The accurate measurement of physical activity energy expenditure (PAEE) in children is a challenge. The specific problems involved with measuring physical activity (PA) in children relate to the intermittent nature of their activity (1) and the limited accuracy of recalled activity (1,28). Gold standard methods for the measurement of total energy expenditure, such as doubly labeled water and indirect calorimetry, are not suitable for measuring free-living PA in large populations (27,34). A key issue for the physical activity epidemiology community is to identify accurate objective methods of measuring PA suitable for use in large populations (2,18). Available methods include accelerometry and heart rate (HR) monitoring, both validated for measuring energy expenditure (EE) in controlled settings (12,35) and during free-living situations (11,20) in children. However, both methods have limitations. HR monitoring overestimates EE at low intensities, where anxiety, increases in body temperature, and a lag in HR response postexercise may increase HR without an associated increase in EE (21). Accelerometry is unable to account for increases in EE associated with carrying a load or cycling activity, when using a single hip-mounted sensor. In addition, the biomechanics of running lead to limitations of uniaxial accelerometry at high-energy expenditures (16).
It has been suggested that HR monitoring and accelerometry have uncorrelated or even negatively correlated error (7). It has been shown that combining the two methods provides better estimates of PAEE than either method alone (7,15). A new combined HR and movement sensor, the Actiheart (Cambridge Neurotechnology, Cambridge, UK), may exploit the advantages of both methods in an easy to use one-piece device, improving the accuracy and ease of assessment of PAEE in children. The Actiheart has been validated on adult volunteers during treadmill walking and running, indicating that the instrument is valid for prediction of PAEE in adults (8). The aim of this study was to assess the validity of the Actiheart in 12–13-yr-old children during treadmill walking and running, and to compare the estimated PAEE from the Actiheart with those of two other commercially available accelerometers.
METHODS
Participants
A total of 39 children, 16 female and 23 male (13.2 ± 0.3 yr), were recruited from a junior school in Ely, Cambridgeshire, UK. All students received an explanation of the study, all provided assent, and one parent provided written informed consent for their participation in the study. The Cambridge local ethics committee approved the study.
Measurements
Anthopometrics.
Height and body weight were measured using standard anthropometric techniques (22). Weight was measured to the nearest 0.1 kg using calibrated scales (Seca 761 mechanical flat scale, Hamburg, Germany), with the child wearing light clothes and without shoes or socks. Height was measured to the nearest 0.1 cm using a portable stadiometer (Chasmors Ltd., London, UK) without shoes and socks. Resistance (Ω) was assessed using a standard bioimpedance technique (Bodystat, Isle of Man, UK). This device has previously been shown to be a reasonably valid (29) and reliable (30) measure of percentage body fat. Total body water (TBW) and fat-free mass (FFM) were calculated using the impedance index (height squared divided by resistance), according published equations (32). Fat mass (FM) was calculated as body weight minus FFM.
Resting energy expenditure (REE).
The children were transported to the laboratory by car during school days. After at least a 2-h fasting and 10 min of supine rest, REE was measured for 6 min using indirect calorimetry. Oxygen consumption (V̇O2) and carbon dioxide production (V̇CO2) were measured with an online system (Jaegar Oxycon Pro, Viasys Health Care, Warwick, UK) on a breath-by-breath basis. Data were averaged over a 15-s epoch, disregarding the two most extreme values in each interval. EE was then computed using the de Weir equation (10). Before every test, vanes were calibrated using known volumes of flow rate (0.2 and 2.0 L·s−1) and the gas analyzers against known concentrations of gases (5% CO2 and 0% O2) and room air. HR was simultaneously measured by the Actiheart and a Polar HR monitor (Polar Electro Oy, Kempele, Finland). REE was determined by averaging EE for the last 2.5 min of the test, and resting HR was determined by averaging the HR data from the Actiheart for the same time interval.
