Modified Taiwan Atrial Fibrillation Score for the Prediction of Incident Atrial Fibrillation

BACKGROUND: We have proposed the Taiwan AF score consisting of age, male sex, hypertension, heart failure, coronary artery disease, end-stage renal disease, and alcoholism to predict incident atrial fibrillation (AF) in Asian population. We hypothesized that the modified Taiwan AF score (mTaiwan AF score) excluding alcoholism remained useful for predicting new onset AF. METHODS: A total of 7,220,654 subjects aged ≥ 40 years without a past history of cardiac arrhythmia were identified from a national cohort, and 438,930 incident AF occurred during a 16-year follow-up with an incidence of 0.42 per 100 person-years. The mTaiwan AF score ranging between −2 and 14 and its predictive accuracy of incident AF was analyzed. RESULTS: The areas under the receiver operating characteristic curve (AUCs) of the mTaiwan AF scores in predicting AF are 0.861 for 1-year follow-up, 0.829 for 5-year follow-up, 0.795 for 10-year follow-up, and 0.751 for 16-year follow-up. The risk of incident AF increased from 0.05%/year for patients with a score of −2 to 6.98%/year for those having a score of 14. Patients were classified into three groups based on the tertile values of the mTaiwan AF scores—group 1 (score −2-3), group 2 (score 4-9) and group 3 (score 10-14). The annual risks of incident AF were 0.20, 1.33, and 3.36% for group 1, 2, and 3, respectively. Compared to patients in group 1, the hazard ratios of incident AF were 5.79 [95% confidence interval (CI) 3.75-7.75] for group 2 and 8.93 (95% CI 6.47-10.80) for group 3. CONCLUSIONS: We demonstrated that the mTaiwan AF score based on age and clinical comorbidities could be used to predict incident AF in Asian population.