Does accounting for seizure frequency variability increase clinical trial power?

•The expected range of seizure frequencies can be predicted several months in advance.•Using these predictions, a new trial analysis method ZV was introduced.•Compared to 50%-responder rates (RR50), ZV has higher statistical power to distinguish the placebo arm from the therapeutic arm.•Use of ZV in trial analysis may allow for design of epilepsy trials with decreased sample size and cost.