Derivation and validation of a new prediction score for bacteremia in the emergency department.

Early detection and treatment of bacteremia are crucial; however, blood culture takes days. We aimed to develop a prediction score using objective emergency department (ED)-available parameters. This single-center, retrospective cohort study included patients aged ≥ 16 years admitted to our ED with suspected bacteremia. The study was divided into derivation and validation periods. Data on vital signs, routine blood tests, and blood cultures were analyzed. Using multivariate logistic regression, a prediction score was developed and validated during the derivation and validation periods, respectively. A total of 3,725 and 3,471 patients were eligible for the derivation and validation periods, respectively. Continuous variables were converted to binary forms and used in logistic regression. The resulting prediction score was: Body temperature + Platelets × 2 + Neutrophil-lymphocyte ratio × 3 + Albumin + Bilirubin × 2 + Creatinine + Lactate × 2. The area under the curve in the receiver operating characteristic curve of the prediction score in the derivation and the validation period was 0.78 (95% confidence interval: 0.76–0.80) and 0.75 (95% confidence interval: 0.73–0.78), respectively. A prediction score based on routine blood tests and vital signs can effectively predict bacteremia in ED settings.