Fitness Trackers Are Only 67% Accurate, New Research Finds

Fitness trackers are among the most widely used health tech devices. But are they as accurate as their manufacturers claim? According to an analysis of scientific literature conducted by the Wellnesspulse research team, fitness trackers offer only moderate accuracy, with the Apple Watch coming out on top.

Manufacturers of fitness trackers claim that the data provided by their devices are highly accurate, allowing users to make informed decisions about their health. However, some users and experts are not convinced, fueling the ongoing debate on wearables' accuracy and potential risks.

To determine the accuracy of data provided by commonly used fitness trackers, three Wellnesspulse researchers analyzed 45 scientific studies and extracted 168 data points. The meta-analysis focuses on three metrics: energy expenditure or calories burned, heart rate, and step count, partly because most fitness trackers measure them.

The analysis reveals that commonly worn fitness trackers are the most accurate for measuring heart rate, providing a strong accuracy of 76.35%. However, they are only moderately accurate for tracking step count and energy expenditure, 68.75% and 56.63%, respectively.

The highest level of accuracy is provided by the Apple Watch, which exhibits 86.31% accuracy for heart rate and 71.02% for energy expenditure. Garmin is the most accurate for tracking step count, as it offers 82.58% accuracy.

Learn more about the methodology here.

Which fitness tracker metrics can you trust, and how much?

Most research is available on five fitness tracker brands — Fitbit, Apple, Garmin, Polar, and TomTom. Google-owned Fitbit is the most researched wearable, accounting for 36.90% of data points among all metrics. It is followed by Apple (14.29 %) and Garmin (13.10 %).

The overall cumulative accuracy of HR, SC, and EE metrics provided by the analyzed fitness trackers is moderate, ranging between 62.09% and 73.53%. On average, the analyzed devices scored 67.40% for accuracy.

Some wearables record steps with only 53% accuracy

Three of the five fitness trackers included in the additional analysis showed a strong accuracy in step counting, while two were only moderately accurate. Garmin leads the list, with 82.58% accuracy, followed by Apple (81.07%) and Fitbit (77.29%). Jawbone and Polar may be the least accurate wearables for counting steps from the list, as they only provide moderate accuracy of 57.91% and 53.21%, respectively.

While many may find the lack of accuracy for this metric surprising, it is noteworthy that wearables rely on wrist movement to count steps, easily mistaking other movements — even made in the sedentary position — for walking.

The inaccuracy of such data may negatively impact the user's health. In a 2023 experiment, individuals who were provided with a deflated step count recorder by their fitness tracker saw their physical activity as less adequate, ate fewer healthy foods, and even had increased blood pressure and heart rate.

Heart rate precision can be as low as 67%

Apple is the most accurate fitness tracker for measuring heart rate, offering 86.31% accuracy, although Fitbit also exhibits a strong accuracy of 73.56%. Of the four devices with enough data to be included in the additional analysis, Garmin and TomTom were the least accurate for tracking heart rate, with only 67.73% and 67.63% accuracy, respectively.

Rather than measuring actual heartbeat, wearables track changes in the blood flow in your wrist using the light sensor (the green blinking light) on the back of the device. Some studies on fitness trackers suggest a significantly reduced accuracy of heart rate measurement in dark-skinned individuals compared to people with lighter skin tones.

Energy expenditure accuracy can be around 50%

When it comes to energy expenditure, Apple is the only fitness tracker providing a strong accuracy of 71.02%. Jawbone, Fitbit, and Polar yield only moderate accuracy, ranging between 50.23% and 65.57%. Compared to other trackers included in the additional analysis, Garmin may be the least accurate wearable for measuring calories burned: although technically it is considered moderately accurate, it shows only 48.05% accuracy.

The low level of accuracy for measuring energy expenditure comes as no surprise. Counting the amount of calories burned is a complex process that considers not only physical activity but also basal functions such as sleeping and breathing. Moreover, fitness trackers estimate energy expenditure without taking into account personal factors like muscle mass. Evidence suggests that wearables may provide less accurate measurement of EE in men, people with greater body mass index, those with darker skin tones, and those engaging in walking.

Growing popularity of fitness trackers

Terms like ‘heart rate zones’ or ‘blood oxygen levels’ are no longer reserved for professional athletes. Thanks to the increasing use of fitness trackers, many healthy living enthusiasts now track these metrics, along with steps taken or calories burned.

