Significance of Prior Culture History for Predicting Urinary Tract Infection Caused by Multi-drug Resistant Enterobacteriaceae

Abstract

BackgroundExtended-spectrum β-lactamase (ESBL) -producing E. coli, Klebsiella spp., and Proteus spp. (EKP), that cause urinary tract infections (UTI) are resistant to first-line therapies (e.g., ceftriaxone). Prediction of UTI caused by ESBL-producing organisms is important for selection of empirical therapy. The objective was to develop a prediction model to identify UTI caused by ceftriaxone (CRO)-resistant EKP and compare the model to other commonly cited predictive models (Tumbarello M et al. AAC Jul 2011; Johnson SW et al. ICHE Apr 2013).MethodsA single-center, matched, case–control of Veterans Affairs (VA) outpatients with a positive (≥10^4 CFU/mL) urine culture was conducted. Patients were excluded if they had no UTI diagnosis or documented symptoms, age <18, transfer from another hospital, or a significant urine culture result. Cases were defined as any patient with a CRO-resistant EKP; controls were matched 4:1 to cases based on incident density (≤ 30 days) by random selection. Logistic regression and receiver operator curves were used to develop and assess models.ResultsOne hundred subjects were included in the analysis. Demographics were similar except for age [Case 73.5 years (13.7); Control 64.5 years (15.2); P = 0.02] and history of CRO-resistant EKP in last 6 months (Case 40%; Control 0%; P < 0.01). Predictor variables in the final model (Likelihood Ratio 44.2, P < 0.01) included history of CRO-resistant EKP in last 6 months (131.5, 12.2–18308.0), cephalosporin use in past 60 days (12.7, 1.9–94.5), residence in a skilled nursing or assisted living facility (8.0, 1.6–40.5), and hospitalization in last 6 months (OR 3.0, 95% CI 0.7–12.5). In the VA population, the other models predicted significantly although less accurately (Figure 1).ConclusionPrior cephalosporin use, hospitalization, and residence were important predictors of UTI caused by CRO-resistant EKP; however, prior history of CRO-resistant EKP was the most important predictor. A Model that included prior culture results predicted CRO-resistant UTIs better than other commonly cited models that do not contain prior ESBL history. Prior culture data should be considered when selecting empirical antibiotics for UTI. Validation in a larger cohort is warranted.Disclosures All authors: No reported disclosures.