Accuracy of Heart-Rate-Recovery Parameters Assessed From a Wrist-Worn Photoplethysmography Monitor (Polar Unite)

Abstract

Purpose: The accuracy of heart rate (HR) measured with a wrist-worn photoplethysmography (PPG) monitor is altered during rest-exercise and exercise-rest transitions, which questions the validity of postexercise HR-recovery (HRR) parameters estimated from this device.

Methods: Thirty participants (50% female) randomly performed two 13-minute sequences (3' rest, 5' submaximal-intensity exercise, and 5' passive recovery) on treadmill and bicycle ergometers. HR was measured concomitantly with a 10-lead electrocardiogram (ECG) and a wrist-worn PPG monitor (Polar Unite). HRR was assessed by calculating Δ60 (the difference between HR during exercise and HR 60 s after exercise cessation) and by fitting HRR data into a monoexponential model.

Results: By focusing on Δ60 and τ (the time constant of the monoexponential curve), levels of association (r) of the Unite versus the 10-lead ECG were high to very high (.73 < r < .93), and coefficients of variation were >20% (in absolute value), except for Δ60 in the bicycle ergometer condition (11.7%). In 97% of cases, the decrease in HR after exercise appeared later with the Unite. By adjusting the time window used for the analysis according to this time lag, coefficients of variation of Δ60 decreased below 10% in the bicycle ergometer condition.

Conclusions: If a wrist-worn PPG monitor is used to assess HRR, we recommend performing the submaximal-intensity exercise on a bicycle ergometer and focusing on Δ60. Furthermore, to obtain a more accurate Δ60, the time lag between the end of the exercise and the effective decrease in HR should also be considered before the calculation.

Keywords: ECG signal; PPG signal; interchangeability; level of association.