About one in five Americans uses a smartwatch or fitness tracker, according to 2020 data. The global market size of these health tech devices is projected to increase more than fourfold between 2024 and 2032, from $62.03 billion to $290.85 billion.

There is growing evidence that wearing fitness trackers may make us move more, bringing us closer to the recommended goal of at least 150 minutes of physical activity a week. For example, a 2022 study found that these devices encourage users to walk up to 40 minutes more daily, resulting in an average 1 kg weight loss over five months.

Major health organizations, such as the British Heart Foundation and the Centers for Disease Control and Prevention, endorse using fitness trackers as a part of chronic disease prevention and management, although they emphasize that wearables cannot replace medical tests.

So, can we rely on fitness trackers?

Commonly used fitness trackers are, on average, 67.40% accurate based on metrics like heart rate, step count, and energy expenditure. Based on this analysis, some wearables may be highly accurate for tracking specific metrics in everyday settings. However, users should interpret all the data they provide with caution — they aren’t medical devices and have many limitations.

Apple appears to be the most accurate fitness tracker from those with enough data to be included in additional analysis, but it still has limitations when it comes to gauging energy expenditure.

People with chronic conditions or those aiming to be more active can benefit from using fitness trackers. However, as the Wellnesspulse analysis reveals, over-relying on them may be dangerous not only due to inaccurate data but also due to the increased risk of health anxiety.

Methodology

Heart rate, step count, and energy expenditure metrics were selected because most fitness tracker studies provide data on them. The majority of commercially available devices measure these basic metrics, while other metrics typically vary between different brands and models.

The accuracy of selected metrics can also be gauged using the ‘gold standard’ measurement devices to which commercially available tracker data can be compared. These metrics also have wider applicability, as they can be used to monitor both daily activities like walking and exercise.

Three researchers: Alifia Fernandes Jokubauskienė, Dominyka Sitavičiūtė, and Linas Černiauskas searched scientific articles with relevant data on commercially available fitness trackers across three scientific archives and repositories: PubMed, Web of Science, and Google Scholar.

The selected studies involved adult individuals of one or both sexes of all health statuses using commercially available fitness trackers in any possible setting, such as the laboratory or real world, and under different physical activity levels.

The study compared commercial fitness tracking device measurements of energy expenditure (EE), heart rate (HR), and step count (SC) against ‘gold standard’ device measurements. These include indirect calorimetry for EE, electrocardiogram for HR, pedometer, or any other manual counting form for SC.

Researchers extracted data on commercially available fitness tracker brands, model names, comparator devices, sample size, and correlation coefficients manually and using AI tools from the original publication text or supplementary material. After the data extraction, one researcher cross-checked and validated the data's accuracy.

The collected raw data was checked manually for input errors. The articles that contained multiple correlation coefficients on the same metric in different physical activity settings were aggregated by averaging all the available coefficients to produce a single value and avoid data duplication.

The data was analyzed using a meta-analysis approach employing a random effects model. Overall correlation coefficient estimates and their 95% confidence intervals were calculated for each metric, and subgroup analysis was done for each brand with more than three comparisons available (additional analysis).

The weight of each metric, SC, HR, and EE, was calculated by dividing the number of available data points for each metric by the total number of data points across all metrics. This allowed the weight to be distributed among metrics based on scientific data availability.

Correlation coefficient estimates and their 95% confidence intervals were also expressed in percentages for clarity and readability.

Correlation coefficients and their percentage values were interpreted as having very weak, weak, moderate, strong, or very strong accuracy.

Relevant data was visualized for clarity and readability purposes.

Limitations of the study

Multiple limitations exist in the meta-analysis of scientific studies on fitness tracker accuracy in measuring heart rate, step count, and energy expenditure.

The studies chosen for the meta-analysis varied in sample sizes (the number of participants), biological sex distributions, and participant health statuses. Additionally, different study designs were applied.

Research papers included in the meta-analysis involved different commercially available fitness tracker device brands and models and ‘gold standard’ devices.

Moreover, there are differences in statistical analysis methods for correlation coefficients. It is noteworthy that correlation coefficients do not entirely reflect the device's accuracy but are approximate statistical metrics that indicate whether there is an association between the measures of the two devices and the strength of the association. It also lacks information on whether the measurements are concordant or discordant.