The accurate assessment of energy requirements is of special importance in athletes to provide adequate dietary energy. When energy intake is not sufficient, negative effects on the athlete's health, performance capacity, and/or body composition may occur (21). Recently, energy expenditure (EE) has also been used for the comparison of training loads between different training interventions (9).
One of the major characteristics of athletes is that they are physically more active than the general population and that, consequently, their EE is higher (3). The average physical activity level (PAL), which represents the ratio of total EE (TEE) and basal metabolic rate, lies between 1.6 and 1.7 in the general population. In athletes, PAL values are typically about 2.0 and higher. Maximal PAL values between 4.0 and 5.0 have been reported for elite endurance athletes (3,31,36). In these athletes, up to 70% of TEE was spent during exercise (37).
Average EE or PAL values have been published for numerous sports and activities, but it should be considered that an athlete's individual EE may substantially deviate from the group level (3). Therefore, the individual assessment of EE is necessary for the adequate evaluation of the athlete's individual diet. For practical purposes, EE assessment methods should be easy, reliable, and accurate (20).
There are several methods available for the assessment of EE (20,24). These include indirect calorimetry (IC), which measures respiratory data (oxygen consumption and carbon dioxide production) and subsequently converts these data to EE (35). IC is the method of choice for acute measurements of EE, but it is mainly limited to laboratory settings or field simulations because metabolic chambers or spirographs are required, which usually restricts the range of motion. For continuous long-term EE assessment in the field, the doubly labeled water (DLW) technique is considered as a gold standard. (27). Because of the limited applicability in the field (IC) or high costs (DLW), both methods are mainly used as reference methods in validation or calibration studies for other methods (28,35). Other objective methods include the use of electronic motion detectors or biological sensors, which can be used for the measurement of EE-related parameters such as acceleration, HR, or body heat loss. Using external calibration and reference data, these sensors can be used to estimate EE (19).
The SenseWear Pro3 Armband (BodyMedia, Pittsburgh, PA) is a portable electronic device that synchronically assesses biaxial accelerometry, body heat loss, and galvanic skin response. EE is automatically calculated on the basis of these values. The armband has been validated in several contexts such as in free-living adults (2,8,32), children (1,7), and clinical patients (6,11,26). In principle, the SenseWear Pro3 Armband seems also well suited for the continuous measurement of EE in athletes. So far, the armband has only been used during low- and moderate-intensity exercises (8,13,16). To our knowledge, the reliability of the armband has not been assessed during high-intensity exercise and in endurance athletes during a prolonged period of exercise.
The aim of our study was to assess the validity of the SenseWear Pro3 Armband in endurance athletes during a regular training period with the use of accepted reference methods (IC, DLW).
For the present study, originally, 15 male endurance athletes were recruited from a local triathlon team and from the student body of our university. All participants were of good health, did not smoke or take any medication, and gave written consent before their participation. The approval by the university's local ethics committee was obtained before the start of the study.
During the study, one participant exhibited a considerable weight loss of 2.8 kg. Because the DLW method is based on constant pools of body water, data from this athlete were omitted. The anthropometric characteristics of the remaining participants are listed in Table 1.
After an overnight fast, the participants reported to our facility between 8:00 and 9:00 a.m. on study day 0. Body weight and body composition were assessed with bioimpedance analysis using a BC 418 MA balance (Tanita, Amsterdam, The Netherlands). The assessment of body weight and body composition was repeated on day 8 of the study, again after an overnight fast between 8:00 and 9:00 a.m. The participants were instructed to be adequately hydrated and emptied their bladder before weighing to minimize changes in body mass.
After the initial weighing (day 0), the athletes were individually instructed about their tasks and were familiarized with the armband. In addition, the isotope solutions for DLW measurement were administered on the morning of day 0.
On day 0, the participants were given the SenseWear Pro3 Armband (BodyMedia) and were instructed to wear the armband on the right arm over the triceps brachii. The volunteers were then asked to start wearing the armband before the midnight of day 0 and to continue until midnight of day 7. The participants were only allowed to remove the armband for showering. Armband values were considered incomplete when the armband was worn <95% of the study time. During the monitoring period (days 1-7), the participants came to our facility twice to perform two separate exercise trials. The trials were performed on two nonconsecutive days. Otherwise, the athletes were free to train as they desired but were asked to record their training using an activity record.
