Bedside risk prediction for positive follow-up blood culture in Gram-negative bacilli bacteremia: for whom is follow-up blood culture useful?

Purpose The value of follow-up blood culture (FUBC) in Gram-negative bacteremia (GNB) management is controversial. We evaluated bedside risk predictors and their probabilities of yielding positive FUBCs in GNB. Methods All adult patients with GNB in a 2700-bed tertiary center were retrospectively enrolled between January 2019 and December 2019. Only one initial GNB episode was included per patient. Positive FUBC was defined as isolation of the same organism in blood culture 48-72 h after the initial blood culture. Results A total of 2216 patients with GNB were identified, of whom 34.4% underwent FUBC. Of the 645 patients with FUBCs analyzed in the study, 89 (13.8%) had positive FUBCs. In multivariate analysis, hemodialysis [adjusted odds ratio (aOR), 2.6], fever on the day of FUBCs (aOR 3.6), intravascular device (aOR 2.4), no use of in vitro active antibiotic within 24 h (aOR 2.5), non-fermenting bacteria (aOR 4.7), and multidrug resistance (aOR 5.4) were independent risk factors for positive FUBCs. If microbiological results were excluded in multivariate analysis, hemodialysis, immunosuppressive treatment, fever on the day of FUBCs, and intravascular device were independent bedside risk predictors for positive FUBCs. The yield of FUBCs increased from 3.0% (95% CI 1.0-7.0) to 63.6% (95% CI 25.6-100) as the number of bedside risk predictors increased from 0 to 4. In addition, positive FUBCs were significantly associated with 30 day mortality. Conclusions FUBCs may not need to be routinely used for patients with GNB bacteremia, and bedside risk predictors could be helpful in identifying patients for whom FUBC is likely to be useful.