Prediction of serious infection during prehospital emergency care.

Background: Regionalization of emergency care for patients with serious infections has the potential to improve outcomes, but is not feasible without accurate identification of patients in the prehospital environment.

Objective: To determine the incremental predictive value of provider judgment in addition to prehospital physiologic variables for identifying patients who have serious infections.

Methods: We conducted a prospective study at a single teaching tertiary-care emergency department (ED) where a convenience sample of emergency medical services (EMS) providers and ED clinicians completed a questionnaire about the same patients. Prehospital providers provided limited demographics and work history about themselves. They also reported the presence of abnormal prehospital physiology for each patient (heart rate >90 beats/min, systolic blood pressure <100 mmHg, respiratory rate >20 breaths/min, pulse oximetry <95%, history of fever, altered mental status) and their judgment about whether the patient had an infection. At the end of formal evaluation in the ED, the physician was asked to complete a survey describing the same patient factors in addition to patient disposition. The primary outcome of serious infection was defined as the presence of both 1) ED report of acute infection and 2) patient admission. We included prehospital factors associated with serious infection in the prediction models. Operating characteristics for various cutoffs and the area under the curve (AUC) were calculated and reported with 95% confidence intervals (95% CIs).

Results: Serious infection occurred in 32 (16%) of 199 patients transported by EMS, 50% of whom were septic, and 16% of whom were admitted to the intensive care unit. Prehospital systolic blood pressure <100 mmHg, EMS-elicited history or suspicion of fever, and prehospital judgment of infection were associated with primary outcome. Presence of any one of these resulted in a sensitivity of 0.59 (95% CI 0.40-0.76) and a specificity of 0.81 (95% CI 0.74-0.86). The AUC for the model was 0.71.

Conclusions: Including prehospital provider impression to objective physiologic factors identified three more patients with infection at the cost of overtriaging five. Future research should determine the effect of training or diagnostic aids for improving the sensitivity of prehospital identification of patients with serious infection.