Ithan Peltan, MD, MSc
Assistant Professor
Intermountain Medical Center
Salt Lake City, UT
Disclosure information not submitted.
Sierra McLean, MD
Resident physician
University of North Carolina, United States
Disclosure information not submitted.
Emily Murnin, MD
Resident physician
University of Wisconsin, United States
Disclosure information not submitted.
Edward Stenehjem, MD
Associate Professor
Intermountain Healthcare, United States
Disclosure information not submitted.
Brandon Webb, MD
Associate Professor
Intermountain Healthcare/Stanford University, United States
Disclosure information not submitted.
Joseph Bledsoe, MD
Associate Professor
Intermountain Healthcare/Stanford University, United States
Disclosure information not submitted.
Samuel Brown, MD, MS
Associate Professor
Center for Humanizing Critical Care, Intermountain Healthcare; Pulmonary Division, Department of Medicine, Intermountain Medical Center; Pulmonary Division, Department of Medicine, University of Utah School of Medicine, United States
Disclosure information not submitted.
Catherine Hough, MD, MS
Dr
Oregon Health and Science University, United States
Disclosure information not submitted.
Title: Agreement of Initial and Final Infection Diagnosis Among ED Patients Meeting Sepsis Criteria
Introduction: Efforts to improve emergency department (ED) sepsis care may also drive misdiagnosis and antibiotic overtreatment of patients with sepsis “mimics.” Because prior studies were generally small and predated current sepsis consensus criteria, we investigated the accuracy of ED-based infection identification in a large, contemporary, multihospital Sepsis-3 cohort.
Methods: We performed a secondary analysis of a retrospective cohort of adult, non-trauma patients who met Sepsis-3 criteria (acute SOFA score increase ≥2 points, IV antibiotics, and blood culture) while in the ED and were admitted to four hospitals in Utah from 2013-2017. Trained abstractors employed standardized methods to identify the infection source diagnosed in the ED and adjudicated a final infection status/source using all data from the hospitalization.
Results: Among 8,267 eligible patients, the 691 (8%) patients with infection ruled out during their hospital stay had slightly higher Charlson Comorbidity Indices (median 4 [IQR 2-7] vs 3 [IQR 1-7]) and were slightly more likely to arrive via ambulance (38% vs 31%) but were otherwise demographically and clinically similar to patients with confirmed or possible infection on final assessment. Mortality was similar between non-infected vs infected patients (10.4% vs 8.8%, p=0.17) but hospital stays were shorter (101±92 vs 120±134 hours, p< 0.001). The 794 (10%) patients whose ED source of infection was “unknown” were more likely to have infection ruled out on final assessment (29%) compared to those with a specific infection source identified by the ED physician (6%). Among 75 physicians who saw at least 40 eligible patients, the percentage of patients who ultimately had infection ruled out ranged from 1% to 16%. ED physician remained a significant predictor of subsequent infection rule out after adjustment for case mix (p=0.01).
Conclusions: Eight percent of patients who met Sepsis-3 criteria and received IV antibiotics in the ED had infection ruled out prior to hospital discharge. This figure is lower than in prior studies. However, our findings that ruled-out infection was more common if no specific source was identified in the ED and infection overdiagnosis varied substantially between physicians still suggest opportunities to reduce misdiagnosis and unnecessary antibiotic treatment.