摘要
需要治疗的病例数(NNT)是近年来国际上用于评价临床疗效的一个简单而有效的指标。运用鞍点逼近法构造了NNT的区间估计,并将其与Wald、Wilson score和Delta法进行比较。蒙特卡洛模拟研究结果表明,鞍点逼近法得到的区间估计覆盖率与名义水平接近程度总体上更高,平均区间长度更短,即鞍点逼近法优于Wald、Wilson score和Delta法。
The number needed to treat (NNT) is a simple and effective index for clinical therapeutic efficacy assessment internationally. In this paper, the saddle-point approximation method is applied to construct confidence interval for NNT. Monte Carlo simulation results show that coverage probability is higher and average interval length is shorter, compared with Wald method, Wilson score method and Delta method. So saddle-point approximation method is better than other methods.
出处
《统计与信息论坛》
CSSCI
2014年第9期11-14,共4页
Journal of Statistics and Information
基金
中国传媒大学理工科规划项目<一类非线性积分理论及其应用>(XNL1402)
北京高等学校"青年英才计划"项目(YETP0611)
中国传媒大学优秀中青年教师培养工程(YXJS201330)
关键词
需要治疗的病例数
置信区间
鞍点逼近法
the number needed to treat (NNT)
confidence interval
saddle-point approximation method