The trial was designed so that 165 deaths were neededto provide a high probability of recording a numericallypositive treatment eff ect. We estimated that about222 patients were needed (148 patients in the sorafenibgroup and 74 patients in the placebo group) to record165 death events. Assuming a true underlying improvement in overall survival of 40% with 165 events (deaths),the probability of recording any numerically positivetreatment eff ect on survival was about 98·5%; theprobability of noting a greater than 20% improvementin overall survival with sorafenib compared with placebowas about 83·9%
đang được dịch, vui lòng đợi..