Abstract
RATIONALE: Processes of care are potential determinants of outcomes in patients with severe sepsis. Whether hospitals with more experience caring for patients with severe sepsis also have improved outcomes is unclear. OBJECTIVES: To determine associations between hospital severe sepsis caseload and outcomes. METHODS: We analyzed data from U.S. academic hospitals provided through University HealthSystem Consortium. We used University HealthSystem Consortium's sepsis mortality model (c-statistic, 0.826) for risk adjustment. Validated International Classification of Disease, 9th Edition, Clinical Modification algorithms were used to identify hospital severe sepsis case volume. Associations between risk-adjusted severe sepsis case volume and mortality, length of stay, and costs were analyzed using spline regression and analysis of covariance. MEASUREMENTS AND MAIN RESULTS: We identified 56,997 patients with severe sepsis admitted to 124 U.S. academic hospitals during 2011. Hospitals admitted 460 ± 216 patients with severe sepsis, with median length of stay 12.5 days (interquartile range, 11.1-14.2), median direct costs $26,304 (interquartile range, $21,900-$32,090), and average hospital mortality 25.6 ± 5.3%. Higher severe sepsis case volume was associated with lower unadjusted severe sepsis mortality (R2 = 0.10, P = 0.01) and risk-adjusted severe sepsis mortality (R2 = 0.21, P < 0.001). After further adjustment for geographic region, number of beds, and long-term acute care referrals, hospitals in the highest severe sepsis case volume quartile had an absolute 7% (95% confidence interval, 2.4-11.6%) lower hospital mortality than hospitals in the lowest quartile. We did not identify associations between case volume and resource use. CONCLUSIONS: Academic hospitals with higher severe sepsis case volume have lower severe sepsis hospital mortality without higher costs.