rationale: In order to attain optimal results from multicenter, multinational clinical trials, it is important to ensure homogeneity and accuracy in seizure classification. The Epilepsy Study Consortium reviewed seizure classifications for accuracy in subjects with treatment resistant partial onset seizures enrolled in a multinational investigational study sponsored by Supernus Pharmaceuticals, Inc. Methods: Sites were required to submit seizure identification forms immediately following Visit 1. Each form contained a detailed description of the subject s seizures and the seizure classification. This information was submitted to the Consortium and reviewed for accuracy for the first two subjects enrolled at each site. Discrepancies were followed up by one of the consortium s reviewers until resolved. After two forms were reviewed with no corrections, the site was approved. If corrections were required, the Consortium continued to review every subject s seizure identification form until approval criteria was met. Results: A total of 231 forms, containing 514 separate seizure descriptions and classifications were reviewed and reconciled. The following countries participated in this study: North America (USA, Canada, Mexico) and Eastern Europe (Bulgaria, Croatia, Poland, Romania, Russia). One or more seizures were misclassified on 29/231(13%) of the forms. Overall, 36/514 (7%) of the seizures were misclassified. The % of seizures within category that were misclassified were as follows: SPS w/o motor: 4%, SPS w/motor: 11%, CPS: 8%, 2 GTCC: 1%. Misclassifications by region: North America submitted 13/82 forms (16%) that contained at least 1 misclassified seizure and Eastern Europe submitted 16/149 (11%). Misclassifications by country: Only countries with more than 10 forms were assessed. The country with the highest number of misclassifications submitted 5/24 incorrect forms (21%) and the lowest had 0 (0%). See table below. Other Misclassifications: ? 2 seizure types were combined into 1 (should have been captured separately) ? Seizure was non-epileptic and removed in the 2nd version ? Investigator was unsure how to classify seizure ? After asking for clarification, new seizure types were added One subject experienced only simple partial seizures that were not observable and did not qualify for the study. The site classified the seizure correctly; however this subject may have been randomized if the review had not been performed. For this particular trial, SPS without an observable component will not be analyzed in the efficacy analysis. The 4 seizures that were classified incorrectly as SPS w/o motor would not have been included in the analysis. Conclusions: Seizure misclassification is frequent enough to warrant oversight at the time of enrollment. In a number of cases, identified errors in classification could have impacted trial results.