Multidrug-Resistant Urinary Tract Infections in Resource-Limited Settings: Resistance Trends and Predictive Modeling
Description
A retrospective cross-sectional study was conducted on 317 culture-confirmed UTI cases at Al-Rafa Laboratory, Ibb University, Yemen (March 2023–November 2024). Bacterial identification and antimicrobial susceptibility testing were performed using standard protocols. MDR-UTI was defined as resistance to three or more antibiotic classes. Multivariable logistic regression identified independent MDR risk factors, and model performance was assessed using receiver operating characteristic (ROC) analysis.
Files
Steps to reproduce
Urine samples were collected and processed following standardized microbiological protocols to ensure consistency and accuracy. Specimens were inoculated onto appropriate agar media (HiMedia Laboratories Pvt. Limited, India) and incubated aerobically at 37°C for 24 hours. Bacterial identification was conducted using established laboratory methods. Antimicrobial susceptibility testing employed the disk diffusion method according to Clinical and Laboratory Standards Institute (CLSI) guidelines. Quality control was maintained by including reference strains from the American Type Culture Collection (ATCC): Escherichia coli ATCC 25922, Staphylococcus aureus ATCC 25923, and Pseudomonas aeruginosa ATCC 25853. The Multiple Antibiotic Resistance (MAR) index was calculated as the ratio of the number of antibiotics to which an isolate was resistant (a) to the total number of antibiotics tested (b): MAR Index = a/b [9, 13]. Outcome Measures The primary outcome was multidrug-resistant urinary tract infection (MDR-UTI), defined as resistance to three or more antibiotic classes, including β-lactams, fluoroquinolones, and aminoglycosides. Secondary outcomes included pathogen-specific resistance rates to empiric antibiotics and clinical failure, defined as persistence of symptoms with a positive follow-up culture within 30 days. Handling of Missing Data Missing data were minimal (<5%) and handled by complete case analysis. No imputation methods were applied given the low proportion and random distribution of missing values. Sample Size The multivariable risk prediction model was developed following TRIPOD guidelines, emphasizing an adequate events-per-variable (EPV) ratio for model robustness [14]. Among 317 patients with culture-confirmed UTIs, 62 (19.6%) had multidrug-resistant infections. With four predictors included (prior antibiotic use, healthcare contact, diabetes, catheterization), the EPV was 15.5, exceeding the recommended minimum of 10. This sample size provides sufficient power (>90%) to detect meaningful associations (OR ≥1.8) for moderately prevalent predictors and supports internal validation via 1,000 bootstrap resamples, which indicated minimal overfitting (shrinkage factor = 0.95). Although limited to a single center, the sample size aligns with prior regional studies such as Badulla et al. and meets methodological standards for prediction modeling in low-resource settings [9].
Institutions
- Ibb University