AF risk prediction in COPD achieved moderate discrimination across several modeling approaches, and the developed models showed higher discrimination than published AF risk scores (CHARGE-AF and HARMS2-AF) in several analyses, particularly in temporal validation, supporting further evaluation of individualized AF risk stratification in patients with COPD.
Among 8385 UK Biobank COPD patients without prior AF, 1402 developed AF over a median follow-up of 14.17 years.
All four modeling approaches achieved moderate discrimination in internal validation, with Elastic Net slightly highest.
Discrimination was lower in temporal validation than in internal validation, with DeepSurv highest in that set.
In temporal validation, the Elastic Net model discriminated better than the established AF scores CHARGE-AF and HARMS2-AF.
Decision-curve analysis suggested potential clinical utility of the models across selected 5-year risk thresholds.
The Elastic Net model was implemented as an interactive web application for individualized AF risk estimation.
This study built and tested tools to predict which people with chronic obstructive pulmonary disease (COPD) are likely to develop atrial fibrillation (AF), an irregular heart rhythm. The researchers used data from 8385 people with COPD in the UK Biobank who did not have AF at the start. Over a median of about 14 years, 1402 of them developed AF. They built four types of prediction models using clinical information, blood tests, treatments, and genetic data. All four models predicted AF risk moderately well. Performance was somewhat lower when tested on a later group of participants (recruited in 2010) than when tested internally. In that later group, the best-performing simple model (Elastic Net) did better than two widely used AF risk scores, CHARGE-AF and HARMS2-AF, which were designed mainly for the general population. The authors also built an online tool for estimating individual risk. This research suggests that a COPD-specific approach may identify AF risk better than general-population scores. However, discrimination was only moderate, the data came from one cohort, and the authors say further evaluation is needed before the tool could be used in routine care.
Xiao X, Qin Y, Peng C, Li D, Kuang M, Gao Y, et al.. (2026). Development and Validation of a Risk Prediction Model for Atrial Fibrillation in Patients with Chronic Obstructive Pulmonary Disease.. International journal of chronic obstructive pulmonary disease. https://doi.org/10.2147/COPD.S641195