Clinical Impact of Multidrug-Resistant Bacterial Infections in Critically Ill Patients: A Retrospective Cohort Study.

Background/

Objectives

Antimicrobial-resistant bacterial infections are frequent in intensive care units (ICUs), but the independent effect of multidrug resistance relative to infection with susceptible organisms remains uncertain. We compared clinical characteristics, microbiology, treatment intensity, and in-hospital mortality among patients with no documented bacterial infection, antimicrobial-susceptible infection, or multidrug-resistant (MDR) infection.

Methods

We conducted a single-center retrospective cohort study of the first eligible ICU hospitalization for each patient at the University Clinical Hospital in Bialystok, Poland, from 1 January 2017 to 1 June 2023. Infection groups were assigned using clinical documentation, microbiological results, and local laboratory resistance-phenotype coding; patients with multiple isolates were classified according to the most resistant clinically relevant isolate. Multivariable logistic regression evaluated associations with in-hospital mortality.

Results

Among 3326 patients, 1413 (42.5%) had a documented bacterial infection: 481 (34.0% of infected patients) had susceptible infection and 932 (66.0%) had MDR infection. Crude mortality was 45.5%, 46.8%, and 48.4% in the no-infection, susceptible-infection, and MDR-infection groups, respectively (p = 0.354), whereas mean ICU length of stay was 10.0, 15.5, and 25.2 days (p < 0.001). In the complete-case whole-cohort model (n = 678), MDR infection (adjusted OR 2.70, 95% CI 1.84-3.96) and susceptible infection (adjusted OR 2.16, 95% CI 1.26-3.69) were associated with in-hospital death versus no infection. Within the infected cohort (n = 352), MDR status was not independently associated with mortality versus susceptible infection (adjusted OR 1.09, 95% CI 0.61-1.93).

Conclusions

MDR infection identified patients with prolonged ICU exposure and greater organ-support requirements but did not independently increase mortality relative to susceptible infection. Interpretation is limited by the retrospective single-center design, complete-case analysis, and absence of time-resolved MDR acquisition data.