Toward Standardization and High Reliability: Improved Sepsis Screening in Emergency Department Triage Across an Academic Health System

Background

Sepsis is a life-threatening emergency, and early recognition and treatment in the emergency department (ED) is critical to improving outcomes.

Methods

The authors implemented an interdisciplinary quality improvement (QI) project to standardize sepsis screening workflow across an academic health system consisting of a large tertiary care urban hospital, one freestanding ED, and two small rural affiliate hospitals (RA-1 and RA-2). The research team used the Institute for Healthcare Improvement Model for Improvement framework, consisting of iterative Plan-Do-Study-Act (PDSA) cycles. The primary outcome was rates of screening for sepsis at each site. Secondary outcomes included sepsis mortality and Centers for Medicare & Medicaid Services (CMS) sepsis bundle (SEP-1) compliance at our main medical center. Primary outcome was assessed using electronic dashboards extracting the ratio of ED encounters with electronic health record (EHR)–documented sepsis screening per total ED encounters. The SEP-1 bundle was assessed as percent compliance, and mortality was calculated as average observed to expected (O:E). Averages were compared from preintervention to after initiating improvements using two-tailed t-tests.

Results

This QI project took place from December 2022 to December 2023 across four EDs that experience around 138,000 visits annually. A standardized workflow was established at ED triage with an EHR–based question and an associated nurse and physician defined response. Preintervention (October 2022 to November 2022) triage rates for sepsis were 1.7% (163/9,560), 25.3% (523/2,068), 11.0% (360/3,272), and 36.5% (915/2,506) at our main hospital, freestanding ED, RA-1, and RA-2, respectively. After four PDSA cycles, triage rates rose to 91.9% (4,927/5,360), 97.5% (1,032/1,059), 99.0% (1,845/1,863), and 97.4% (1,328/1,363), respectively (p < 0.005). Sepsis triage rates rose most slowly at the large academic medical center, where progressive PDSA cycles were needed to achieve > 90% screening for sepsis. Mean O:E mortality was 0.99 for the 9 months of available data preintervention and 0.83 in the 17 months postintervention (p = 0.07). CMS sepsis bundle compliance was 28.4% for the 15 months preintervention and 40.5% in the 17 months postintervention, (p = 0.14).

Conclusion

An interdisciplinary QI project leveraged EHR optimization to integrate with human workflows over four PDSA cycles to achieve standardized and improved screening for sepsis in the ED. This resulted in lower sepsis mortality and increased sepsis bundle compliance, though results were not statistically significant.