Next generation activity trackers are wrist-worn trackers measuring heart rate with light-emitting diode (LED) reflection from the skin, reflecting tiny changes in skin blood volume. It is easy to understand their convenience, but their accuracy is questionable.

We’ve discussed a variety of health apps and wearables used for heart rate monitoring. The New York Attorney General recently fined several developers for heart rate monitoring apps that didn’t have the evidence to back up accuracy claims. And despite some well publicized uses of wearables like Fitbit for heart rate monitoring, the data supporting that use has been lacking.

Cadmus-Bertram et al tested four trackers on the market (Fitbit Surge, Basis Peak, Fitbit Charge, and Mio Fuse) against the gold standard ECG at rest and at 65% of maximum heart rate during exercise on the treadmill. Participants were healthy adults aged 30 to 65 years without cardiovascular conditions (N=40). Two trackers were randomly placed on right versus left wrist or proximal versus distal location on the wrist. Heart rates were measured at 1-minute intervals for 10 minutes at rest and during exercise.

Fitbit Surge (Fitbit) had the highest agreement with the ECG results (range 5.1 to 4.5 beats/min), followed by the Mio Fuse (Mio Global; 7.8 to 9.9 beats/min), Fitbit Charge (Fitbit; range 10.5 to 9.2 beats/min), and Basis Peak (Basis; range 17.1 to 22.6 beats/min).

The reproducibility of the results was measured by the repeatability coefficient, where a smaller value is more beneficial. For ECG, the repeatability coefficient was 5.3 beats/min at rest and 9.1 beats/min during exercise. The repeatability coefficient was lowest for the Fitbit Surge (4.2 beats/min at rest; 20.6 beats/min during exercise), followed by the Fitbit Charge (10.9 beats/min at rest; 21.6 beats/min during exercise), Mio Fuse (10.9 beats/min at rest; 23.7 beats/min during exercise), and Basis Peak (19.3 beats/min at rest; 20.2 beats/min during exercise). The location of the tracker on the wrist did not seem to affect measurements.

Although the level of agreement with ECG results and the repeatability coefficient was comparable to ECG at rest for some devices such as the Fitbit Surge, the accuracy and precision of the measurements seems to vary a lot during exercise. This is similar to another study we covered from the Cleveland Clinic last year in which 50 patients were enrolled. Both studies are limited by enrollment of generally young and healthy patients.

While these wrist devices may provide a sufficient approximation for assessing daily activity and exercise, more work is needed before using the data clinically. As a side note, safety of using these trackers should also be tested carefully, as we can see with the recent recall of Basis Peak due to overheating issues.