Comparison of Time-Domain Short-Term Heart Interval Variability Analysis Using a Wrist-Worn Heart Rate Monitor and the Conventional Electrocardiogram

Background: Wrist-worn heart rate monitors have not been extensively validated for heart rate variability analysis. The purpose of this study was to compare time-domain variability of heart interval series (R-Ri) recorded by the Polar S810 monitor (Polar Electro Oy, Kempele, Finland) and the conventional electrocardiogram (ECG).
Methods: Agreement was verified between variability indices of 5-minute R-Ri simultaneously recorded by both devices and processed by unique software, from 33 subjects aged 18 to 42 years, normal or with different clinical conditions, in rest supine and active standing. ECG minus Polar differences were quantified by the Bland-Altman analysis, and tested by the one-sample t-test or Wilcoxon test.
Results: In the supine position, the Polar overestimates (P < 0.0001) the absolute and percentage mean or median of the number (−2.00; −0.49%) and mean of R-Ri (–1.85 ms; –0.20%) and pNN50 (−2.20%; −8.68%), and underestimates the standard deviation (SDNN) (0.32 ms; 0.59%) (P = 0.08; P = 0.02) and root mean square successive difference (RMSSD) (0.90 ms; 1.56%) (P = 0.0008; P < 0.0001). The coefficient of variation (CV) showed null difference. On standing, differences were overestimated for the number (−2.61 intervals; −0.64%) and mean of R-Ri (−0.70 ms; −0.09%), and underestimated for rMSSD (1.70 ms; 10.84%) (P < 0.0001 to < 0.02). The SDNN, CV, and pNN50 indices did not show differences (P = 0.12 to 0.73).
Conclusions: The Polar S810 monitor was feasible and reliable for recording short-term R-R interval series, showing excellent agreement with the ECG in providing the time-domain indexes of heart interval variability with differences functionally not relevant. The CV showed the higher agreement in both postures, and the SDNN and pNN50 in the standing posture.