Relationship Between a Sepsis Intervention Bundle and In-Hospital Mortality Among Hospitalized Patients: A Retrospective Analysis of Real-World Data.

ABSTRACT

BACKGROUND: Sepsis is a systemic response to infection that can lead to tissue damage, organ failure, and death. Efforts have been made to develop evidence-based intervention bundles to identify and manage sepsis early in the course of the disease to decrease sepsis-related morbidity and mortality. We evaluated the relationship between a minimally invasive sepsis intervention bundle and in-hospital mortality using robust methods for observational data. METHODS: We performed a retrospective cohort study at the University of California, San Francisco, Medical Center among adult patients discharged between January 1, 2012, and December 31, 2014, and who received a diagnosis of severe sepsis/septic shock (SS/SS). Sepsis intervention bundle elements included measurement of blood lactate; drawing of blood cultures before starting antibiotics; initiation of broad spectrum antibiotics within 3 hours of sepsis presentation in the emergency department or 1 hour of presentation on an inpatient unit; administration of intravenous fluid bolus if the patient was hypotensive or had a lactate level >4 mmol/L; and starting intravenous vasopressors if the patient remained hypotensive after fluid bolus administration. Poisson regression for a binary outcome variable was used to estimate an adjusted incidence-rate ratio (IRR) comparing mortality in groups defined by bundle compliance measured as a binary predictor, and to estimate an adjusted number needed to treat (NNT). RESULTS: Complete bundle compliance was associated with a 31% lower risk of mortality (adjusted IRR, 0.69, 95% confidence interval [CI], 0.53-0.91), adjusting for SS/SS presentation in the emergency department, SS/SS present on admission (POA), age, admission severity of illness and risk of mortality, Medicaid/Medicare payor status, immunocompromised host status, and congestive heart failure POA. The adjusted NNT to save one life was 15 (CI, 8-69). Other factors independently associated with mortality included SS/SS POA (adjusted IRR, 0.55; CI, 0.32-0.92) and increased age (adjusted IRR, 1.13 per 10-year increase in age; CI, 1.03-1.24). CONCLUSIONS: The University of California, San Francisco, sepsis bundle was associated with a decreased risk of in-hospital mortality across hospital units after robust control for confounders and risk adjustment. The adjusted NNT provides a reasonable and achievable goal to observe measureable improvements in outcomes for patients diagnosed with SS/SS.