Exercise Test
Before the exercise test, the children were allowed to practice on the treadmill for a minimum of 2 min to allow familiarization with the treadmill. Respiratory gas exchange data were collected throughout the treadmill test using the same equipment and procedures as described above. PAEE (J·kg−1·min−1) was calculated by subtracting REE from EE measured during the exercise test.
The treadmill test involved 3 min of walking at 3.2 km·h−1. The speed was then gradually increased at a rate of 0.33 km·h−1·min−1 over the next 6 min (to 5.2 km·h−1). The gradient was then gradually increased at a rate of 2% per minute during 3 min (to 6%). Both speed and gradient were then increased gradually to reach a speed of 5.8 km·h−1 and a gradient of 10.2% by 15 min, when the gradient decreased back to the flat over a period of 30 s. This marks the end of the walking part and the beginning of the running part of the protocol. Treadmill speed then increased to 9 km·h−1 and gradually increased at a rate of 0.8 km·h−1·min−1 during the next 4 min (to 12.2 km·h−1 by 20.5 min), after which time the treadmill slowed down gradually. The treadmill test was terminated sooner if the HR reached 90% of age-predicted maximal HR (33), or if a HR above 80% of predicted maximum HR had been sustained for more than 3 min. Only four children finished the protocol, 25 completed 17 min, and all children completed 13 min of the protocol. Predicted V̇O2max was estimated by extrapolating individual HR-V̇O2 regression lines to age-predicted maximum HR.
Activity Monitors
We evaluated the Actiheart combined HR and movement sensor (Cambridge Neurotechnology, Cambridge, UK), hip- and ankle-mounted MTI Actigraphs (Model 7164, Manufacturing Technologies Inc. Health Systems, Shalimar, FL), and a hip-mounted Actical (Mini Mitter Co., Inc., Bend, OR) in the present study. A detailed description of the MTI Actigraph (formerly known as the CSA activity monitor) is available elsewhere (13). Briefly, the Actigraph is a uniaxial accelerometer sensitive to movements in the 0.51- to 3.6-Hz range, and when mounted to the hip, most sensitive to vertical movements of the torso. The Actical is also a uniaxial accelerometer, it is sensitive to movements in the 0.5- to 3-Hz range but is mounted slightly differently (“omnidirectional”). Similar to the Actigraph, the Actical is most sensitive to vertical movements of the torso when placed on the hip. All monitors were attached to the child before the treadmill test, and set to record data in 15-s epochs. The hip MTI and Actical were mounted on an elastic waist strap, and the hip placement alternated between the right and left hip for every other child. The ankle-mounted MTI was attached to a Velcro ankle strap and placed on the lateral midline of the right ankle. All monitors were initialized before the test on time-synchronized computers.
A detailed description of the Actiheart monitor is available elsewhere (8). Briefly, the monitor weighs approximately 8 g and is attached to the chest with two standard ECG electrodes. The Actiheart is able to measure acceleration, HR, HR variability, and ECG magnitude for epoch settings of 15, 30, and 60 s. HR, HR variability, and ECG magnitude can be recorded for a set time with a memory capacity of 128 kbyte, for example, 11 d with an epoch of 60 s. Data on interbeat intervals (IEI logging) and ECG waveforms can also be recorded for about 24 h and 13 min, respectively. Acceleration is measured by a piezoelectric element within the Actiheart with a frequency range of 1–7 Hz (3 dB). For every participant, the Actiheart monitor was tested for adequate HR pickup by recording ECG waveforms for approximately 30 s before the rest test. If pickup was adequate, the Actiheart was set up to record HR and movement continuously for the rest of the session, using the IEI logging mode. One electrode was placed at the base of the child’s sternum and the other horizontally to the child’s left side, with the Actiheart spaced so that the wire between the two sections of the Actiheart was straight but not taut.
