Public Insurance Design and Health at Older Ages

项目摘要 在过去的十年里，医疗保险一直处于美国公共政策辩论的前沿。为老年人服务 受益于医疗保险几乎全民覆盖的美国人，关于他们的范围的决定 自1965年联邦医疗保险成立以来，保险覆盖面一直是政策关注的核心问题。医疗保险的承保范围 是广泛的，但根据法律，该计划只涵盖“合理和必要的”医疗服务，这 这可能是一个有争议的定义。一些关于联邦医疗保险覆盖范围的决定引发了激烈的争论 辩论，强调了该计划的覆盖决定对数百万患者和医生的重要性。 未来几十年，围绕医疗保险覆盖范围的辩论可能会加剧，因为 医疗和检测迅速扩大，人口老龄化和公共保险制度难以应对 如何解决获得医疗创新方面的不平等问题。然而，尽管关于 医疗保险内外的保险覆盖范围，我们对覆盖范围(如 与患者分担费用相反)，影响老年人的治疗选择、临床实践和健康结果。 这项提案中概述的研究旨在开始填补这一空白。该项目调查了存在或 缺乏对特定程序或药物治疗的保险覆盖会影响治疗决定和 阿尔茨海默病和相关痴呆症(ADRD)老年人的健康结局。这是一个人口 这可能特别容易受到保险覆盖范围的变化和限制的影响，因为患者可能 决策能力，并可能不成比例地接受被认为是实验性的治疗 缺乏有效性，无法清除“合理和必要”的门槛。该项目的目标1是估计 处方药和门诊程序的覆盖决定对治疗的平均影响 老年ADRD老年医疗保险参与者的决策和健康结果。目标2是预测和描述 ADRD患者中最有可能受到限制范围的决定影响的亚群 使用机器学习方法的保险覆盖。所提出的经验方法是使用准- 由保险范围突然变化的自然实验引起的实验变化 医疗保险计划的不同部分。分析利用了公式范围的变化。 跨Medicare Part D计划，以及医生服务和 医疗保险B部分下的门诊程序这些变异来源加上准- 用于异质性分析的处理效果和机器学习方法的实验估计，允许 评估不同药物对老年ADRD患者治疗决策和健康的影响 程序和ADRD患者的亚群。

PROJECT ABSTRACT Health insurance has been at the forefront of US public policy debate throughout the last decade. For elderly Americans, who benefit from nearly universal coverage under Medicare, decisions about the scope of their insurance coverage have been a central policy concern since Medicare’s inception in 1965. Medicare’s coverage is extensive, but by statute the program only covers medical services that are “reasonable and necessary,” which can be a controversial definition. Some decisions about what Medicare should cover have engendered intense debate, underscoring the importance of the program’s coverage decisions for millions of patients and doctors. The debate around the scope of Medicare coverage is likely to intensify in the coming decades as options for medical treatment, and testing, expand rapidly, the population ages, and public insurance systems grapple with how to address inequality in access to medical innovations. Yet, despite the importance of decisions about the scope of insurance coverage in and outside of Medicare, we know little about how the scope of coverage (as opposed to patient cost-sharing), affects treatment choices, clinical practice, and health outcomes for the elderly. The research outlined in this proposal aims to start filling this gap. The project investigates how the presence or lack of insurance coverage for specific procedural or pharmacologic therapies affects treatment decisions and health outcomes for the elderly with Alzheimer’s Disease and related Dementias (ADRD). This is a population that may be particularly vulnerable to changes and limits in insurance coverage, as the patients may have limited decision-making capacity and may be disproportionately exposed to treatments that are deemed experimental and lacking effectiveness to clear the “reasonable and necessary” threshold. Aim 1 of the project is to estimate the average effect of coverage decisions across prescription drugs and outpatient procedures on treatment decisions and health outcomes of elderly Medicare enrollees with ADRD. Aim 2 is to predict and characterize the subgroups of ADRD patients that are most likely to be affected by decisions that restrict the scope of insurance coverage using machine learning methods. The proposed empirical method is to use quasi- experimental variation that arises from natural experiments of abrupt changes in insurance coverage within different parts of the Medicare program. The analysis takes advantage of variation in the scope of formularies across Medicare Part D plans, as well as the variation in local coverage decisions for physician services and outpatient procedures under Medicare Part B. These sources of variation coupled with methods for quasi- experimental estimation of treatment effects and machine learning methods for heterogeneity analyses, allow estimating the response of treatment decisions and health of the elderly with ADRD across different drugs, procedures, and subgroups of ADRD patients.