Abstract

--- - Background Infections have a poor prognosis in inpatients with cirrhosis. We aimed to determine regional variations in infections and their association with clinical outcomes in a global cohort of inpatients with cirrhosis. Methods In this prospective cohort study initiated by the CLEARED Consortium, we enrolled adults (aged >18 years) with cirrhosis who were non-electively admitted to 98 hospitals from 26 countries or regions across six continents between Nov 5, 2021, and Dec 10, 2022. Data at admission, during hospitalisation, and for 30 days after discharge were collected through patient reports and chart reviews. Collected data included demographics; country and country income level per World Bank classifications (high-income countries [HICs], upper-middle-income countries [UMICs], and low-income or lower-middle-income countries [L-LMICs]); comorbidities; characteristics related to cirrhosis and the infections, including types, culture results, and drug resistance profile; antibiotic use; and disease course while hospitalised and for 30 days post-discharge. The primary outcome was in-hospital death or hospice referral in those with versus those without an admission infection (defined by the presence of infection on or within 48 h of admission). Multivariable log-binomial regression for in-hospital death or hospice referral was performed to identify risk factors. Findings Of 4550 patients screened, 4238 patients (mean age 561 years [SD 133]; 2711 [640%] male and 1527 [360%] female) with complete data were enrolled. 1351 (319%) had admission infections. A higher proportion of patients in L-LMICs had infections (318 [417%] of 762 vs 444 [583%] without infection) than in UMICs (588 [306%] of 1922 vs 1334 [694%]) or HICs (445 [286%] of 1554 vs 1109 [714%]). Patients with admission infections had worse severity of cirrhosis and were more likely to have had an infection or been hospitalised in the preceding 6 months. The most common specific infection types were spontaneous bacterial peritonitis (391 [289%] of 1351), pneumonia (233 [172%]), and urinary tract infections (193 [143%]). 549 (406%) patients were culture-positive for bacterial or fungal infections, with the lowest culture-positive rates in Africa and mainland China. Most of the isolated organisms were Gram-negative (345 [63%] of 549), then Gram-positive (157 [29%]), and then fungi or mixed (47 [9%]), with Escherichia coli, Klebsiella pneumoniae, and Enterococcus spp being the top three isolated pathogens. The overall rate of drug resistance was 40% (220 of 549 with positive cultures), being highest in UMICs. The most used empirical antimicrobials were third-generation cephalosporins (453 [37%] of 1241), followed by the broad-spectrum beta-lactams and beta-lactamase inhibitors (289 [23%]). De-escalation was observed in 62 (20%) of 304 patients who had their antibiotics changed. Patients with versus without admission infections had a higher rate of in-hospital death or hospice transfer (299 [221%] of 1351 vs 232 [80%] of 2887; p<00001), a result replicated in multivariable analysis (adjusted risk ratio 175 [95% CI 142-206]; p<00001). - Older age, self-reported female gender, not being in a HIC, lactulose use, and higher MELD-Na score were also associated with in-hospital death or hospice transfer on multivariable analysis. Interpretation In the CLEARED Consortium cohort of inpatients with cirrhosis, the rates and types of infections, causative organisms, and culture-positivity varied substantially across regions, and infections were associated with a higher mortality risk. Culture positivity, which guides appropriate antibiotic use, was low. Taking a global perspective, considering regional variations in infections, drug resistance, and resources, could help to alleviate disparities in burden and outcomes.