Appropriateness of empiric antimicrobial therapy and short- term mortality in critically ill patients with urosepsis: a retrospective cohort study

Abstract Background Empiric antimicrobial therapy must be initiated before culture results are available in patients with urosepsis. Whether the appropriateness of empiric coverage affects outcomes in critically ill patients with urosepsis remains poorly defined. We previously reported that, in this cohort, early antibiotic initiation alone was not significantly associated with survival. We therefore evaluated the association between the appropriateness of empiric antimicrobial therapy and short-term mortality. Methods We conducted a retrospective cohort study of adult ICU patients with urosepsis at a tertiary hospital (January 2015–December 2024). Empiric therapy, defined as all agents started within 24 h of the first antimicrobial dose, was classified as appropriate if at least one agent had in vitro activity against the pathogen from the first urine culture after disease onset, using rules based on species identification, extended-spectrum β-lactamase (ESBL) status, and resistance patterns. Culture-negative patients were not assessable. The primary outcome was 28-day all-cause mortality. Associations were estimated by logistic regression, with Firth penalization, propensity-score weighting, and multiple imputation as sensitivity analyses. Clinical trial number: not applicable. Results Of 92 included patients, 62 (67.4%) had a positive first urine culture and were assessable: 51 (82.3%) received appropriate and 11 (17.7%) inappropriate empiric therapy. Inappropriate therapy comprised Candida spp. without empiric antifungal coverage (n = 6), ESBL-producing Enterobacterales without a carbapenem (n = 4), and Enterococcus faecium treated with meropenem (n = 1). Baseline severity was comparable between groups. Twenty-eight-day mortality was 54.5% versus 21.6% (P = 0.056) and in-hospital mortality 27.3% versus 7.8% (P = 0.099) in the inappropriate versus appropriate groups. After adjustment for APACHE II score, inappropriate empiric therapy was associated with increased 28-day mortality (adjusted OR 5.75, 95% CI 1.06–31.3; P = 0.043), with consistent estimates across sensitivity analyses (adjusted ORs 3.97–5.75). Conclusions In critically ill patients with culture-positive urosepsis, inappropriate empiric antimicrobial therapy was associated with higher short-term mortality, although wide confidence intervals indicate imprecision. Unrecognized candiduria and ESBL-producing Enterobacterales without a carbapenem were the principal sources of inappropriate coverage. In obstructive urosepsis, the correctness of empiric coverage may be as important as the speed of initiation. These findings require validation in prospective cohorts.