Risk of extended-spectrum β-lactamase-producing Enterobacterales colonization and infection in patients with cirrhosis admitted to the intensive care for gastrointestinal bleeding: a multicenter retrospective observational study

The changing epidemiology of multidrug-resistant organisms, community spread of extended-spectrum β-lactamase-producing Enterobacterales (ESBL-E), and need for prudent carbapenem use underscore the importance of identifying patients at risk of ESBL-E colonization and infection. We assessed the prevalence, risk factors, and clinical impact of rectal ESBL-E colonization in patients with cirrhosis admitted to intensive care units (ICUs) for gastrointestinal bleeding. This retrospective multicenter cohort study included 522 adults with cirrhosis admitted for gastrointestinal bleeding to 15 French ICUs. Patients without rectal screening at admission were excluded. We assessed ESBL-E colonization at ICU admission, associated factors, infectious outcomes, and the predictive value of ESBL-E carriage for ESBL-E-related ICU-acquired infection (ICUAI). ESBL-E colonization at ICU admission was identified in 64 patients (12%), mainly Escherichia coli ( n = 28). Factors independently associated with colonization were immunosuppression [adjusted odds ratio (aOR)=2.34 (95%CI 1.23–4.43), p = 0.009] and antibiotic use during the previous 3 months [aOR = 1.94 (95% CI 1.00–3.68), p = 0.047], whereas chronic alcohol use was inversely associated with carriage [aOR = 0.41 (95%CI 0.22–0.78), p = 0.006]. Infection at ICU admission was present in only 33 patients (6%). Among 504 patients with an ICU stay > 24 hours, ICUAI occurred in 103 (20%) and was associated with higher ICU and hospital mortality. Independent predictors of ICUAI were transjugular intrahepatic portosystemic shunt before admission [adjusted subdistribution hazard ratio (aSHR)=2.84 (95% CI 1.17–6.88), p = 0.021], vasopressor use at admission [aSHR = 2.60 (95%CI 1.32–5.10), p = 0.006], and mechanical ventilation > 24 hours [aSHR = 3.93 (95%CI 1.78–8.68), p = 0.001]. ESBL-E-related infections were infrequent ( n = 15, 2.8%). Among patients who developed ICUAI, previous ESBL-E colonization was strongly associated with ESBL-E-related ICUAI [OR = 36.36 (95%CI 8.3–258.4), p < 0.001]. Admission ESBL-E carriage predicted ESBL-E-related ICUAI with 83% sensitivity, 88% specificity, 48% positive predictive value, and 98% negative predictive value. Among patients with cirrhosis admitted to the ICU for gastrointestinal bleeding, ESBL-E colonization was present in 12% at admission, whereas ESBL-E-related infections were uncommon. These findings do not support routine carbapenem-based prophylaxis. However, because most ESBL-E-related ICUAIs occurred in previously colonized patients and the absence of carriage had a high negative predictive value, rectal ESBL-E screening may help guide empiric antibiotic therapy when ICUAI is suspected. Not applicable.