Thanks for the quick response. I re-read the article again and realized that the authors actually defined what they meant by it -- time it took from randomisation to time randomized treatment was stopped because of ineadequate seizure control, intolerable side effect or both.
So, if the study concluded that drug X was significantly better than drug Y in terms of time to treatment failure would it mean that drug X worked better than drug Y and therefore had a longer time to treatment failure?