Objective: To identify characteristics associated with high- and low-quality multi-hospital systems for major cancer surgery Background: Although multi-hospital health systems provide most inpatient healthcare in the US, our understanding of how these systems can optimize surgical quality among their hospitals remains limited. Identifying the structural characteristics (e.g., number of hospitals, procedural volume, geographic dispersion) that distinguish high- and low-quality systems may inform actionable strategies to improve surgical quality. Methods: We conducted a retrospective cross-sectional observational study of 270,491 Medicare beneficiaries (2016-2020) undergoing major cancer surgery at a multi-hospital health system. Systems were classified into quartiles of quality based on risk- and reliability-adjusted rates of 30-day mortality using a hierarchical multivariable logistical regression model to adjust for patient, procedural, and hospital factors. Results: The adjusted 30-day operative mortality rate in the highest- versus lowest-quality quartile of systems was 1.7% versus 3.1%,(P<0.001). High-quality systems had fewer hospitals per system (median [IQR], number of system hospitals, 5 [3-11] vs. 12 [8-30];P<0.001), with each performing more procedures per hospital (median [IQR] annual procedure volume, 104 [52-218] vs. 45 [22-90];P<0.001). High-quality systems were also more geographically concentrated (median [IQR] maximum distance between hospitals, 62 [19-194] vs. 321 [91-1125] miles;P<0.001). Furthermore, high-quality systems demonstrated less variation in quality between hospitals (mean [SD] within-system absolute variation in mortality, 0.8% [0.3%] vs. 2.6% [1.0%];P<0.001). Conclusions: The highest-quality multi-hospital systems had fewer, more geographically concentrated hospitals, with each performing more procedures per hospital. Among the highest-quality systems, diverse system phenotypes were represented, suggesting the potential to overcome structural limitations and achieve high quality.
Read full abstract