1835. Evaluation of Predicting Performance of POSITIVE, PREDICT and VIRSTA Score for Infective Endocarditis in Patients with Staphylococcus aureus Bacteremia

Abstract Background Prompt identification of Infective Endocarditis (IE) in patients with Staphylococcus aureus bacteremia (SAB) is important as delayed diagnosis results in 15-20% increased mortality. Predictive clinical scoring strategies have been developed to optimize the diagnostic process, particularly the use of invasive methods such as Transesophageal Echocardiography(TEE), known to be superior to Transthoracic Echocardiography but is not suitable for all patients. Three scoring systems have been proposed to predict the risk of IE: VIRSTA, PREDICT and POSITIVE. We compared these scoring strategies in combination with other clinical data to test the hypothesis that inclusion of time to blood culture positivity(TTP) in the scoring strategies might improve the sensitivity and specificity of clinical risk stratification. Methods Adults (≥ 18y) with SAB admitted to Ochsner LSU Health Shreveport in 2020-2021 were retrospectively screened. Patients with polymicrobial bacteremia, with index blood cultures obtained at a different institution and without cardiac imaging were excluded. In this preliminary analysis, 60 patients (30 with and without IE) were studied. IE was defined as patients who met modified Dukes criteria for definite IE. The three scores were calculated for each subject. TTP, defined as the time from incubation to automated detection, was obtained from the microbiology laboratory. Clinical predictors of IE were identified using multivariable logistic regression analysis. Results The demographic, baseline characteristics and mode of presentation, was no different in both groups based on analysis of >50 factors. The mean TTP for subjects with and without IE were no different (9.65h vs 11h, p=0.12) TTP≤13h as an independent variable was found to be sensitive but not specific. VIRSTA was found to be the most sensitive scoring method. POSITIVE, which includes TTP as a component, was the most accurate, but least sensitive method. Addition of TTP≤13h to VIRSTA and PREDICT increased sensitivity but lowered specificity. Conclusion While the sensitivity of scoring strategies is acceptable, none of the proposed methods has adequate specificity. Larger studies are needed for the development of a highly specific and accurate scoring method that includes classic risk factors and TTP. Disclosures All Authors: No reported disclosures.