College of Medicine and Division of Health Care Policy and Research, Mayo Clinic, Rochester, MN 55905, USA.
Health Econ. 2011 Sep;20(9):1025-42. doi: 10.1002/hec.1757. Epub 2011 Jun 1.
High deductible health plans (HDHPs) have become an increasingly common form of benefit design used by employers to manage health-care costs. Numerous studies have evaluated the uptake and impact of HDHPs on health-care utilization. Most studies have employed the standard difference-in-differences (DID) methodology. In this paper, we employ three alternative methodologies to evaluate a natural experiment in which a traditional health plan was fully replaced by a HDHP. We implement the standard DID and the quantile difference-in-differences (QDID) estimators to evaluate the impact of the HDHP on following six outcomes: overall cost, medical cost, pharmacy cost, outpatient visit count, inpatient visit count and emergency room visit count. We compare these results to a changes-in-changes (CIC) estimator, a generalized version of the standard (DID) estimator. We find that both the DID and CIC models yielded similar results, while the QDID model provided additional insights on the HDHP impact across different parts of the outcome distributions. Overall, introduction of HDHP had no impact on health-care costs, positive impact on the number of outpatient visits and mixed impacts on the inpatient and emergency room visit counts. The QDID estimates suggest HDHP introduction generally impacted subjects in upper percentiles (50th, 75th and 90th).
高免赔额健康计划 (HDHPs) 已成为雇主用来管理医疗保健成本的一种越来越常见的福利设计形式。许多研究已经评估了 HDHPs 对医疗保健利用的采用和影响。大多数研究都采用了标准的差异（DID）方法。在本文中，我们采用了三种替代方法来评估一项自然实验，其中一项传统健康计划被 HDHP 完全取代。我们实施了标准 DID 和分位数差异（QDID）估计器来评估 HDHP 对以下六个结果的影响：总费用、医疗费用、药房费用、门诊就诊次数、住院就诊次数和急诊就诊次数。我们将这些结果与变化中的变化（CIC）估计器进行比较，这是标准（DID）估计器的广义版本。我们发现 DID 和 CIC 模型都产生了相似的结果，而 QDID 模型则提供了有关 HDHP 在不同结果分布部分的影响的更多见解。总体而言，HDHP 的引入对医疗保健费用没有影响，对门诊就诊次数有积极影响，对住院和急诊就诊次数的影响则较为复杂。QDID 估计表明，HDHP 的引入通常会影响处于较高百分位（第 50、75 和 90 百分位）的受试者。
