Retrospective Analysis of Infection Characteristics and Prognostic Factors in ICU Sepsis Patients in Jiaxing Area

Sepsis remains a leading cause of morbidity and mortality in intensive care units (ICUs), with antimicrobial resistance (AMR) complicating therapeutic decisions. This study aimed to characterize pathogen distribution, AMR patterns, and prognostic indicators in ICU patients with sepsis at Jiaxing First Hospital.

A retrospective analysis of 352 septic patients (2021-2024) was conducted. Clinical outcomes, pathogen profiles, and antimicrobial susceptibility were evaluated. Multivariable logistic regression and ROC curve analysis identified predictors of non-recovery.

Gram-negative bacteria predominated (67.31%), with Klebsiella pneumoniae (K. pneumoniae), Escherichia coli (E. coli), Acinetobacter baumannii (A. baumannii), and Pseudomonas aeruginosa (P. aeruginosa) as the most frequent isolates. Notably, A. baumannii exhibited extensive resistance (>80% to ceftazidime, ciprofloxacin, and carbapenems) and the highest Non-recovery rate (60.98%). Among Gram-positive isolates, Enterococcus faecium (E. faecium), Staphylococcus aureus (S. aureus), and Enterococcus faecalis (E. faecalis) were prevalent, showing high susceptibility to vancomycin, tigecycline, and quinupristin-dalfopristin. Pulmonary infection was the predominant source (37.8%), primarily involving A. baumannii and K. pneumoniae, while intestinal translocation was twice as frequent in non-recovery patients. Independent predictors of non-recovery included age, pneumonia, shock, respiratory failure (RF), and lactate (LA). Their combination yielded a robust prognostic model (AUC 0.813), with RF as the strongest contributor.

This study highlights the epidemiological landscape of ICU sepsis and emerging AMR patterns, particularly for A. baumannii. A prognostic model using routine parameters enables early identification of high-risk patients. Targeted interventions-updated guidelines, enhanced infection control, and antimicrobial stewardship-may improve outcomes and warrant prospective validation.