998. Missclassification of Community and Hospital Onset Bloodstream Infections Using Laboratory-Identified Events

Abstract Background Laboratory-identified bloodstream infections (LAB-ID-BSI) are classified as community onset (CO) if blood culture (BC) is collected within 3 days after facility admission and hospital onset if ≥4 days. This classification is often based on a computer-generated subtraction of the day of admission from day of onset. This method may miss recent prior hospitalizations at the same or different facilities. Methods We reviewed BC results (January 1, 2010–December 31, 2016), selected patients with BSI and defined the place of onset as CO (day 0–3) and HO (≥4 days) of admission based on LABID-BSI. All patients with CO were further evaluated to determine whether they were recently hospitalized. The source and microbiology of patients with hospitalization within 14 days of the onset of BSI was compared with HO and CO without prior admission within 6 months. Results We encountered 5,179 BSI episodes, 3866 (74.6%) were CO. Prior hospitalization in any hospital within 1–14 and 15–180 days of onset was documented in 659 (17.0%) and 1,465 (37.9%), respectively. Source of bacteremia and type of organisms in patients with prior hospitalization within 1–14 days were closer to HO than patients without prior hospitalization with higher frequency of Intravenous catheters (IVC), polymicrobial bacteremia, and candidemia (table). Conclusion Using Lab-ID events to classify BSI, one in six patients may risk being misclassified as CO. This underestimates BSI related to hospital setting. Onset classification should be based on thorough historical information and not a computer-generated subtraction of admission and Lab event dates. Infective endocarditis; soft tissue/bone; pneumonia; abdomen; unknown/miscellaneous; polymicrobial. Gram-positive; Gram-negative; anaerobes; Candida spp. a: P < 0.01; chi square test. Disclosures All authors: No reported disclosures.