Association of Academic Medical Center Presence With Clinical Outcomes at Neighboring Community Hospitals Among Medicare Beneficiaries

• Published in: JAMA network open Vol. 6; no. 2; p. e2254559
• Main Authors: Burke, Laura G, Burke, Ryan C, Orav, E John, Duggan, Ciara E, Figueroa, Jose F, Jha, Ashish K
• Format: Journal Article
• Language: English
• Published: United States American Medical Association 01.02.2023
• Subjects: Academic Medical Centers; Aged; Cohort Studies; Female; Health Policy; Hospitals, Community; Humans; Medicare; Online Only; Original Investigation; Retrospective Studies; United States - epidemiology; United States
• ISSN: 2574-3805; 2574-3805
• DOI: 10.1001/jamanetworkopen.2022.54559

Studies suggest that academic medical centers (AMCs) have better outcomes than nonteaching hospitals. However, whether AMCs have spillover benefits for patients treated at neighboring community hospitals is unknown. To examine whether market-level AMC presence is associated with outcomes for patients treated at nonteaching hospitals within the same markets. This retrospective, population-based cohort study assessed traditional Medicare beneficiaries aged 65 years and older discharged from US acute care hospitals between 2015 and 2017 (100% sample). Data were analyzed from August 2021 to December 2022. The primary exposure was market-level AMC presence. Health care markets (ie, hospital referral regions) were categorized by AMC presence (percentage of hospitalizations at AMCs) as follows: no presence (0%), low presence (>0% to 20%), moderate presence (>20% to 35%), and high presence (>35%). The primary outcomes were 30-day and 90-day mortality and healthy days at home (HDAH), a composite outcome reflecting mortality and time spent in facility-based health care settings. There were 22 509 824 total hospitalizations, with 18 865 229 (83.8%) at non-AMCs. The median (IQR) age of patients was 78 (71-85) years, and 12 568 230 hospitalizations (55.8%) were among women. Of 306 hospital referral regions, 191 (62.4%) had no AMCs, 61 (19.9%) had 1 AMC, and 55 (17.6%) had 2 or more AMCs. Markets characteristics differed significantly by category of AMC presence, including mean population, median income, proportion of White residents, and physicians per population. Compared with markets with no AMC presence, receiving care at a non-AMC in a market with greater AMC presence was associated with lower 30-day mortality (9.5% vs 10.1%; absolute difference, -0.7%; 95% CI, -1.0% to -0.4%; P < .001) and 90-day mortality (16.1% vs 16.9%; absolute difference, -0.8%; 95% CI, -1.2% to -0.4%; P < .001) and more HDAH at 30 days (16.49 vs 16.12 HDAH; absolute difference, 0.38 HDAH; 95% CI, 0.11 to 0.64 HDAH; P = .005) and 90 days (61.08 vs 59.83 HDAH; absolute difference, 1.25 HDAH; 95% CI, 0.58 to 1.92 HDAH; P < .001), after adjustment. There was no association between market-level AMC presence and mortality for patients treated at AMCs themselves. AMCs may have spillover effects on outcomes for patients treated at non-AMCs, suggesting that they have a broader impact than is traditionally recognized. These associations are greatest in markets with the highest AMC presence and persist to 90 days.