Urinary tract infections (UTI) caused by carbapenem-resistant Gram-negative bacilli (CRGN) represent a significant threat to kidney transplant (KT) recipients, potentially resulting in complications and graft loss. The Desirability of Outcome Ranking (DOOR) strategy allows a global, patient-centered assessment of outcomes. This study aimed to validate a DOOR model adapted specifically for KT recipients by comparing clinical outcomes of UTI caused by carbapenem-resistant versus carbapenem-susceptible Gram-negative (CSGN) bacteria. Retrospective cohort study conducted between 2013 and 2023 at a tertiary hospital in Brazil. Patients ≥18 years, with previous KT and UTI due to Gram-negative bacilli, were included. Patients with CRGN infections were matched 1:1 with CSGN controls by year of infection. The primary outcome was a 14-day DOOR. DOOR categories were defined to reflect outcome severity, including occurrence of adverse events, lack of clinical response, infectious complications, transplant-related complications (e.g., rejection, CMV infection, and graft dysfunction), and death. Adjusted analyses were performed using propensity score weighting. A total of 194 patients were included (97 CRGN and 97 CSGN). The probability of a more favorable 14-day DOOR outcome was 41.3% (95% CI: 34.2%–48.7%; p = 0.02) for CRGN in the crude analysis, and 43.4% (95% CI: 36.2%–50.9%; p = 0.08) after adjustment. Transplant-related complications occurred in 29.9% of CRGN versus 15.5% of CSGN patients. In a subgroup analysis excluding patients on baseline dialysis, the results were maintained. Patients who developed UTI after the first post-transplant year had significantly better outcomes. UTI caused by CRGN in KT recipients were associated with worse 14-day clinical outcomes in the unadjusted analysis. The higher frequency of events such as graft dysfunction contributed to these poorer outcomes. The adapted DOOR model proved to be a promising tool for global assessment of outcomes in KT recipients with UTI and may be refined and applied in different epidemiological settings.
Baldissera et al. (2026) studied this question.