Department of Healthcare Quality Assessment, Graduate School of Medicine, University of Tokyo, Tokyo, Japan.
Health Economics and Outcomes Research, Regeneron Pharmaceuticals, Inc, Tarrytown, New York, United States of America.
PLoS One. 2022 Aug 15;17(8):e0272975. doi: 10.1371/journal.pone.0272975. eCollection 2022.
Confounding by indication is a serious threat to comparative studies using real world data. We assessed the utility of automated data-adaptive analytic approach for confounding adjustment when both claims and clinical registry data are available.
We used a comparative study example of carotid artery stenting (CAS) vs. carotid endarterectomy (CEA) in 2005-2008 when CAS was only indicated for patients with high surgical risk. We included Medicare beneficiaries linked to the Society for Vascular Surgery's Vascular Registry >65 years old undergoing CAS/CEA. We compared hazard ratios (HRs) for death while adjusting for confounding by combining various 1) Propensity score (PS) modeling strategies (investigator-specified [IS-PS] vs. automated data-adaptive [ada-PS]); 2) data sources (claims-only, registry-only and claims-plus-registry); and 3) PS adjustment approaches (matching vs. quintiles-adjustment with/without trimming). An HR of 1.0 was used as a benchmark effect estimate based on CREST trial.
The cohort included 1,999 CAS and 3,255 CEA patients (mean age 76). CAS patients were more likely symptomatic and at high surgical risk, and experienced higher mortality (crude HR = 1.82 for CAS vs. CEA). HRs from PS-quintile adjustment without trimming were 1.48 and 1.52 for claims-only IS-PS and ada-PS, 1.51 and 1.42 for registry-only IS-PS and ada-PS, and 1.34 and 1.23 for claims-plus-registry IS-PS and ada-PS, respectively. Estimates from other PS adjustment approaches showed similar patterns.
In a comparative effectiveness study of CAS vs. CEA with strong confounding by indication, ada-PS performed better than IS-PS in general, but both claims and registry data were needed to adequately control for bias.
在使用真实世界数据进行比较研究时，混杂因素是一个严重的威胁。我们评估了在既有索赔数据又有临床注册数据的情况下，自动化数据自适应分析方法在混杂因素调整方面的效用。
我们使用了一个 2005-2008 年颈动脉支架置入术（CAS）与颈动脉内膜切除术（CEA）的比较研究示例，当时 CAS 仅适用于高手术风险的患者。我们纳入了 Medicare 受益人与血管外科学会的血管注册中心相链接的年龄在 65 岁以上的行 CAS/CEA 的患者。我们比较了调整混杂因素后的死亡风险比（HR），方法是结合以下各项对各种 1）倾向评分（PS）建模策略（研究者指定[IS-PS]与自动化数据自适应[ada-PS]）；2）数据来源（仅索赔数据、仅注册数据以及索赔加注册数据）；3）PS 调整方法（匹配与五分位数调整，有无修剪）进行组合。根据 CREST 试验，使用 1.0 的 HR 作为基准效应估计值。
该队列包括 1999 例 CAS 和 3255 例 CEA 患者（平均年龄 76 岁）。CAS 患者更有可能有症状和高手术风险，并且死亡率更高（未经调整的 CAS 患者的粗 HR 为 1.82 与 CEA 相比）。未经修剪的 PS 五分位数调整的 HR 分别为仅索赔 IS-PS 和 ada-PS 的 1.48 和 1.52，仅注册 IS-PS 和 ada-PS 的 1.51 和 1.42，以及索赔加注册 IS-PS 和 ada-PS 的 1.34 和 1.23。其他 PS 调整方法的估计值表现出类似的模式。
在 CAS 与 CEA 的比较有效性研究中，混杂因素严重，ADA-PS 总体上优于 IS-PS，但需要同时使用索赔数据和注册数据才能充分控制偏倚。
