The effect of duration of antimicrobial treatment for bacteremia in critically ill patients on in-hospital mortality - Retrospective double center analysis.

Excessive duration of antibiotic treatment is a major factor for inappropriate antibiotic consumption. Although in some instances shorter antibiotic courses are as efficient as longer ones, no specific recommendations as to the duration of antimicrobial treatment for bloodstream infections currently exist. In the present study, we investigated the effect of antibiotic treatment duration on in-hospital mortality using retrospective data from two cohorts that included patients with bacteremia at two Swiss tertiary Intensive Care Units (ICUs). Overall 8227 consecutive patients requiring ICU admission were screened for bacteremia between 01/2012-12/2013 in Lausanne and between 07/2016-05/2017 in Bern. Patients with an infection known to require prolonged treatment or having single positive blood culture with common contaminant pathogens were excluded. The primary outcome of interest was the time from start of antimicrobial treatment to in-hospital death or hospital discharge, whichever comes first. The predictor of interest was adequate antimicrobial treatment duration, further divided into shorter (≤10 days) and longer (>10 days) durations. A time-dependent Cox model and a cloning approach were used to address immortality bias. The secondary outcomes were the median duration of antimicrobial treatment for patients with bacteremia overall and stratified by underlying infectious syndrome and pathogens in the case of secondary bacteremia. Out of the 707 patients with positive blood cultures, 382 were included into the primary analysis. Median duration of antibiotic therapy was 14 days (IQR, 7-20). Most bacteremia (84%) were monomicrobial; 18% of all episodes were primary bacteremia. Respiratory (28%), intra-abdominal (23%) and catheter infections (17%) were the most common sources of secondary bacteremia. Using methods to mitigate the risk of confounding associated with antibiotic treatment durations, shorter versus longer treatment groups showed no differences in in-hospital survival (time-dependent Cox-model: HR 1.5, 95% CI (0.8, 2.7), p = 0.20; Cloning approach: HR 1.0, 95% CI (0.7,1.5) p = 0.83). Sensitivity analyses showed that the interpretation did not change when using a 7 days cut-off. In this restrospective study, we found no evidence for a survival benefit of longer (>10 days) versus shorter treatment course in ICU patients with bacteremia. The study was retrospectively registered on clinicatrials.gov (NCT05236283), 11 February 2022. The respective cantonal ethics commission (KEK Bern # 2021-02302) has approved the study