Background: >Recent sepsis definitions proposed using the quick sequential organ failure assessment (qSOFA) to screen non-ICU patients for sepsis. This study tests how qSOFA, assessed alone and in combination with physician judgment, compares with unstructured physician judgment in the prediction of mortality among infected Emergency Department (ED) patients.
Methods: We performed a secondary analysis of a prospective, observational study of potentially septic ED patients, conducted between two urban, academic medical centers from July 2016 – December 2017. We enrolled ED patients with 1) two or more systemic inflammatory response syndrome criteria and severe sepsis qualifying organ dysfunction, 2) sBP <90 mmHg, or 3) lactate ≥4.0 mmol/L. Infectious etiology was adjudicated retrospectively by paired physician review with a third physician breaking ties. We excluded non-infected patients. For each patient, the treating physician provided a likelihood of in-hospital mortality (0-100%) at hospital admission. qSOFA scores were calculated using ED data only. The primary outcome was in-hospital mortality within 28 days. Physician prediction of mortality was modeled using logistic regression to generate an area under the curve (AUC). Models using qSOFA alone and a combination of physician judgment with qSOFA were likewise created. We used bootstrap methods, with 1000 replicate datasets, to derive 95% confidence intervals for each model AUC and assess differences between models.
Results: Of 405 patients meeting inclusion criteria, 195 (48.1%) were determined to have infection and analyzed. Of analyzed patients, 16 (8.2%) suffered in-hospital mortality within 28 days. Analyzed patients had a mean age of 58.3 (SD 16.5) years and 78 (40%) were female. qSOFA alone (0.63; 95%CI 0.53-0.73) was not superior to unstructured physician judgment (0.80; 95%CI 0.70-0.89) when predicting 28-day in-hospital mortality. Adding qSOFA to physician judgment did not improve model performance (0.79; 95%CI 0.69-0.89) compared to physician judgment alone.
Conclusion: When predicting 28-day in-hospital mortality, qSOFA did not outperform or add value to physician judgment among infected ED patients, suggesting a limitation to the tool’s clinical value.