528. Bacterial vs viral etiology of fever: a prospective study to estimate a host score’s potential to support ED physicians' etiologic accuracy

Abstract Background A host-protein score (BV) combining the circulating levels of TRAIL, IP-10 and CRP was developed for distinguishing bacterial from viral infection. BV’s potential to impact decision making at emergency department (ED) assessed by comparing BV physician’s etiological suspicion. Methods Rosetta study participants, aged 3 months 18 years with febrile respiratory tract infection or fever without source, were prospectively recruited in a tertiary care pediatric ED. The main eligibility criteria symptoms less than 7 days immunocompetence. ED initial suspicion (bacterial/viral/I don’t know) degree confidence (low/medium/high) recorded questionnaire. interpreted based on pre-defined thresholds (viral/bacterial/equivocal); results not available physician. etiologic reference standard diagnoses (bacterial/viral/indeterminate) independent pediatricians who provided all patient data but blinded BV. accuracy estimated according alignment between as filled questionnaire (assuming full adoption physician). Results Among 348 patients recruited, 290 met had requisite serum sample (Figure 1A). Median age 1.3 (interquartile range 1.7), 37.5% female. 211 questionnaires filled. matched label 72% cases (151/211; Figure 1B); these, 52% (78/151) labeled physician low/medium confidence, i.e., could reinforce In 12% (26/211) aligned suspicion, therefore correct physician's diagnosis. did align 7% (14/211) equivocal 9% (20/211). Overall, potentially diagnosis reduce error >2-fold, 16.1% 8.1%. 1.Assessing accuracy. (A) Reference outcomes scores. (B) versus adjudication-based Cases missing omitted; n=211. Conclusion has aid diagnostic process. Future studies are warranted assess real-world practice. Disclosures Meital Paz, MD, MeMed: employee Oded Scheuerman, Grant/Research Support Ron Berant, GE: Advisor/Consultant.