Microbiological Colonization Dynamics and 28-Day Mortality in Critically Ill Patients with Decompensated Cirrhosis: A Prospective Observational Study

Background: Repeated microbiological screening may reveal colonization dynamics associated with adverse outcomes in critically ill patients with decompensated cirrhosis. We investigated the association between bacterial and/or fungal colonization trajectories and 28-day mortality. Methods: This prospective single-center cohort included 150 patients screened using rectal, inguinal, and pharyngeal swabs at ICU admission and day 7. Patients were classified as persistently colonized (C-C), newly colonized (NC-C), or persistently non-colonized (NC-NC). Survival was assessed using Kaplan–Meier analysis and exploratory day 7 landmark Cox regression, with subsequent mortality evaluated through day 28 after ICU admission and continuous covariates entered without dichotomization. Sensitivity analyses included bootstrap resampling and substitution of SOFA for Child–Pugh class. Results: All patients completed both screenings and 28-day follow-up. The C-C, NC-C, and NC-NC groups comprised 91, 24, and 35 patients, with 28-day mortality of 65.9%, 87.5%, and 8.6%, respectively. Overall, 84 patients died (56.0%). ICU-acquired colonization remained associated with subsequent mortality after adjustment for disease severity, infectious complications, and other covariates (adjusted HR = 2.00, 95% CI: 1.17–3.43; p = 0.011). The association persisted when SOFA replaced Child–Pugh class (adjusted HR = 3.03, 95% CI: 1.71–5.37; p < 0.001). Bootstrap resampling yielded a similar median adjusted HR of 2.03 (95% bootstrap CI: 1.10–4.01). Conclusions: Acquisition of colonization identified a subgroup of critically ill patients with decompensated cirrhosis at increased risk of death by day 28, with the association persisting after adjustment for disease severity and infectious complications. Changes in colonization status may therefore carry prognostic information, although their clinical utility and the benefit of repeated screening require further validation.