Problem: Hospital-onset sepsis is understudied and was not considered in the development and implementation of Centers for Medicare and Medicaid Services (CMMS) core quality measure related to sepsis, SEP-1.

Methods: Samples of patients with sepsis were identified by diagnosis codes from the electronic health records of 4 university hospitals and publicly available administrative data from the California Department of Public Health. Multilevel models with random effects were used to evaluate for an association between risk factors and outcomes of interest. Endogenous treatment effects models were used to estimate treatment effects from observational data.

Findings: Providers are about 3-times less likely to deliver SEP-1-adherent care for patients with hospital-onset sepsis than patients with community-onset sepsis. The care bundle outlined in SEP-1 was not associated with a treatment benefit (reduced mortality, decreased requirement for vasopressors) in the total sample of eligible patients, in the cohort with community-onset sepsis, or in the cohort with hospital-onset sepsis. However, multiple components of the bundle appeared to improve outcomes in patients with community-onset sepsis. Only the use of antibiotics was associated with reduced mortality in patients with hospital-onset sepsis.

Meaning: Hospital-onset and community-onset sepsis are distinct clinical entities—they are managed differently by providers and respond differently to SEP-1-adherent care. Low levels of adherence to the SEP-1 core quality measure for patients with hospital-onset sepsis may not represent a quality gap. Rather, consideration should be given to excluding patients with hospital-onset sepsis from the core quality measure.