Returns to Medicare Spending: Evidence from Variation Across Physicians

描述（由申请人提供）：医疗支出的收益是否超过其成本是一个关键的政策问题，特别是考虑到政府在医疗保健领域的巨大和不断扩大的作用。本项目旨在发展一种衡量医疗支出边际收益的新方法，并研究其对几种不同类型医疗保健的适用性。医疗保健占支出的很大一部分，特别是老年人的医疗保健，由于其对医疗保险等公共政策的设计的重要性，因此特别令人感兴趣。研究医疗保健边际收益的新方法的基础是治疗强度变化的自然来源：医生之间治疗强度的广泛变化。在患者被准随机分配给医生的经验背景下——例如，当分配给患者的医生是由紧急情况下随叫随到的医生决定的——我们可以更好地理解医生之间这种差异的本质，并将其应用于工具变量框架中，以检查强化治疗对医疗保健支出和健康结果的影响。拟议研究的主要目的是确定可适用拟议战略的特定类型的医疗保健。首先，相关的经验背景必须有明确负责治疗决策的医生或医生团队，并且医生的（匿名）身份必须在索赔数据中可见。其次，医生之间的治疗强度必须有所不同。第三，将病人分配给医生必须是随机的。通过识别数据中昂贵和普遍的背景并与医生交谈，确定了一组候选的经验背景之后，关键的分析步骤将是使用医疗保险索赔数据来缩小一组经验背景，其中每个标准似乎都在数据中持有。理想情况下，这一过程将产生几个经验背景，其中将有可能估计医疗支出的边际回报，最终导致新的估计。

DESCRIPTION (provided by applicant): Whether the benefits of medical spending exceed their costs is a key policy question, especially given the government's large and expanding role in the health care sector. This project aims to develop a new approach for measuring the marginal returns to medical spending, and to investigate its applicability to several different types of medical care. Medical care that results in a large share of expenditures, especially among the elderly, is of particular interest because of its importance for the design of public policies such as Medicare. The basis for the new approach to investigating marginal returns to medical care is a natural source of variation in treatment intensity: the wide variation in treatment intensity across physicians. In empirical contexts in which patients are quasi-randomly assigned to doctors - such as when the doctors to which patients are assigned are determined by which physician is on-call at the time of an emergent condition - we can better understand the nature of this variation across physicians, and employ it in an instrumental variables framework to examine the impact of intensive treatments on health care spending and health outcomes. The main objective of the proposed research is to identify specific types of medical care to which the proposed strategy can be applied. First, relevant empirical contexts must have a doctor or team of doctors that is clearly responsible for treatment decisions, and the (anonymized) identity of the doctors must be visible in claims data. Second, there must be variation in treatment intensity across doctors. Third, patient assignment to doctors must be as good as random. After identifying a set of candidate empirical contexts by identifying costly and prevalent contexts in the data and speaking with physicians, the key analytical step will be to use the Medicare claims data to narrow in on a set of empirical contexts where each of these criteria appear to hold in the data. This process will ideally result in several empirical contextsin which it will be possible to estimate the marginal returns to medical spending, ultimately leading to new estimates. PUBLIC HEALTH RELEVANCE: This project will broaden the toolkit available to measure the marginal returns to medical spending, laying the groundwork for the development of new evidence on the value of health spending. By identifying the treatments that produce the largest health gains, society can allocate resources to improve health more efficiently, ultimately resulting in more health per dollar of health expenditure.