Carbapenem-resistant Gram-negative bacteremia in a medical intensive care unit: predictors, antimicrobial susceptibilities, and its effects on mortality.

BACKGROUND

Carbapenem-resistant Gram-negative bacteremia (CR-GNB) is a serious threat to critically ill patients. The present study aimed to (a) uncover drug resistance of pathogens and CR-GNB’s effect on prognosis; and (b) distinguish predictors of CR-GNB among patients from a medical intensive care unit (ICU) setting.

METHODS

This was a single-center, retrospective cohort study of all consecutive patients with nosocomial GNB in the medical ICU between January 1, 2015 and February 29, 2024. The predictors of CR-GNB were analyzed by logistic regression. Survivals of patients undergoing CR-GNB or carbapenem-susceptible GNB (CS-GNB) were compared using the Kaplan-Meier method.

RESULTS

Seventy-four episodes of GNB occurred in 67 patients out of 4914 patients, among them, 48 episodes of CR-GNB occurred in 42 (0.9%) patients. The most common primary source of infection leading to CR-GNB was the respiratory tract (accounting for 71.4% of all cases) and Klebsiella pneumoniae (41.7%) the most frequent isolate. Antimicrobial susceptibility testing revealed that CR-GNB isolates exhibited extremely high resistance rates (> 90%) to most tested antibiotics. After excluding intrinsically resistant organisms, the resistance rate was 14.3% for polymyxin B and 48.8% for tigecycline. Most notably, this analysis confirmed that all evaluable isolates of Klebsiella spp. and pseudomonas aeruginosa remained fully susceptible to ceftazidime/avibactam (0% resistance). Among the 67 patients with GNB, the 30-day all-cause mortality was 68.7% (46/67). The mortality was particularly high in the 42 patients with CR-GNB, reaching 81.0% (34/42). Within this high-risk group, 69.0% (29/42) received inappropriate empirical antibiotic therapy, and their mortality was 82.8% (24/29). On multivariate analysis, risk factors for CR-GNB were invasive mechanical ventilation before bacteremia [odds ratio (OR) = 7.680, 95% confidence interval(CI): 1.119–52.709, P = 0.038], carbapenem use > 2 days before GNB (OR = 5.626, 95% CI: 1.074–29.480, P = 0.041), and length of hospitalization ≥ 7 days before bacteremia [OR = 8.683, 95%CI: 2.009–37.538, P = 0.004]. CR-GNB was related to increased 30-day all-cause mortality compared to CS-GNB (P = 0.005). The poor prognosis in patients with CR-GNB was discovered in the analysis of Kaplan-Meier curve compared to those with CS-GNB (P = 0.018).

CONCLUSIONS

CR-GNB, predicted by prior carbapenem exposure, prior mechanical ventilation and long hospital stay, is characterized by near-pan-drug resistance and confers an over 80% mortality rate in the medical ICU, demanding urgent preventive and diagnostic measures. CLINICAL TRIAL NUMBER: Not applicable.