Clinical efficacy of empiric antibiotic therapy in Chinese patients with cirrhosis and bacterial infections: A multicenter study.

Abstract

Bacterial and fungal infections are major drivers of acute decompensation and acute-on-chronic liver failure (ACLF) in cirrhosis, but real-world data on the effectiveness of empiric antibiotic therapy in China are scarce. We evaluated clinical response to empiric antibiotics, identified predictors of non-response, and assessed the impact of response and secondary infection on short-term outcomes in Chinese patients with cirrhosis. In this multicenter retrospective cohort from 24 tertiary centers, 1,401 adults with cirrhosis and bacterial or fungal infections were included. Clinical response to first-line empiric therapy occurred in 913 patients (65%). Non-responders had higher MELD/MELD-Na scores, more organ failures and ACLF, and more intense systemic inflammation. In multivariable models, greater ACLF grade 2-3, systemic inflammatory response syndrome, higher neutrophil-to-lymphocyte ratio, C-reactive protein, bilirubin, culture positivity, and lower albumin and mean arterial pressure independently predicted non-response. Non-response was independently associated with 28-day and 90-day mortality (HR 4.20 and 3.16, respectively), with consistent findings in time-dependent and center-clustered sensitivity analyses. Secondary infection developed in 7.0% of patients, was four-fold more frequent in non-responders (13.9% vs 3.3%), and nearly doubled 90-day mortality, whereas microbiological features at secondary infection did not clearly distinguish second-course responders from non-responders. Comparative analyses revealed substantial center-to-center and China-global differences in etiology, infection profile, and multidrug-resistant (MDR) burden. In Chinese patients with cirrhosis, lack of early clinical response to empiric antibiotics independently predicts short-term mortality and is associated with a higher risk of clinically distinct secondary infection. Our findings support risk-adapted, locally informed empiric strategies.