Risk prediction for new-onset atrial fibrillation using the Minnesota code electrocardiography classification system

Area of Science:

• Occupational Health
• Cardiology
• Predictive Modeling

Background:

• Limited risk models exist for predicting atrial fibrillation (AF) in occupational settings.
• Annual health checkups (HC) with electrocardiogram (ECG) data offer potential for AF risk assessment.

Purpose of the Study:

• Develop and validate risk prediction models for new-onset AF in a large worker population.
• Evaluate the utility of Minnesota code information from ECGs in AF risk prediction.

Main Methods:

• Retrospective analysis of 56,288 workers (≥40 years, no prior AF) with annual HC data (2013-2016).
• Cox models developed prediction models with and without Minnesota code information.
• External validation performed using the Hisayama cohort.

Main Results:

• During a 3.0-year follow-up, 209 workers (0.37%) developed AF.
• Key predictors included age, sex, waist circumference, blood pressure, LDL cholesterol, and γ-GTP.
• Models with Minnesota code improved AUC (0.82 to 0.84 derivation, 0.78 to 0.79 validation).
• Net Reclassification Improvement (NRI) and Integrated Discrimination Improvement (IDI) showed significant enhancements.

Conclusions:

• A practical risk model incorporating Minnesota code information was developed for predicting new-onset AF in workers.
• The model demonstrated utility and was validated in both internal and external cohorts.
• While AUC improvement was modest, the model offers valuable predictive capabilities for AF.