The DLW method was used as a reference method for TEE from day 1 to 7. Each athlete ingested 0.5 g·kg−1 body weight of H218O (10.7% enrichment) and 0.3 g·kg−1 D2O. H218O and D2O were obtained from Euriso-Top GmbH (Saarbrücken, Germany). The application scheme has been reported in detail elsewhere (17). In brief, urine samples obtained on day 0 were used to assess the amount of tracers ingested. Subsequently, morning urine samples were collected between days 1 and 8. On the basis of these samples, TEE was assessed using a multipoint approach. The time shift between SenseWear recording and DLW collection was neglected because lead time of SenseWear (day 1, midnight to morning urine) and lag time of DLW (day 8, midnight to morning urine) were considered equal in length and types of activities performed.
Isotope ratio mass spectrometry analysis for 2H and 18O was performed in triplicate on a PDZ Europa ANCA 20-20 according to standard procedures (23,25). The measurement error was 1.088% (coefficient of variation) for 2H and 0.142% for 18O. TEE was derived from isotope depletion using a multipoint approach and a fixed respiratory quotient of 0.88 (4).
Controlled exercise trials.
The first controlled exercise was an incremental running test on a treadmill (Woodway, Weil am Rhein, Germany). Running speed was started at 2.4 m·s−1, which was maintained for 5 min. Subsequently, running speed was increased every 5 min by 0.4 m·s−1 until individual exhaustion. To account for wind resistance, the inclination of the treadmill was set at 1% (15). Between each step, the treadmill was stopped for 30 s, and capillary blood samples were taken from the earlobe for the analysis of lactic acid concentrations.
The second exercise trial was performed on a stationary bicycle ergometer (SRM, Jülich, Germany). Power output was started at 140 W, and the volunteers were asked to maintain a pedaling frequency between 75 and 105 rpm. Every 5 min, power output was increased by 40 W until individual exhaustion was reached. Blood samples were taken from the earlobe within the last 30 s of each step.
Capillary lactic acid concentrations were measured with an EBIO plus analyzer system (Eppendorf, Hamburg, Germany). On the basis of capillary lactic acid concentrations, individual lactate thresholds (running velocity at a lactate concentration of 2 mmol·L−1 (v2), power output at a lactate concentration of 2 mmol·L−1 (P2)) were calculated according to Heck (10).
During both exercise trials, IC was performed by continuous breath-by-breath analysis with a ZAN 600 spirograph (ZAN, Oberthlba, Germany). Exercise EE (ExEE) was calculated on the basis of the average respiratory gas exchange of minutes 3-5 of each running/bicycling step using the Weir equation (34).
Data and statistical analysis.
Armband data were downloaded from the armband on day 8 and subsequently analyzed with SenseWear Professional software (version 6.1; BodyMedia, Pittsburgh, PA). TEE was computed for each minute according to the manufacturer's algorithm and was expressed in kilocalories per minute. Statistical analysis was performed using R version 2.8.0 (The R Foundation for Statistical Computing, 2008). For the assessment of validity, correlation coefficients were determined using simple linear regression (12). In addition, Bland-Altman analysis was used for the assessment of mean and proportional bias (5). An error of probability of α < 0.05 was assumed for the assessment of statistical significance. If not stated otherwise, values are given as mean ± SD.
Armband recording duration.
On average, the armband was worn 98.4% ± 1.2% of the study time (165 ± 2 h). None of the participants wore the armband <95% of the study period, so data were available for all 14 participants.
There was a positive correlation (r = 0.73, P < 0.01) between TEE assessed with the SenseWear Armband (TEEArmband) and TEE measured with DLW (TEEDLW) (Fig. 1). The difference between TEEDLW (3620 ± 900 kcal·d−1) and TEEArmband (3555 ± 411 kcal·d−1) was 65 ± 665 kcal·d−1. According to the Bland-Altman analysis, the 95% limits of agreement lay between −1368 and 1238 kcal·d−1. In addition, there was a highly significant proportional bias toward overestimating low and underestimating high EE (P < 0.001).
The difference between armband values and DLW data was significantly associated with the participants' performance capacity (Fig. 2). Lactate thresholds assessed during the incremental running and bicycling tests (v2, P2) were negatively correlated with TEEArmband − TEEDLW (v2: r = −0.56, P < 0.05; P2: r = −0.62, P < 0.05).
