Varying Estimates of Sepsis among Adults Presenting to US Emergency Departments: Estimates from a National Dataset from 2002-2018

Background: A variety of approaches to defining sepsis using administrative datasets have been previously reported. We aimed to compare estimates, demographics, treatment factors, outcomes and longitudinal trends of patients identified with sepsis in United States emergency departments (EDs) using differing sets of sepsis criteria. Methods: We performed a cross-sectional study using the National Healthcare Ambulatory Medical Care Survey, a complex survey of nonfederal US ED encounters between 2002 to 2018. We obtained survey-weighted population-adjusted encounters of sepsis using the following criteria: explicit sepsis, severe sepsis, and quick Sequential Organ Failure Assessment (qSOFA) score combined with the presence of infection. Results: Age-adjusted for US adults, 18.6, 16.1 and 8.9 encounters per 10 000 population were identified when using the explicit, severe sepsis and qSOFA definitions, respectively. A higher proportion of the explicit cohort was hospitalized and had blood cultures performed, compared to cohorts ascertained using severe sepsis and qSOFA criteria, though confidence intervals overlapped. Antibiotic use was highest in encounters meeting qSOFA criteria. When inspecting unweighted encounters meeting each set of criteria, there was minimal overlap, with only 3% meeting all three. Encounters meeting the explicit and severe sepsis criteria were increasing over time. Conclusion: The explicit, severe sepsis and qSOFA criteria generated similar annual rates of presentation when applied to US ED encounters, with some evidence of the explicit sepsis cohort being higher acuity. There was minimal overlap of cases and instability in estimates when assessed longitudinally. Our findings inform research efforts to accurately identify sepsis among ED encounters using administrative data.