Background Visit-to-visit heart rate variability (VVV-HR) has been associated with adverse cardiovascular outcomes. We aimed to determine the predictive value of VVV-HR for adverse clinical outcomes in patients with nonvalvular atrial fibrillation (AF).

Methods We used data from a prospective multicenter AF registry of 27 hospitals in Thailand during 2014 to 2017. After the baseline visit, patients were followed up every 6 months until 3 years. VVV-HR was calculated from the standard deviation of heart rate data from baseline visit and every follow-up visit. VVV-HR was categorized into four groups according to the quartiles. Clinical outcomes were all-cause death, ischemic stroke/systemic embolism (SE), and heart failure. Cox proportional hazard models were used for multivariable analysis.

Results There were a total of 3,174 patients (mean age: 67.7 years; 41.8% female). The incidence rates of all-cause death, ischemic stroke/SE, and heart failure were 3.10 (2.74–3.49), 1.42 (1.18–1.69), and 2.09 (1.80–2.42) per 100 person-years respectively. The average heart rate was 77.8 ± 11.0 bpm and the average of standard deviation of heart rate was 11.0 ± 5.9 bpm. VVV-HR Q4 was an independent predictor of all-cause death, ischemic stroke/SE, and heart failure with adjusted hazard ratios of 1.45 (95% confidence interval: 1.07–1.98), 2.02 (1.24–3.29), and 2.63 (1.75–3.96), respectively. VVV-HR still remained a significant predictor of clinical outcomes when analyzed based on coefficient of variation and variability independent of mean.

Conclusion VVV-HR is an independent predictor for adverse clinical outcomes in patients with AF. A J-curve appearance was demonstrated for the effect of VVV-HR on all-cause death.