Clinical predictors of hospital-acquired bloodstream infections: A healthcare system analysis

Abstract

INTRODUCTION This study was performed to identify patient factors associated with hospital-acquired bloodstream infections (HABSI) to guide blood culture collection and empiric antibiotic therapy. METHODS A retrospective case-control study reviewed the medical records of 350 patients admitted to our health system from September 2017 to April 2020. The patients were 18 years and older and had at least one set of new positive non-contaminant blood cultures collected after 48 hours of admission, defined as HABSI. We developed clinical variables through a literature review associated with it. Univariate relationships between each variable and bacteremia were evaluated by chi-square test. A predictive model was developed through stepwise multivariate logistic regression. RESULTS The univariate analysis and stepwise regression analysis showed that temperature >100.4° F (OR: 1.9, CI 1.1 to 3.4), male sex (OR: 1.8, CI 1.0 to 3.0), and platelet count <150,000/µL (OR: 1.8, CI 1.0 to 3.2) were statistically associated with a positive blood culture. CONCLUSIONS This model helps identify patients with clinical characteristics associated with the likelihood of HABSI. This model can help guide the appropriate initiation of empiric antibiotics in clinical situations and assist with antibiotic stewardship.