ABSTRACT
BACKGROUND: Urinary tract infections (UTI) due to multidrug-resistant bacteria are a frequent reason for visiting the emergency department (ED). OBJECTIVES: The aim of this study was to evaluate the applicability of a predictive model of infection by multidrug-resistant microorganisms in UTIs treated in an ED. METHODS: This is a retrospective observational study. Adult patients admitted to an ED with a diagnosis of UTI and positive urine culture were included. The main objective was to evaluate the area under the curve of the receiver operating characteristic (AUC-ROC), the scale proposed by González-del-Castillo, considering infection by a resistant pathogen as the dependent variable and the scale score of the predictive model used as the independent variable. RESULTS: The study included 414 patients with UTIs, 125 (30.2%) of which were caused by multidrug-resistant microorganisms. A total of 38.4% of patients were treated with antibiotics during the previous 3 months and a multidrug-resistant pathogen was isolated from 10.4% of the total during the previous 6 months. The AUC-ROC of the scale for predicting UTIs due to multidrug-resistant microorganisms was 0.79 (95% confidence interval 0.76-0.83), the optimal cut-off point being 9 points, with a sensitivity of 76.8% and a specificity of 71.6%. CONCLUSIONS: The use of the predictive model evaluated is a useful tool in real clinical practice to improve the success of empirical treatment of patients presenting to the ED with a diagnosis of UTI and positive urine culture pending identification.
