Clinical Characteristics and Prognosis of Hospital-Acquired Klebsiella pneumoniae Bacteremic Pneumonia versus Escherichia coli Bacteremic Pneumonia: A Retrospective Comparative Study.

Abstract

This research aimed to investigate the variations in clinical features and prognosis of HABP caused by E. coli and K. pneumoniae. We also aimed to evaluate the risk variables related to 30-day death in the investigated groups. A single-center retrospective cohort research lasting four years was performed. A total of 117 patients with HABP were involved in this research. The primary prognosis was 30-day death. Among 117 patients with HABP, 60 patients were infected with K. pneumoniae (KP-HABP), and 57 patients were infected with E. coli (E. coli-HABP). A higher proportion of males, ICU admission, undergoing tracheotomy and trachea cannulation, carbapenem-resistant strains, inappropriate empirical therapy (IET), immune compromise, diabetes mellitus, and sepsis were observed in the patients with KP-HABP (all P < 0.05). Meanwhile, the median SOFA score and Pitt score were significantly (P < 0.001) higher in the KP-HABP group compared to the E. coli-HABP group. The 30-day death was 48.33% in the KP-HABP group and 24.56% in the E. coli-HABP group (P = 0.008). After adjusting for the main covariates, the hazard ratios for 30-day mortality in KP-HABP were 1.58 (95% CI:0.80-3.12), 3.24 (95% CI:1.48-7.06), 5.67 (95% CI:2.00-16.07), and 5.99 (95% CI:2.10-17.06), respectively. Multivariate logistic regression models revealed that IET, hypoproteinaemia, cerebral vascular disease (CVD), and SOFA score ≥ 5.0 were the independent risk variables for 30-day death in KP-HABP. Simultaneously, SOFA score ≥ 4.0 and Pitt score ≥ 2.0 were independent risk factors for 30-day mortality in E. coli-HABP. The clinical features of HABP vary depending on whether it is caused by Escherichia coli or K. pneumoniae. KP-HABP patients have higher 30-day mortality than E. coli-HABP patients. To ensure greater validity, it is necessary to further verify this conclusion using a larger sample size.