Prediction Score for Identification of ESBL Producers in Urinary Infections Overestimates Risk in...

Area of Science:

• Infectious Diseases
• Clinical Microbiology
• Antimicrobial Resistance

Background:

• Urinary tract infections (UTIs) caused by extended-spectrum beta-lactamase (ESBL) Enterobacterales are a growing concern.
• Developing effective strategies to identify patients at risk for ESBL-producing infections is crucial for appropriate antibiotic selection.
• Local validation of risk assessment tools is necessary, especially in regions with high ESBL prevalence.

Purpose of the Study:

• To validate the Tumbarello scoring system for identifying ESBL producers in hospitalized UTI patients in Colombia.
• To assess the diagnostic performance of a modified Tumbarello score in a high-prevalence setting for ESBL-producing *Escherichia coli*.

Main Methods:

• Retrospective cohort study of 817 patients hospitalized with pyelonephritis in Bogotá, Colombia (2013-2020).
• Evaluation of a modified Tumbarello score (excluding inter-facility transfer) to predict ESBL presence.
• Analysis of diagnostic performance using Area Under the Curve (AUC) for ESBL and cephalosporin resistance.

Main Results:

• The modified Tumbarello score (cut-off ≥6) demonstrated low sensitivity (18%) and moderate specificity (83%).
• The AUC for predicting ESBL presence was 0.47, indicating poor predictive ability.
• The score failed to efficiently predict resistance to third-generation (AUC = 0.49) and second-generation (AUC = 0.51) cephalosporins.

Conclusions:

• The Tumbarello index is not recommended for managing UTIs in Colombia due to high ESBL *E. coli* prevalence.
• Its use could lead to unnecessary broad-spectrum antibiotic utilization.
• Further research is needed to develop accurate risk assessment tools for ESBL producers in high-prevalence settings.