During both incremental exercise tests, ExEE was simultaneously recorded using the SenseWear Pro3 Armband (ExEEArmband) and IC (ExEEIC) as a reference method. The participants exercised until individual exhaustion and maximal running speeds of 4.2 ± 0.3 m·s−1 and maximal power outputs of 323 ± 26 W were reached.
At 2.4 m·s−1, the difference between ExEEArmband and ExEEIC did not reach statistical significance (P = 0.09), but for all running speeds between 2.8 and 4.8 m·s−1, the armband significantly underestimated ExEE (Fig. 3, top). According to the Bland-Altman analysis, the difference between armband and IC values was 4.5 kcal·min−1 (95% limits of agreement = −11.4 to 2.4 kcal·min−1) and increased significantly with increasing values (P < 0.001). During stationary bicycling, ExEE was also significantly underestimated for all power steps between 140 and 380 W (Fig. 3, bottom). On average, ExEEArmband − ExEEIC was −6.6 kcal·min−1, and the 95% limits of agreement lay between −14.8 and 1.6 kcal·min−1. Again, the degree of underestimation increased with increasing values of ExEE (P < 0.001).
The aim of the present study was to test the validity of the SenseWear Pro3 Armband in endurance athletes. To evaluate TEE and ExEE independently, we used DLW and IC, respectively, as separate reference methods. The use of DLW enabled our participants to follow their habitual lifestyle and training regimen, whereas IC measurements made the evaluation of exercise-specific aspects possible. To our knowledge, this was the first study to systematically assess the performance of the SenseWear Pro3 Armband under these conditions in endurance athletes, particularly during high exercise intensities.
Our results indicate that there is considerable interindividual variation between armband and DLW values. Although TEE was estimated correctly on the group level (mean error < 2% of mean TEEDLW), our data showed a highly significant proportional bias toward underestimating high values of TEE. As a consequence, the 95% limits of agreement between TEEDLW and TEEArmband were unacceptably wide for practical purposes (−1368 to 1238 kcal·d−1, i.e., −37% to 34% of mean TEEDLW). In the only other validation study against DLW, there was also a significant trend toward overestimating low and underestimating high levels of TEE, although the differences were smaller (32). In this study, the 95% limits of agreement lay between −571 and 337 kcal·d−1, which corresponded to −23% and 14% of mean TEEDLW. However, this study investigated the validity of the armband in the general population, and consequently, TEE (mean TEEDLW = 2492 ± 444 kcal·d−1) was lower than that in our sample of endurance athletes (mean TEEDLW = 3620 ± 900 kcal·d−1). This is in agreement with our observation that the error was associated with measures of the athletes' performance capacity, i.e., the individual lactate thresholds during running and bicycling.
Furthermore, this is supported by our finding that, also during the two controlled exercise bouts, the level of underestimation increased with the level of exercise intensity. ExEEArmband was significantly lower than IC values at higher intensities during running (2.8-4.8 m·s−1) and over the full intensity range from moderate to highly intensive during bicycling (140-340 W). The difference between ExEEArmband and ExEEIC increased with running speed or power output. Berntsen et al. (2) also found that the armband underestimated EE during more vigorous activity although the level of intensity was considerably lower than that in our study. In children, the degree of underestimation was also reported to increase with exercise intensity (1). Also, in most studies performed during walking (1,8,13), a general trend toward overestimating low levels of EE and underestimating high EE values was described. Only in one study was ExEE during treadmill walking and running systematically overestimated by the armband (16). However, exercise intensity was lower than that in our study (maximal speed = 3.5 m·s−1), and the degree of overestimation decreased substantially with increasing exercise intensity.
For bicycling, the results were more inconclusive because both underestimation (1,13) and overestimation (22) were reported in the literature. However, in the latter study, exercise-although very intense-lasted only roughly 6 min.
Because the parameters measured by the SenseWear Armband (acceleration, heat loss, skin conductivity) are directly associated with physical workload also in higher-intensity ranges, theoretically, it should be possible to develop exercise-specific prediction equations to improve the algorithm used for the estimation of EE. The adjustment of SenseWear Armband data for specific situations has been successful in other studies. During treadmill walking, stationary bicycling, and stair stepping, the application of exercise-specific algorithms provided accurate predictions of EE (13), but exercise intensity and, consequently, ExEE (maximal values = 8.3 ± 1.5 kcal·min−1 (walking), 6.6 ± 1.2 kcal·min−1 (bicycling), 9.2 ± 1.90 kcal·min−1 (stair stepping)) were much lower than those in our study (maximal values = 21.3 ± 1.6 kcal·min−1 (running), 26.9 ± 2.0 kcal·min−1 (bicycling)). In children, the application of specific equations during rest and moderate exercise also improved the accuracy of EE predictions with the armband (7).
