The Fitbit Charge HR systematically underestimated heart rate by an average of 5.96 bpm compared to the Polar H6 monitor under free-living conditions, with larger errors during moderate to vigorous physical activity.
Observational (n=10)
No
Does the Fitbit Charge HR activity tracker accurately measure heart rate and detect moderate to vigorous physical activity compared to the Polar H6 monitor in healthy volunteers?
Under free-living conditions, the Fitbit Charge HR systematically underestimates heart rate, especially during moderate to vigorous physical activity, indicating it may not be sufficiently accurate for clinical exercise prescription.
Mean Difference: -5.96 (95% CI -6.33–-5.6)
p-value: p=<0.001
BACKGROUND: Many modern smart watches and activity trackers feature an optical sensor that estimates the wearer's heart rate. Recent studies have evaluated the performance of these consumer devices in the laboratory. OBJECTIVE: The objective of our study was to examine the accuracy and sensitivity of a common wrist-worn tracker device in measuring heart rates and detecting 1-min bouts of moderate to vigorous physical activity (MVPA) under free-living conditions. METHODS: Ten healthy volunteers were recruited from a large university in Singapore to participate in a limited field test, followed by a month of continuous data collection. During the field test, each participant would wear one Fitbit Charge HR activity tracker and one Polar H6 heart rate monitor. Fitbit measures were accessed at 1-min intervals, while Polar readings were available for 10-s intervals. We derived intraclass correlation coefficients (ICCs) for individual participants comparing heart rate estimates. We applied Centers for Disease Control and Prevention heart rate zone cut-offs to ascertain the sensitivity and specificity of Fitbit in identifying 1-min epochs falling into MVPA heart rate zone. RESULTS: We collected paired heart rate data for 2509 1-min epochs in 10 individuals under free-living conditions of 3 to 6 hours. The overall ICC comparing 1-min Fitbit measures with average 10-s Polar H6 measures for the same epoch was .83 (95% CI .63-.91). On average, the Fitbit tracker underestimated heart rate measures by -5.96 bpm (standard error, SE=0.18). At the low intensity heart rate zone, the underestimate was smaller at -4.22 bpm (SE=0.15). This underestimate grew to -16.2 bpm (SE=0.74) in the MVPA heart rate zone. Fitbit devices detected 52.9% (192/363) of MVPA heart rate zone epochs correctly. Positive and negative predictive values were 86.1% (192/223) and 92.52% (2115/2286), respectively. During subsequent 1 month of continuous data collection (270 person-days), only 3.9% of 1-min epochs could be categorized as MVPA according to heart rate zones. This measure was affected by decreasing wear time and adherence over the period of follow-up. CONCLUSIONS: Under free-living conditions, Fitbit trackers are affected by significant systematic errors. Improvements in tracker accuracy and sensitivity when measuring MVPA are required before they can be considered for use in the context of exercise prescription to promote better health.
Gorny et al. (Fri,) conducted a observational in Healthy volunteers (n=10). Fitbit Charge HR vs. Polar H6 heart rate monitor was evaluated on Difference in heart rate measurement (bpm) (MD -5.96 bpm, 95% CI -6.33 to -5.60, p=<0.001). The Fitbit Charge HR systematically underestimated heart rate by an average of 5.96 bpm compared to the Polar H6 monitor under free-living conditions, with larger errors during moderate to vigorous physical activity.