Division of Pharmacoepidemiology and Clinical Pharmacology, Utrecht Institute of Pharmaceutical Sciences, Utrecht University, Utrecht, The Netherlands.
Department of Healthcare Development, Stockholm County Council, Public Healthcare Services Committee, Stockholm, Sweden.
Pharmacoepidemiol Drug Saf. 2021 Jul;30(7):934-951. doi: 10.1002/pds.5232. Epub 2021 Mar 25.
Greedy caliper propensity score (PS) matching is dependent on randomness, which can ultimately affect causal estimates. We sought to investigate the variation introduced by this randomness.
Based on a literature search to define the simulation parameters, we simulated 36 cohorts of different sizes, treatment prevalence, outcome prevalence, treatment-outcome-association. We performed 1:1 caliper and nearest neighbor (NN) caliper PS-matching and repeated this 1000 times in the same cohort, before calculating the treatment-outcome association.
Repeating caliper and NN caliper matching in the same cohort yielded large variations in effect estimates, in all 36 scenarios, with both types of matching. The largest variation was found in smaller cohorts, where the odds ratio (OR) ranged from 0.53 to 10.00 (IQR of ORs: 1.11-1.67). The 95% confidence interval was not consistently overlapping a neutral association after repeating the matching with both algorithms. We confirmed these findings in a noninterventional example study.
Caliper PS-matching can yield highly variable estimates of the treatment-outcome association if the analysis is repeated.
贪婪卡尺倾向评分（PS）匹配依赖于随机性，这最终可能会影响因果估计。我们试图研究这种随机性带来的变化。
基于文献检索来定义模拟参数，我们模拟了 36 个不同大小、治疗流行率、结局流行率、治疗结局关联的队列。我们在同一个队列中进行了 1:1 卡尺和最近邻（NN）卡尺 PS 匹配，并重复了 1000 次，然后计算治疗结局关联。
在相同的队列中重复卡尺和 NN 卡尺匹配会导致效应估计值的大幅变化，在所有 36 种情况下，两种匹配类型都存在这种情况。在较小的队列中发现了最大的变化，其比值比（OR）范围为 0.53 至 10.00（OR 的 IQR：1.11-1.67）。在重复使用两种算法进行匹配后，95%置信区间并不总是与中性关联重叠。我们在一项非干预性的示例研究中证实了这些发现。
如果重复分析，卡尺 PS 匹配可能会产生高度可变的治疗结局关联估计值。
