Predictive modeling of multidrug resistance in female genital infections: implications for early urinary tract infection detection

Abstract

Abstract Background Multidrug-resistant (MDR) organisms are increasingly recognized for complicating the treatment of urinary tract infections (UTIs), particularly in females. Traditionally, urine samples are used for UTI diagnosis; however, high vaginal swabs (HVS) have the potential to serve as an early indicator of uropathogens, especially in cases of genital tract colonization. This study explores the predictive value of HVS specimens in identifying MDR organisms and their role in the early detection of UTIs. By analyzing the clinical presentation and organisms isolated, this study aims to enhance diagnostic accuracy and facilitate timely intervention for UTI management. Method We analyzed 824 HVS specimens from female patients aged 0–79 years with suspected genital tract infections over four years. HVS was chosen to detect urogenital pathogens before ascending infections occur. Data on age and clinical signs were collected via structured questionnaires. Specimens underwent culture-based and molecular analyses, including polymerase chain reaction (PCR). Patients with diabetes, hypertension, or immunocompromised states were excluded to minimize confounding. Results Most specimens were from young adults (20–39 years, 75%). Common symptoms included inflammation (51.3%), vaginal discharge (21.2%), and menstrual disorders (11.5%). MDR organisms were detected in 21.8% of cases. Pathogens were isolated in 83.4% of specimens, with Candida albicans (27.1%) and Staphylococcus aureus (26.7%) being the most common isolates. Enterococcus faecalis exhibited the highest MDR rate (40%), while Escherichia coli was significantly associated with MDR (B = 3.220, p < 0.001). E. coli was the strongest predictor of MDR, with an odds ratio of 1.38 ( p < 0.001). The predictive model showed moderate explanatory power (Nagelkerke R² = 0.233), good model fit (Hosmer-Lemeshow test, p = 0.961), and acceptable discriminatory ability (AUC = 0.753, p < 0.001), but had low sensitivity for MDR classification (2.8%). Conclusion Our findings suggest that HVS specimens may serve as early indicators of MDR and possible UTI risk. However, the predictive model’s low sensitivity highlights the need for further refinement to improve clinical utility. Future research should focus on enhancing predictive accuracy for better clinical decision-making.