Clinical-Microbiological Discordance and Delayed Mortality in OXA-48/ESBL-Producing Proteus mirabilis Infections

OXA-48/extended-spectrum β-lactamase (ESBL)-producing Proteus mirabilis may appear susceptible to carbapenems in routine testing, potentially resulting in inappropriate therapeutic decisions. We aimed to evaluate clinical outcomes and the association between antimicrobial therapy and mortality. We conducted a single-center observational study including 1254 patients with P. mirabilis infections (2017–2024). Seventy-nine OXA-48/ESBL cases (6.3%) were matched 2:1 with non-OXA-48/ESBL controls based on age, comorbidity, infection source, and severity. OXA-48 production was detected using the NG-Test CARBA 5 immunochromatographic assay, while ESBL production was inferred phenotypically using cefotaxime, ceftazidime, and cefepime discs alone and in combination with clavulanic acid. Predictors of 30-day mortality were assessed using time-dependent Cox regression models. OXA-48/ESBL-producing infections were associated with higher 30-day mortality (absolute difference 19.7%; population attributable fraction 18.2%). Differences in mortality emerged after day 8, suggesting delayed clinical deterioration. In time-dependent models, OXA-48/ESBL status was not associated with early mortality (days 0–7) but strongly predicted late mortality [hazard ratio (HR) 14.5; 95% confidence interval (CI) 3.19–65.93; p = 0.001]. Among patients with OXA-48/ESBL infections, carbapenem-based regimens were associated with higher mortality than alternative active therapies (50% vs. 20%; p = 0.04), remaining significant after adjustment (HR 3.37; 95% CI 1.27–8.9), although residual confounding by indication cannot be excluded. These observational findings support cautious interpretation of carbapenem susceptibility results and highlight the value of early carbapenemase detection to guide appropriate therapy. Graphical abstract available for this article. Some bacteria may appear susceptible to certain antibiotics in laboratory tests, even though treatment may not be effective in clinical practice. In this study, we focused on a type of Proteus mirabilis that produces OXA-48 enzymes and extended-spectrum beta-lactamases. These bacteria may appear susceptible to carbapenem antibiotics, even though treatment may not work as expected. We studied patients with Proteus mirabilis infections and compared patients with OXA-48/ESBL-producing strains with matched patients with non-OXA-48/ESBL strains. We found that patients with OXA-48/ESBL-producing bacteria had a higher risk of death within 30 days. Importantly, this increased risk did not appear early. Instead, patients often seemed stable at first but worsened after the first week of infection. This suggests that standard laboratory results may not fully reflect how these infections behave in clinical practice. We also observed that patients treated with carbapenems had higher mortality, an association that remained after adjustment for severity, although it should not be interpreted as causal. These findings show a mismatch between laboratory results and clinical outcomes. Doctors should be cautious when interpreting antibiotic susceptibility results and consider using rapid tests to detect resistance mechanisms. Early recognition of these bacteria may help guide better treatment decisions and improve patient outcomes.