Exercise-independent factors such as the estimation of resting metabolic rate (RMR) may also have affected the accuracy of the armband. For example, in a recent study in children, RMR was overestimated by the armband by 17% (7). Therefore, the accuracy of RMR assessment with the armband should also be investigated in athletes in further studies.
With respect to the interpretation of our results, it should be considered that although DLW and IC are highly accepted reference methods, errors may also result from the limitations of these methods. The accuracy of the DLW method has been reported to lie between 2% and 12%, depending on the loading dose, the length of the sampling period, the number of urine samples, and the food quotient used for the calculation of TEEDLW from carbon dioxide enrichment (4,33). The limitations of the present DLW method have been discussed in detail elsewhere (17). The precision of IC lies between 0.5% and 2% when ventilated open-circuit spirographs are used. IC is considered to be valid during steady-state exercise (30), and steady state is generally reached quickly during incremental exercise protocols. The last 3 min of each 5-min step, which has been used to calculate ExEEIC in the present study, is considered representative of the actual gas exchange (38). However, during high-intensity exercise, IC may underestimate actual ExEE because the anaerobic component is neglected (29). Furthermore, we did not cover excess postexercise oxygen consumption, which may be of special relevance after high-intensity and intermittent exercise. Excess postexercise oxygen consumption has been shown to contribute to the increase of TEE in athletes (14,18).
In the present study, only adult male endurance athletes were included as participants. Therefore, it is necessary to further investigate the validity of the SenseWear Pro3 Armband in other groups of athletes such as females and adolescents and also in other sports with intermittent or high-intensity exercise components.
Our results demonstrate that the SenseWear Pro3 Armband does not provide accurate estimates of TEE or ExEE in male endurance athletes, mainly because of the underestimation of EE during higher exercise intensities. Researchers and practitioners using the instrument in endurance athletes should be aware of the potential limitations. Further research should be directed at the development of exercise-specific corrections to make a more reliable assessment of EE in athletes possible.
This study was supported by institutional funding by the German Research Center of Elite Sport.
The authors thank Mrs. Anett Rümmler and Mr. Matthias Hägele for their practical assistance and support.
The use of commercial names in this article is solely for information purposes and does not represent any endorsement.
There are no conflicts of interest to declare.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Arvidsson D, Slinde F, Larsson S, Hulthén L. Energy cost of physical activities in children: validation of SenseWear Armband. Med Sci Sports Exerc
2. Berntsen S, Hageberg R, Aandstad A, et al. Validity of physical activity monitors in adults participating in free living activities. Br J Sports Med
3. Black AE. Dietary assessment for sports dietetics. Nutr Bull
4. Black AE, Prentice AM, Coward WA. Use of food quotients to predict respiratory quotients for the doubly-labelled water method of measuring energy expenditure. Hum Nutr Clin Nutr
5. Bland JM, Altman DG. Comparing methods of measurement: why plotting difference against standard method is misleading. Lancet
6. Cole PJ, LeMura LM, Klinger TA, Strohecker K, McConnell TR. Measuring energy expenditure in cardiac patients using the Body Media Armband versus indirect calorimetry: a validation study. J Sports Med Phys Fitness
7. Dorminy CA, Choi L, Akohoue SA, Chen KY, Buchowski MS. Validity of a multisensor armband in estimating 24-h energy expenditure in children. Med Sci Sports Exerc
8. Fruin ML, Rankin JW. Validity of a multi-sensor armband in estimating rest and exercise energy expenditure. Med Sci Sports Exerc
9. Gibala MJ, McGee SL. Metabolic adaptations to short-term high-intensity interval training: a little pain for a lot of gain? Exerc Sport Sci Rev
10. Heck H. Energiestoffwechsel und Medizinische Leistungsdiagnostik [Energy Metabolism and Medical Performance Diagnostics]
. Schorndorf (Germany): Hofmann; 1990. p. 138-49. German.
