ABSTRACT
OBJECTIVES: We examined factors associated with follow-up blood cultures (FUBCs) in patients with monomicrobial Gram-negative (GN) bloodstream infection (BSI) and investigated the impact of FUBCs on therapeutic management and patient outcome. METHODS: A retrospective cohort analysis was conducted of adult patients diagnosed with GN-BSI at a tertiary-care university hospital during 2013-2016. FUBCs performed between 24 hours and 7 days after index BCs was the exposure variable. Risk factors for 30-day mortality were analysed by multivariate Cox analysis on the overall cohort, including FUBCs as a time-varying covariate and on 1:1 matched patients according to Sequential Organ Failure Assessment (SOFA) score and time to FUBC. RESULTS: In 278 (17.6%) of 1576 patients, FUBCs were performed within a median of 3 and 2 days after index BCs and active antibiotic therapy initiation. Persistent BSI was found in 107 (38.5%) of 278 patients. FUBCs were performed in more severely ill patients, with nonurinary sources, difficult-to-treat pathogens and receipt of initial inappropriate therapy. Source control and infectious disease consultation rates were higher among patients with preceding FUBCs and was associated with longer treatment duration. Thirty-day mortality was 10.4%. Independent risk factors for mortality were Charlson comorbidity index (hazard ratio (HR) 1.12) SOFA (HR 1.11), septic shock (HR 2.64), urinary source (HR 0.60), central venous catheter source (HR 2.30), complicated BSI (HR 2.10), carbapenem resistance (HR 2.34), active empiric therapy (HR 0.68), source control (HR 0.34) and FUBCs (HR 0.48). Association between FUBCs and lower mortality was confirmed in the 274 matched pairs. CONCLUSIONS: FUBCs were performed in more severe GN-BSIs, yielding a high rate of persistent BSI. In this context, FUBCs were associated with lower mortality.
