Reliability of subjective fever in triage of adult patients.

Abstract

To determine whether a historical complaint of fever is predictive of fever on emergency department triage. We prospectively questioned 651 ambulatory adult patients in a military tertiary care emergency department as to whether they had fever before oral temperature was taken. Fever was defined as a temperature of 38 degrees C (100.4 degrees F) or greater. Sensitivity and specificity were 84% (95% confidence interval [CI], 71% to 95%) and 83% (95% CI, 80% to 86%), respectively. The prevalence of objective fever was 6.4%, yielding positive and negative predictive values of 25% (95% CI, 18% to 32%) and 99% (95% CI, 93% to 100%), respectively. Overall accuracy was 83% (95% CI, 80% to 86%). In this study, outpatients were fairly accurate in predicting fever. However, in an outpatient population with a low overall prevalence of objective fever, the predictive value of a complaint of fever representing an objective fever remained low. Therefore the complaint of subjective fever should be interpreted with caution when it is used to support a given diagnosis in an ambulatory care setting.