Loss of colonization resistance within the gastrointestinal microbiome facilitates the expansion of pathogens and has been associated with death and infection in select populations. We tested whether gut microbiome features at the time of intensive care unit (ICU) admission predict death or infection.

This was a prospective cohort study of medical ICU adults. Rectal surveillance swabs were performed at admission, selectively cultured for vancomycin-resistant Enterococcus (VRE), and assessed using 16S rRNA gene sequencing. Patients were followed for 30 days for death or culture-proven bacterial infection.

Of 301 patients, 123 (41%) developed culture-proven infections and 76 (25%) died. Fecal biodiversity (Shannon index) did not differ based on death or infection (p = 0.49). The presence of specific pathogens at ICU admission was associated with subsequent infection with the same organism for Escherichia coli, Pseudomonas spp., Klebsiella spp., and Clostridium difficile, and VRE at admission was associated with subsequent Enterococcus infection. In a multivariable model adjusting for severity of illness, VRE colonization and Enterococcus domination (≥ 30% 16S reads) were both associated with death or all-cause infection (aHR 1.46, 95% CI 1.06–2.00 and aHR 1.47, 95% CI 1.00–2.19, respectively); among patients without VRE colonization, Enterococcus domination was associated with excess risk of death or infection (aHR 2.13, 95% CI 1.06–4.29).

Enterococcus status at ICU admission was associated with risk for death or all-cause infection, and rectal carriage of common ICU pathogens predicted specific infections. The gastrointestinal microbiome may have a role in risk stratification and early diagnosis of ICU infections.