Abstract Body (Do not enter title and authors here): Background
Atrial fibrillation (AF) is associated with sudden cardiac death (SCD). It is known that AF is associated with multiple traditional risk factors of SCD, such as coronary artery disease (CAD), history of myocardial infarction (MI), left ventricular hypertrophy (LVH) and low left ventricular ejection fraction (LVEF). However, AF per se could predispose SCD not mediated by these cardiovascular risk factors, and predictors of SCD in AF patients without these apparent SCD risk factors are unknown.

Methods
Using the NTUH iMED database, AF patients without apparent SCD risk factors were longitudinal followed up from Jan. 2010 to Dec. 2021. The primary outcome was implantation of implantable cardioverter defibrillator (ICD) due to SCD. Cox regression was used to identify the predictors of SCD.

Results
A total of 11250 AF patients without cardiovascular SCD risk factors and LVEF ≥55 % were longitudinally followed up. We first found CHA2DS2-VASc score did not predict SCD event. We found male gender (hazard ratio [HR ] 1.71 [1.05-2.78 ], p =0.032), hyperthyroidism (HR 3.50 [1.40-8.77 ], p=0.007）and paroxysmal AF (PAF)(HR 2.86 [1.63-5.04 ], p<0.001）were associated with higher SCD risks. We created a AFSCD score using these parameters. We found SCD risk incrementally increased with increasing AFSCD score (p <0.001 for trend). Patients with highest score (5 points) had more than 3-fold SCD risk compared with those with 0-point score (1.18 % person-year vs 0.36 %, p<0.001). The C statistic of AFSCD score was significantly higher than CHA2DS2-VASc score (0.64 [0.57-0.70] vs 0.40 [0.33-0.48], p <0.01).

Conclusions
Male gender, hyperthyroid AF and PAF were predictors of incident SCD in AF patients and AFSCD scoring scheme based on these risk factors could possibly be implemented in clinical practice.