In-Hospital Mortality Predictors and a Bayesian Weighted-Incidence Antibiogram in Infective Endocarditis: A...

Area of Science:

• Infectious Diseases
• Cardiology
• Health Services Research

Background:

• Infective endocarditis (IE) poses significant mortality risks, especially in middle-income countries with distinct patient demographics and microbial patterns.
• Existing international prognostic scores may lack local applicability.
• Limited availability of syndrome-specific, locally relevant decision aids for empirical therapy.

Purpose of the Study:

• Identify predictors of in-hospital mortality for infective endocarditis.
• Externally evaluate the performance of the RiskE and ICE prognostic scores.
• Develop a Bayesian weighted-incidence syndromic combination antibiogram (WISCA) for IE empirical therapy.

Main Methods:

• Retrospective cohort study of adult IE patients (January 2019-January 2026).
• Model development using maximum-likelihood and Firth penalization with bootstrap correction for predictor identification.
• External validation of RiskE and ICE scores; construction of a Bayesian WISCA stratified by IE type.

Main Results:

• In-hospital mortality was 22.9% in a cohort with high hemodialysis prevalence and right-sided IE.
• Vasopressor-requiring shock and acute heart failure were key mortality predictors (AUC 0.922), outperforming RiskE and ICE.
• Bayesian WISCA identified carbapenem- and anti-MRSA-sparing regimens with adequate empirical coverage (≥80%) for community-acquired IE.

Conclusions:

• A concise, three-variable local model accurately predicts IE mortality in specific populations, surpassing European scores.
• The Bayesian WISCA provides guidance for selecting stewardship-compatible empirical regimens, reducing reliance on broad-spectrum agents.
• Further external validation is recommended for clinical adoption and multicenter application.