Clinical and microbiological determinants of treatment failure in complicated urinary tract infection: a four-hospital cohort of enterobacterales infections in Misurata, Libya

Background Complicated urinary tract infection (cUTI) is influenced by host vulnerability, urinary tract abnormalities, antimicrobial resistance, and bacterial persistence. We evaluated whether phenotypic beta-lactamase co-production and biofilm formation were independently associated with treatment failure among adults with Enterobacterales cUTI in Misurata, Libya. Methods This prospective multicenter cohort included 963 cUTI episodes and 1,017 Enterobacterales isolates collected continuously at four hospitals from 1 March 2024 through 31 March 2026. Identification combined conventional biochemical testing, API 20E, and MALDI Biotyper confirmation. Antimicrobial susceptibility testing used CLSI disk diffusion; selected isolates underwent agar-dilution minimum inhibitory concentration testing, phenotypic beta-lactamase testing, and triplicate microtiter-plate biofilm assessment. The prespecified composite outcome was reconciled against its recorded component fields before modeling; documented same-organism persistence at 3–7 days constituted treatment failure under the prespecified rule. Multivariable logistic regression included prespecified covariates and hospital fixed effects and was used for explanatory association estimation rather than clinical prediction. Results Rule-based outcome reconciliation identified 188/963 treatment failures (19.5%), including 36 episodes with same-organism persistence that had been discordantly coded as successes. Obstruction (adjusted odds ratio [aOR] 1.55, 95% CI 1.06-2.26), diabetes (aOR 1.48, 95% CI 1.06-2.09), and sepsis (aOR 1.59, 95% CI 1.10-2.29) were associated with failure after adjustment. Phenotypic beta-lactamase co-production was not independently associated with failure (aOR 0.62, 95% CI 0.35-1.09). Biofilm positivity showed an inverse association in the primary model (aOR 0.70, 95% CI 0.49-0.99), but this finding was not robust across sensitivity analyses. In a post hoc sensitivity analysis excluding microbiological persistence from the composite outcome (147 events), neither co-production (aOR 0.53, 95% CI 0.27-1.03) nor biofilm positivity (aOR 0.76, 95% CI 0.51-1.13) was statistically significant. Conclusion Host severity and remediable urinary-tract factors showed more consistent associations with treatment failure than the microbiological phenotypes evaluated. The inverse biofilm pattern is observational, sensitive to outcome definition, and should not be interpreted as protective or causal; the internally corrected model had limited discrimination and was not developed as a prediction tool.