Accurate estimation of energy expenditure is a key element in determining the relationships between aspects of human behavior, physical activity, and overall health.1,2 Although wearable devices for estimating energy expenditure are becoming increasingly popular, there is little evidence regarding their validity.3,4 This study was performed to examine the validity of total energy expenditure estimates made by several wearable devices compared with gold standard measurements for a standardized day (metabolic chamber method) and free-living days (doubly labeled water [DLW] method).
All protocols were reviewed and approved by the ethics review board of the National Institute of Health and Nutrition, Tokyo, Japan. Written informed consent was obtained from all participants, who were compenstated for their participation. Participants were 19 healthy adults (9 men and 10 women) aged 21 to 50 years who were not obese and had no problems performing regular daily activities. Total energy expenditure was measured using 12 wearable devices. Eight were consumer devices selected because the manufacturers claim that they measure total energy expenditure and they are popular according to Japanese sales rankings (JAWBONE UP24, Fitbit Flex, Misfit Shine, EPSON PULSENCE PS-100, Garmin Vivofit, TANITA AM-160, OMRON CaloriScan HJA-403C, and Withings Pulse O2). The remaining 4 devices are validated for use in research (OMRON Active style Pro HJA-350IT, Panasonic Actimarker EW4800, SUZUKEN Lifecorder EX, and ActiGraph GT3X). All 12 devices were worn simultaneously at randomly assigned positions on the wrist, chest, or waist as appropriate to minimize possible bias owing to placement (Figure 1).
Detailed procedures for total energy expenditure measurement using the metabolic chamber and DLW methods have been described.5,6 For the metabolic chamber experiment, participants visited the laboratory at 7:30 am after an overnight fast. After setting and applying all devices, participants entered the metabolic chamber from 9:00 am to 9:00 am the following day. They completed 24-hour indirect calorimetry under a standardized protocol simulating normal daily life, which included 3 meals, desk work, watching TV, housework, treadmill walking, and sleeping.
For the DLW experiment, DLW dosing was performed in the laboratory after collection of baseline urine samples. Each participant collected urine in airtight containers on 8 days spread over a 15-day free-living period. Concurrently, the participants wore all 12 devices while awake except when bathing, special activities in which wearing the devices would be difficult, or when charging the battery. The 5 wearable devices worn on the wrist were worn while sleeping. After 15 days, urine samples and all wearable devices were recovered to analyze mean daily total energy expenditure during 15 free-living days.
Total mean (SD) energy expenditure measured by the metabolic chamber (2093  kcal/d) was significantly lower than that measured by the DLW method (2314  kcal/d; P < .05). For both gold standard measures, Spearman rank correlation coefficients for most devices were greater than 0.8. Measurements from the 12 devices for a standardized day ranged from 278 kcal/d lower to 204 kcal/d higher than the metabolic chamber. Compared with the DLW measure for free-living days, estimates from the 12 devices ranged from 590 kcal/d lower to 69 kcal/d lower (Figure 2).
The wearable devices that we tested were able to rank daily total energy expenditure between individuals, but absolute values differed widely among devices and varied significantly from the gold standard measures. Furthermore, all wearable devices underestimated total energy expenditure under free-living conditions. The large variance may be associated with epoch lengths and posture detection (sitting or standing), and underestimation might be due to periods of not wearing the devices during bathing and battery charge.1,5 Our study was limited by the small sample size and including only nonobese, healthy participants. Although further studies are required, the findings presented herein suggest that most wearable devices do not produce a valid measure of total energy expenditure.
Corresponding Author: Motohiko Miyachi, PhD, Department of Health Promotion and Exercise, National Institute of Health and Nutrition, NIBIOHN. 1-23-1 Toyama, Shinjuku, Tokyo 162-8636, Japan (firstname.lastname@example.org).
Published Online: March 21, 2016. doi:10.1001/jamainternmed.2016.0152.
Author Contributions: Drs Murakami and Miyachi had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Kawakami, Nakae, Nakata, Ishikawa-Takata, Tanaka, Miyachi.
Acquisition, analysis, or interpretation of data: Murakami,Nakae, Nakata, Tanaka, Miyachi.
Drafting of the manuscript: Kawakami, Nakae, Nakata, Miyachi.
Critical revision of the manuscript for important intellectual content: Nakae, Nakata, Ishikawa-Takata, Tanaka, Miyachi.
Statistical analysis: Murakami, Nakae, Nakata, Tanaka, Miyachi.
Obtaining funding: Miyachi.
Administrative, technical, or material support: Kawakami, Nakae, Ishikawa-Takata, Tanaka, Miyachi.
Study supervision: Nakae, Nakata, Tanaka, Miyachi.
Conflict of Interest Disclosures: The authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Tanaka reported receiving research funding from Omron Health Care Inc. No other disclosures were reported.
Funding/Support: This research was supported by the Practical Research Project for Lifestyle-related Diseases including Cardiovascular Diseases and Diabetes Mellitus from the Japan Agency for Medical Research and Development (AMED).
Role of the Funder/Sponsor: AMED had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
NM. Assessment of physical activity and energy expenditure: an overview of objective measures. Front Nutr
. 2014;1:5.PubMedGoogle ScholarCrossref
et al. Daily activity energy expenditure and mortality among older adults. JAMA
. 2006;296(2):171-179.PubMedGoogle ScholarCrossref
GJ. Validity of consumer-based physical activity monitors. Med Sci Sports Exerc
. 2014;46(9):1840-1848.PubMedGoogle ScholarCrossref
KR. Daily physical activity assessment with accelerometers: new insights and validation studies. Obes Rev
. 2013;14(6):451-462.PubMedGoogle ScholarCrossref
et al. Effects of intermittent physical activity on fat utilization over a whole day. Med Sci Sports Exerc
. 2013;45(7):1410-1418.PubMedGoogle ScholarCrossref
et al. The relationship of body composition to daily physical activity in free-living Japanese adult men. Br J Nutr
. 2014;111(1):182-188.PubMedGoogle ScholarCrossref