Your tracker's calorie burn figure is off by more than 30%

Wrist trackers now sit on millions of arms, and people make daily decisions based on what they display: whether to take another walk, whether a workout counted, whether they have earned a bigger dinner. Yet the market is crowded with devices making similar promises, and the evidence on how well they actually measure activity has been scattered across dozens of separate validation studies. This review set out to unify that evidence, asking two practical questions: which devices give accurate readings, and which ones do people actually keep wearing?

Researchers searched four major medical databases for studies measuring physical activity in the general population using wrist-wearable trackers. They screened the results and included 65 articles, recording each study's setting, population, outcomes and risk of bias. Accuracy was assessed for 14 different outcomes, grouped into step counts and other activity counts, time spent active, intensity of activity including energy expenditure, heart rate, distance and speed. Acceptability was judged mostly through data availability and how long people kept the device on. The studies varied so much that their results could not be pooled statistically.

For step counts, the Fitbit Charge and Fitbit Charge HR kept mean error below 25% across 20 studies, the most consistent performance in the review. For heart rate, the Apple Watch kept error below 10% in 2 studies. Energy expenditure was the clear failure: error exceeded 30% for every brand assessed, which the authors describe as poor accuracy across devices. On acceptability, data availability was at least 75% for the Fitbit Charge HR, Fitbit Flex 2 and Garmin Vivofit, and wearing time reached 89% for both the GENEActiv and Nike FuelBand.

This review pulls together a fragmented validation literature and lands on a consistent pattern: counting steps and reading heart rate are solved problems on the better devices, while estimating calories burned is not solved on any of them. Because the underlying studies used such different methods, the authors could not rank devices formally, and they call for more standardised testing. It remains an open question whether newer generations of these devices have closed the energy expenditure gap, since the review covers models tested up to its search date.

If you use a wrist tracker, this is a reason to treat its different numbers with different levels of trust. Step counts and heart rate readings from the better-performing devices tracked reality reasonably well, so trends in those figures are worth watching. The calorie burn figure, by contrast, missed by more than 30% on every brand tested, so it is better read as a rough indication of relative effort than as a number to build eating decisions around.