11. Hiremath SV, Ding D. Evaluation of activity monitors to estimate energy expenditure in manual wheelchair users. Conf Proc IEEE Eng Med Biol Soc
12. Hopkins WG, Marshall SW, Batterham AM, Hanin J. Progressive statistics for studies in sports medicine and exercise science. Med Sci Sports Exerc
13. Jakicic JM, Marcus M, Gallagher KI, et al. Evaluation of the SenseWear Pro Armband to assess energy expenditure during exercise. Med Sci Sports Exerc
14. Jamurtas AZ, Koutedakis Y, Paschalis V, et al. The effects of a single bout of exercise on resting energy expenditure and respiratory exchange ratio. Eur J Appl Physiol
15. Jones AM, Doust JH. A 1% treadmill grade most accurately reflects the energetic cost of outdoor running. J Sports Sci
16. King GA, Torres N, Potter C, Brooks TJ, Coleman KJ. Comparison of activity monitors to estimate energy cost of treadmill exercise. Med Sci Sports Exerc
17. Koehler K, Braun H, de Marées M, Fusch G, Fusch C, Schaenzer W. Parallel assessment of nutrition and activity in athletes: validation against doubly labelled water, 24-h urea excretion, and indirect calorimetry. J Sports Sci
18. LaForgia J, Withers RT, Gore CJ. Effects of exercise intensity and duration on the excess post-exercise oxygen consumption. J Sports Sci
19. Lamonte MJ, Ainsworth BE. Quantifying energy expenditure and physical activity in the context of dose response. Med Sci Sports Exerc
. 2001;33(suppl 6):S370-8.
20. Levine JA. Measurement of energy expenditure. Public Health Nutr
21. Loucks AB. Energy balance and body composition in sports and exercise. J Sports Sci
22. Mattson EM, Gallagher KI. Validity of the SenseWear Pro Armband during outdoor cycling. Med Sci Sports Exerc
23. Meyer H, Schönicke L, Wand U, Hubberten HW, Friedrichsen H. Isotope studies of hydrogen and oxygen in ground ice-experiences with the equilibration technique. Isotopes Environ Health Stud
24. Montoye HJ. Introduction: evaluation of some measurements of physical activity and energy expenditure. Med Sci Sports Exerc
. 2000;32(suppl 9):S439-41.
25. Prosser SJ, Scrimgeour CM. High-precision determination of 2
H in H2
O by continuous-flow isotope ratio mass spectrometry. Anal Chem
26. Rokuss K, Kalenka A, Bender HJ, Hinkelbein J. Intensive care patients. Determining daily energy expenditure-a comparison of two methods. Anaesthesist
27. Schoeller DA. Measurement of energy expenditure in free-living humans by using doubly labeled water. J Nutr
28. Schoeller DA, Racette SB. A review of field techniques for the assessment of energy expenditure. J Nutr
. 1990;120(11 suppl):1492-5.
29. Scott CB. Oxygen deficit and slow oxygen component relationships between intermittent and continuous exercise. J Sports Sci
30. Simonson DC, DeFronzo RA. Indirect calorimetry: methodological and interpretative problems. Am J Physiol
. 1990;258(3 Pt 1):E399-412.
31. Sjödin AM, Andersson AB, Högberg JM, Westerterp KR. Energy balance in cross-country skiers: a study using doubly labeled water. Med Sci Sports Exerc
32. St-Onge M, Mignault D, Allison DB, Rabasa-Lhoret R. Evaluation of a portable device to measure daily energy expenditure in free-living adults. Am J Clin Nutr
33. Trabulsi J, Schoeller DA. Evaluation of dietary assessment instruments against doubly labeled water, a biomarker of habitual energy intake. Am J Physiol Endocrinol Metab
34. Weir JB. New methods for calculating metabolic rate with special reference to protein metabolism. J Physiol
35. Westerterp KR. Food quotient, respiratory quotient, and energy balance. Am J Clin Nutr
. 1993;57(suppl 5):759-64S.
36. Westerterp KR. Limits to sustainable human metabolic rate. J Exp Biol
. 2001;204(Pt 18):3183-7.
37. Westerterp KR. Impacts of vigorous and non-vigorous activity on daily energy expenditure. Proc Nutr Soc
38. Whipp BJ, Davis JA, Torres F, Wasserman K. A test to determine parameters of aerobic function during exercise. J Appl Physiol