The use of quantile regression in health care research: a case study examining gender differences in the timeliness of thrombolytic therapy
The use of quantile regression in health care research: a case study examining gender differences in the timeliness of thrombolytic therapy
复制标题
DOI：
10.1002/sim.1851
复制
发表时间：
2005-03-15
影响因子：
2
通讯作者：
Alter, DA
中科院分区：
文献类型：
--
作者：
Austin, PC;Tu, JV;Alter, DA
调查人员经常感兴趣的是确定与提供基本医疗延误相关的患者和系统特征。研究者通常使用多元线性回归或考克斯比例风险模型来评估患者和系统特征对医疗及时性的影响。使用这两种方法的缺点是，它们最多只能部分探索治疗延迟或等待时间的分布如何随患者特征而变化。相比之下，分位数回归模型允许人们评估条件分布的任何分位数如何随患者特征而变化。我们通过检查急性心肌梗死患者溶栓治疗的性别差异来说明分位数回归的实用性。我们证明，更丰富的推论，可以通过使用分位数回归。与配偶相比，雌性动物更有可能出现治疗延迟。此外，性别对治疗延迟最严重的患者影响更大。想要确定治疗延迟或等待时间的分布如何随患者或系统特征而变化的研究者应考虑使用分位数回归补充其分析。版权所有（C）2004约翰威利父子有限公司。
Investigators are frequently interested in determining patient and system characteristics associated with delays in the provision of essential medical treatment. Investigators have typically used either multiple linear regression or Cox proportional hazards models to assess the impact of patient and system characteristics on the timeliness of medical treatment. A drawback to the use of these two methods is that they allow, at best, a partial exploration of how a distribution of delays in treatment or of waiting times changes with patient characteristics. In contrast, quantile regression models allow one to assess how any quantile of a conditional distribution changes with patient characteristics. We illustrate the utility of quantile regression by examining gender differences in the delivery of thrombolysis in patients with an acute myocardial infarction. We demonstrate that richer inferences can be drawn through the use of quantile regression. Females were more likely to experience delays in treatment compared to mates. Furthermore, gender had a greater impact upon those patients who had the greatest delays in treatment. Investigators who want to determine how a distribution of delays in treatment or of waiting times changes with patient or system characteristics should consider complementing their analyses with the use of quantile regression. Copyright (C) 2004 John Wiley Sons, Ltd.