Improving risk prediction for death, stroke and bleeding in Asian patients with atrial fibrillation

Area of Science:

• Cardiology
• Epidemiology
• Health Outcomes Research

Background:

• Atrial fibrillation (AF) management relies on risk stratification for adverse events.
• Existing risk scores like CHA 2 DS 2 -VASc and HAS-BLED have limitations in Asian populations.
• The COOL-AF registry provides a platform for developing and validating new predictive models.

Purpose of the Study:

• To compare the predictive performance of the new COOL-AF registry model against established risk scores (GARFIELD Refitted, CHA 2 DS 2 -VASc, HAS-BLED).
• To evaluate the models for predicting all-cause death, ischaemic stroke/systemic embolism (SSE), and major bleeding.
• To assess the accuracy and calibration of the COOL-AF models in Asian patients with AF.

Main Methods:

• Utilized data from the nationwide COOL-AF registry, including 3405 patients with non-valvular AF in Thailand (2014-2017).
• Developed predictive models using multivariable Cox-proportional Hazard models.
• Evaluated model performance using C-statistics, calibration plots, and decision curve analysis (DCA), with internal validation via bootstrapping.

Main Results:

• COOL-AF models demonstrated good predictive ability with C-statistics of 0.727 for mortality, 0.708 for SSE, and 0.706 for major bleeding.
• Excellent calibration was observed, with slopes between 0.94-0.99, indicating strong agreement between predicted and observed outcomes.
• The COOL-AF models generally outperformed the GARFIELD Refitted, CHA 2 DS 2 -VASc, and HAS-BLED models in predictive accuracy.

Conclusions:

• The COOL-AF predictive models exhibit strong predictive capabilities for key adverse events in Asian AF patients.
• The COOL-AF model for all-cause mortality proved superior to the GARFIELD Refitted and CHA 2 DS 2 -VASc models.
• These findings support the utility of the COOL-AF model for enhanced risk stratification in this population.