370 Accuracy of Clinical Assessment in Predicting Source of Infection for Septic Patients in the Emergency Department

Document Type

Conference Proceeding

Publication Date


Publication Title

Ann Emerg Med


Background: Many sepsis cases are first encountered in the Emergency Department (ED), and it is essential to identify and assess the severity of patient illnesses as well as their mortality risks as soon as possible after they present to the ED. The Sepsis Core Measure requires that clinicians rapidly screen patients and administer antimicrobial treatment within 3 hours of identification of severe sepsis and septic shock. Although much research has been done on the importance of early fluid administration, antimicrobial initiation, and hemodynamic resuscitation of septic patients, less is known about clinician ability to diagnose or predict the presumptive source of a septic patient's clinical syndrome. Despite limitations in patient presentation and physical findings, clinicians must make a scientific judgment for the potential source of infection and initiate appropriate therapy swiftly. Objectives: To evaluate the predictive ability of clinicians to determine the likely source or site of infection leading to severe sepsis and septic shock. Methods: This was a prospective observational trial. Data was collected at an urban tertiary care medical center ED from September 2017 to December 2019. Data was collected with the assistance of undergraduate research associates. Results: There were 111 patients included in the analysis, 62 (55.9%) were female, 89 (80.2%) were Black, and the mean age was 53.1 (SD 19.2) years. A high proportion had diabetes (36.9%) and hypertension (54.1%). The median time from patient arrival to treating clinician survey was 2 [IQR 1, 3] hours. The median time to antibiotic administration was 4 [IQR 2, 5] hours. Median ED length of stay was 8 [IQR 6, 12] hours. The accuracy of clinician suspicion for the source of infection was modest: 70.0% (95% CI 60.0 - 78.2%) for skin and soft tissue or abdominal sources, 82.2% (95% CI 74.7 - 89.6%) for urinary, and 42.3% (95% CI 32.6 - 52.3%) for pneumonia. In 8 cases of bacteremia, antibiotics were initiated for all patients. Conclusions: In conclusion, this study provides insight into the clinical characteristics and outcomes of a cohort of patients with suspected sepsis in an ED. The majority of patients were Black and had comorbidities such as diabetes and hypertension. Overall hospital mortality was low, and the accuracy of clinician suspicion for the source of infection was variable. The highest clinician suspicion accuracy was observed for urinary infections, followed by skin and soft tissue or abdominal sources. In such patients, clinical suspicion is a valuable tool in rapid identification of infection sources as it enables swift administration of treatment specific to the source of infection. In contrast, the accuracy for diagnosing pneumonia was particularly low so clinician suspicion is of much less utility in such cases. This highlights the need for improved diagnostic tools and protocols to aid clinicians in accurately identifying the source of infection in patients with suspected sepsis. The study also highlights potential areas for improvement in care, such as reducing the time to antibiotic administration. Addressing these issues could lead to superior, more targeted treatments for patients, ultimately improving outcomes and reducing the risk of morbidity and mortality associated with sepsis. No, authors do not have interests to disclose

PubMed ID

Not assigned




1097-6760, 0196-0644

First Page


Last Page