基于赢率分析临床复合终点的统计方法

A Win Ratio Approach to Comparing Composite Outcomes in Clinical Trials

目的当复合终点(事件)中组分结局事件的临床优先级问题存在时,传统基于风险比的方法常出现一定的估计误差。本文以急性淋巴细胞白血病患者预后研究为背景,讨论能解决该问题的统计方法。方法介绍在组分事件之间的优先级非常重要时,一种新的指标赢率的概念和分析方法;随后将该方法应用于急性淋巴细胞白血病的预后研究,比较风险比和基于赢率的匹配、不匹配和加权赢率法的差异。结果研究发现,赢率法显示GvHD预防是急性淋巴细胞白血病患者预后的危害因素,而风险比的结果与之相反,这是由于在含有多个优先级结局事件的研究中,风险比受第一结局事件的影响,结果往往存在一定的偏差,而赢率能更好的反映实际情况。结论当组分事件之间的优先级很重要时,相对于风险比,赢率是一个更好的统计评价指标。

Objective When the clinical priority of components in the composite endpoints exists,the hazard ratio often has some estimation errors.Based on the prognosis of patients with acute lymphoblastic leukemia,this study discusses the statistical methods that can solve this problem.Methods Firstly,we introduced the win ratio a new approach to the analysis of composite endpoints when the priority between components is important;this method was subsequently applied to the prognosis study of acute lymphoblastic leukemia,comparing the differences between the hazard ratio,the matched win ratio,the unmatched win ratio and the weighted win ratio.Results The win ratio showed that GvHD prevention was a risk factor for prognosis in patients with acute lymphoblastic leukemia,while the hazard ratio was the opposite.This is because in the study with multiple endpoints,the hazard ratio is influenced by the first event,which leads to deviations.However,the win ratio is not affected.Conclusion The win ratio is better statistical measure than the hazard ratio when the clinical priorities among components matters.