Some Methods of Propensity-Score Matching had Superior Performance to Others: Results of an Empirical Investigation and Monte Carlo simulations
Some Methods of Propensity-Score Matching had Superior Performance to Others: Results of an Empirical Investigation and Monte Carlo simulations
复制标题
DOI：
10.1002/bimj.200810488
复制
发表时间：
2009-02-01
影响因子：
1.7
通讯作者：
Austin, Peter C.
中科院分区：
文献类型：
--
作者：
Austin, Peter C.
在使用观察性数据估计因果治疗效果时，倾向性评分匹配越来越多地用于减少治疗选择偏差的影响。目前在医学文献中采用了几种倾向分数匹配方法：使用倾向分数的logit的标准差的0.2或0.6的宽度的卡尺匹配倾向分数的logit;使用0.005、0.01、0.02、0.03和0.1的卡尺匹配倾向分数;以及倾向分数上的5 -> 1位数匹配。我们进行了实证调查和蒙特卡罗模拟，调查这些竞争方法的相对性能。使用大量因心脏病发作住院且在出院时接受他汀类药物处方的患者样本，我们发现8种不同方法产生的倾向评分匹配样本中，治疗和未治疗受试者之间在测量的基线变量方面实现了定性等效平衡。8个倾向评分匹配的样本中有7个导致他汀类药物暴露导致死亡率降低的定性相似估计值。与其他7种方法相比，5 I数字匹配导致相对风险降低的定性不同估计。使用Monte Carlo模拟，我们发现，使用倾向评分logit标准差的宽度为0.2的卡尺进行匹配，以及使用宽度为0.02和0.03的卡尺进行匹配，在估计治疗效果方面往往具有上级性能。
Propensity-score matching is increasingly being used to reduce the impact of treatment-selection bias when estimating causal treatment effects using observational data. Several propensity-score matching methods are currently employed in the medical literature: matching on the logit of the propensity score using calipers of width either 0.2 or 0.6 of the standard deviation of the logit of the propensity score; matching on the propensity score using calipers of 0.005, 0.01, 0.02, 0.03, and 0.1; and 5 -> 1 digit matching on the propensity score. We conducted empirical investigations and Monte Carlo simulations to investigate the relative performance of these competing methods. Using a large sample of patients hospitalized with a heart attack and with exposure being receipt of a statin prescription at hospital discharge, we found that the 8 different methods produced propensity-score matched samples in which qualitatively equivalent balance in measured baseline variables was achieved between treated and untreated subjects. Seven of the 8 propensity-score matched samples resulted in qualitatively similar estimates of the reduction in mortality due to statin exposure. 5 I digit matching resulted in a qualitatively different estimate of relative risk reduction compared to the other 7 methods. Using Monte Carlo simulations, we found that matching using calipers of width of 0.2 of the standard deviation of the logit of the propensity score and the use of calipers of width 0.02 and 0.03 tended to have superior performance for estimating treatment effects.