离家更近：改善医疗服务可及性的结构式“先估计、后优化”方法

中文摘要

医疗服务可及性的地理不平等不仅限于城乡差距，还包括社会经济、种族及其他方面的差异。医院、诊所、医疗服务提供者和药店的邻近程度差异很大，给战略性设施选址带来了挑战。各项服务的需求取决于当地人口健康状况、个人偏好、服务提供能力及其他因素。本研究提出一种新颖的结构式先估计后优化（SETO）框架，将采用改进的 Berry–Levinsohn–Pakes 方法并考虑服务提供能力的结构化需求估计，与基于选择的最优设施选址模型相结合，以最大化医疗服务利用率。我们以加利福尼亚州联邦零售药店计划为案例说明该方法，该计划是一项已实施数百万剂新冠疫苗接种的公私合作项目。需求估计表明，社会经济脆弱社区的居民对前往药店疫苗接种点的出行距离更为敏感。战略性地增设500家服务低收入社区的零售药店，可使预测接种量总体提高2.9%（全州新增77万剂次），在健康状况最差的街区提高5.3%。我们的综合性 SETO 方法优于基于当前接种率、现有服务缺口、人口密度或预测需求来配置资源的启发式方法。该案例研究证明了以下两点的重要性：（1）在估计需求时考虑异质性；（2）通过具有空间异质性的供给选择合作伙伴，以补充现有网络并高效填补服务缺口。本研究提供了一种优化医疗服务提供网络的系统方法，既利用公开可得的总计数据，又考虑个人偏好，凸显了将结构化需求模型与规范性分析相结合的价值。本文由医疗健康管理编辑 Stefan Sholtes 接收。补充材料：在线附录与数据文件可从 https://doi.org/10.1287/mnsc.2024.06274 获取。

Abstract

Geographic inequalities in healthcare access extend beyond rural–urban divides to include socioeconomic, racial, and other disparities. Proximity to hospitals, clinics, healthcare providers, and pharmacies varies widely, posing a challenge in deciding where to strategically locate such facilities. Demand for each service depends on local population health, individual preferences, provider capacity, and other factors. This study introduces a novel structural estimate-then-optimize (SETO) framework, combining structural demand estimation using a modified Berry–Levinsohn–Pakes approach that accounts for provider capacity with a choice-based optimal facility location model to maximize health service utilization. Our methodology is illustrated with a case study on the Federal Retail Pharmacy Program in California, a public–private partnership that administered millions of COVID-19 vaccinations. Demand estimates indicate that residents of socioeconomically vulnerable communities are more sensitive to travel distances to pharmacy-based vaccination sites. Strategically adding 500 retail stores serving lower-income communities increases predicted vaccinations by 2.9% overall (770,000 additional vaccinations statewide) and by 5.3% in the least healthy neighborhoods. Our integrative SETO approach outperforms heuristics that allocate resources based on current vaccination rates, existing service gaps, population density, or predicted demand. The case study demonstrates the importance of (1) accounting for heterogeneity in estimating demand and (2) selecting partnerships to complement existing networks with spatially heterogeneous supply and efficiently fill service gaps. Our study provides a systematic approach to optimize healthcare delivery networks, using publicly available aggregate data while accounting for individuals’ preferences, highlighting the value of combining a structural demand model with prescriptive analytics. This paper was accepted by Stefan Sholtes, healthcare management. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2024.06274 .