Proof of efficacy trials: endpoints.

Seizure frequency data are difficult to work with because they are so variable both between and within patients. Seizures can come in clusters, thereby adding to the variation in seizure frequency. Also, seizure frequency data are non-parametric in nature, making them difficult to analyze by some statistical techniques without transforming them. Therefore, to improve data analysis in proof of efficacy studies, manipulations of seizure frequency are commonly used, such as the proportion of responders above an arbitrary threshold, or the response ratio. Other useful endpoints are time to nth seizure, number of seizure-free days, proportion of seizure-free patients, and assessment of seizure severity using rating scales. The choice of endpoint(s) will depend upon the design of the study.