# The Clinical Implications of Inappropriate Therapy in Community-Onset Urinary Tract Infections and the Development of a Bayesian Hierarchical Weighted-Incidence Syndromic Combination Antibiogram

## TL;DR

This study introduces a new Bayesian model to improve antibiotic choices for urinary tract infections by considering patient-specific factors and local resistance patterns.

## Contribution

The novel Bayesian hierarchical model integrates patient-specific factors and local resistance data to optimize empirical antibiotic therapy for UTIs.

## Key findings

- Inappropriate antibiotic treatment was linked to worse clinical outcomes like extended hospital stays and mortality.
- Combination therapies showed better coverage, especially in multidrug-resistant cases and older adults.
- The Bayesian model improved estimation accuracy, particularly for rare pathogen-antibiotic interactions in high-resistance settings.

## Abstract

Background/Objectives: The rise in multidrug-resistant pathogens complicates UTI management, particularly in empirical therapy. This study aimed to develop and describe a Bayesian hierarchical weighted-incidence syndromic combination antibiogram (WISCA) model to optimize antibiotic selection for adult patients with community-onset UTIs. Methods: A retrospective study was conducted using a Bayesian hierarchical model. Data from microbiology laboratory records and medical databases were analyzed, focusing on age, prior antibiotic exposure, and clinical characteristics. Clinical outcomes, including extended hospital stays and in-hospital mortality, were evaluated before WISCA model development. Unlike traditional antibiograms, a WISCA integrates patient-specific factors to improve antimicrobial coverage estimations. A total of 11 monotherapies and 18 combination therapies were tested against 15 pathogens, with posterior coverage probabilities and 95% highest density intervals (HDIs) used to assess coverage. Results: Inappropriate final antibiotic treatment was associated with worse outcomes. The Bayesian framework improved estimations, particularly for rare pathogen–antibiotic interactions, increasing model applicability in high-resistance settings. Combination regimens showed superior coverage, especially in MDR cases and older adults. Conclusions: This study employed a comprehensive methodological approach for WISCA development, enhancing empirical antibiotic selection by incorporating local resistance data and patient-specific factors in a middle-income Latin American country with a high antimicrobial resistance profile. These findings provide a foundation for future clinical applications and antimicrobial stewardship strategies in high-resistance environments.