Microbiology and predictors of mortality in haematological malignancy patients with gram-negative bacterial bloodstream infections.

Objectives: Gram-negative bacteria (GNB) bloodstream infection (BSI) pose a significant clinical challenge in patients with haematological malignancy, further complicated by rising carbapenem-resistant GNB (CRGNB) prevalence. This study aims to investigate the prevalence and risk factors for GNB BSI and associated mortality in this population.

Methods: A retrospective study was conducted at a tertiary teaching hospital in southern China (2015-2023), including haematological malignancy patients with GNB BSI. Multivariate logistic analyses were performed to identify risk factors for CRGNB BSI and to establish a predictive model for 30-day mortality of GNB BSI.

Results: Among 351 patients with GNB BSIs, acute myeloid leukaemia (51.3 %) was the predominant underlying disease. Escherichia coli (28.8 %) and Klebsiella pneumoniae (29.7 %) were the most common GNB BSI and CRGNB BSI pathogens, respectively. Independent risk factors for CRGNB BSI included chronic liver disease, prior carbapenems therapy (≤30 days before BSI), a platelet count <30 × 10⁹/l and albumin concentration <30 g/l before BSI. The 30-day mortality prediction model of GNB BSI incorporated CRGNB infection, platelet count <30 × 10⁹/l and albumin concentration <30 g/l before BSI, demonstrating good discrimination (training cohort AUC: 0.828; validation cohort: 0.791). Calibration plots and decision curve analysis confirmed robust model performance.

Conclusions: Identified factors enable risk stratification for CRGNB BSI and poor prognosis in GNB BSI, facilitating timely interventions to improve outcomes.