Impact on clinical outcome of follow-up blood cultures and risk factors for persistent bacteraemia in patients with gram-negative bloodstream infections: a systematic review with meta-analysis

Background The clinical usefulness of follow-up blood cultures (FUBCs) in gram-negative bloodstream infections (GN-BSIs) represents a debated issue. Objective To assess the impact on the clinical outcome of FUBCs in patients with GN-BSI and to predict risk factors for persistent bacteraemia. Data sources PubMed-MEDLINE, Scopus, and the Cochrane Library Database were independently searched until 24 June, 2022. Study eligibility criteria Randomized controlled trials, prospective, or retrospective observational studies, including patients affected by GN-BSIs. Primary endpoints were in-hospital mortality rate, and persistent blood stream infections were defined as FUBC-positive for the same pathogen isolated from index blood cultures (BCs). Participants Hospitalized patients with documented GN-BSIs. Intervention Performance of FUBCs (defined as subsequent BCs collected at least 24 hours after index BCs). Assessment of risk of bias Quality of included studies was independently assessed according to the Cochrane Risk of Bias Tool and the Risk Of Bias In Non-randomized Studies of Interventions. Methods of data synthesis Meta-analysis was performed by pooling odds ratio (OR) retrieved from studies providing adjustment for confounders using random-effect model with the inverse variance method. Risk factors for persistent blood stream infections were also assessed. Results A total of 3747 articles were screened, and 11 observational studies (6 assessing impact on outcome (N = 4631), and 5 investigating risk factors for persistent GN-BSI (N = 2566)), conducted between 2002 and 2020 were included. The execution of FUBCs was associated with a significantly lower risk of mortality (OR, 0.58; 95% CI, 0.49–0.70; I2 = 0.0%). The presence of end-stage renal disease (OR, 2.99; 95% CI, 1.77–5.05), central venous catheter (OR, 3.30; 95% CI, 1.82–5.95), infections due to extended-spectrum β-lactamase-producing strains (OR, 2.25; 95% CI, 1.18–4.28), resistance to empirical treatment (OR, 2.70; 95% CI, 1.65–4.41), and unfavourable response at 48 hours (OR, 2.99; 95% CI, 1.44–6.24) emerged as independent risk factors for persistent bacteraemia. Conclusions The execution of FUBCs is associated with a significantly low risk of mortality in patients with GN-BSIs. Our analysis could be useful to stratify patients at a high risk of persistent bacteraemia to optimize the use of FUBCs.