Research Communications of the 27(th) ECVIM-CA Congress: Intercontinental, Saint Julian's, Malta, 14th to 16th September 2017

Disclosures: No disclosures to report. Bloodstream infections are a substantial cause of morbidity and mortality in critically ill veterinary patients. Blood cultures are the gold standard for diagnosis of bacteremia, but are infrequently obtained due to technical and practical difficulties in sample acquisition. In ill patients, urine cultures are sometimes recommended as surrogates for blood cultures as part of a "better than nothing" approach. This study evaluated the ability of urine culture to predict blood-stream infection. We retrospectively evaluated all blood, aerobic, and anaerobic cultures submitted at NC State Veterinary Hospital between 2011 and 2016. We calculated growth rates of 18% (blood), 24% (urine), and 61% (non-urine) from 511 blood, 6797 urine, and 6552 non-urine cultures submitted. Blood isolates were most commonly coagulase-positive Staphylococcus spp (27%) and Escherichia coli (14%); Escherichia coli was the most common urinary isolate (43%), along with Enterococcus (14%) and coagulase-positive Staphylococcus (11%). 324 urine and blood samples were submitted in parallel, of which 21 yielded simultaneous growth. Of these, only 14 samples were concordant, while 7 yielded discordant urinary and bloodstream infections. Overall, urinary isolates were poorly reflective of bloodstream isolates, with a sensitivity of 24% but a specificity of 87%. General concordance, including true positive (n = 14) and true negatives (n = 232), between urinary and bloodstream isolates was 76%. Urine culture isolates had a poor positive predictive value (29%) but a negative predictive value of 84% for bloodstream infection. An apparent exception is patients with suspected urogenital infection sources (renal, prostatic, etc), in which 100% (n = 7) had concordant urinary and bloodstream infections. Coagulase-positive Staphylococcus infections were most likely to be concordant. 133 non-urine samples submitted in parallel with blood cultures were also evaluated; only biliary and intravenous catheter samples carried a PPV >40%. In short, we recommend that if bloodstream infection is suspected, blood cultures be acquired.