Statistical Analysis
All data was pooled in Microsoft Excel and then transferred to Intercooled Stata 8.2 (Statacorp, College Station, TX) for statistical analysis. Descriptive data are summarized as mean ± SD. Backward stepwise linear regression was used to predict PAEE (J·kg−1·min−1) from the output from the activity monitors (counts and HR), gender and height, disregarding the first 2 min. Height was included in the model as a proxy for leg length to account for interindividual differences in stride length that may affect the activity count–PAEE relationship. Three separate models were derived from the Actiheart data. First, the activity model using activity counts, gender, and height was derived. Secondly, the HR model in which HR above rest (HRAR), gender, and height were used to predict PAEE, and finally a combined model using the terms activity counts, HRAR, gender, and height. The interaction term gender-HRAR was introduced in all models that included HRAR. Three separate PAEE prediction equations were then derived for the hip- and ankle-mounted MTI and Actical, respectively. Age was not included into any of these models because of lack of heterogeneity in the sample.
To determine the validity of the derived prediction equations, a cross-validation approach was used. A gender-balanced subsample (N = 15) of the population was randomly selected, and prediction equations derived from the nonselected participants (N = 24) were used to predict PAEE in the subsample. Modified Bland–Altman plots of the difference (estimation error) between predicted and criterion-measured PAEE, plotted against the measured PAEE, were produced for each model. Separate plots for the different activity intensities of level walking, graded walking, and level running were also produced, and the degree of agreement (mean difference ± 95% CI) between criterion-measured PAEE and predicted PAEE were calculated (3). A P value of less than 0.05 was used to determine statistical significance.
RESULTS
A summary of the participant characteristics is shown in Table 1. There were no significant differences between gender according to age, weight, height, body fat percentage, and REE. Predicted V̇O2max was significantly higher in boys (P = 0.003).
;)
TABLE 1:
Descriptive characteristics of participants (N = 39).
Figure 1 (a–c) show the increase in PAEE, HR, and activity counts with increasing work rate. From these figures it is clear that steady state has been reached within 2 min. Further increments (beyond 3 min) in work rate are very small, and because PAEE and HR adapt to a new workload characterized by first-order monoexponential functions, all observations after 2 min are in practice a true reflection of the underlying workload.
The PAEE prediction equations for all derived models are displayed in Table 2. The combined Actiheart (HR and activity) model explained 86% of the variance in PAEE. The Actiheart HR model was the most accurate single-measure predictor of PAEE, explaining 82% of the variance. The PAEE prediction using the Actiheart accelerometer only (activity model) was comparable to that from the Actical, with an explained variance of 69 and 67%, respectively. The two MTI Actigraph models explained least of the variance in PAEE. The hip-mounted MTI monitor explained 50%, and the ankle-mounted monitor explained 37% of the variance in PAEE. The standard error of the estimate (SEE) was lowest for the combined Actiheart model (69 J·kg−1·min−1).
TABLE 2:
Final PAEE (J·kg−1·min−1) prediction equations during treadmill walking and running from Actiheart (activity model, HR model, and combined activity + HR model) and from a hip mounted, ankle mounted MIT accelerometer and a hip mounted Actical (N = 39).
The validity of derived prediction equations was then tested in a gender-balanced, randomly selected subset of fifteen participants. The difference between measured and predicted PAEE (J·kg−1·min−1) was plotted against measured PAEE (J·kg−1·min−1) for each model and presented in Figure 2 (a–f). The two distinct clusters visible on each of the activity monitor graphs represent walking and running phases of the protocol. The positively correlated errors with measured PAEE in the plots of all single-measure accelerometry models indicate a systematic error in these prediction equations. This error manifests as an overestimation of PAEE at low intensity levels and, consequently, an underestimation at higher intensities (bias r = 0.49 to 0.90, P < 0.05). The HR model and the combined HR and activity model indicated less systematic error (r = 0.41 and 0.33, respectively).
Cross-validation statistics for separate activity phases are shown in Table 3; all data points were averaged over each phase. The Actiheart combined model was the only model that did not significantly underestimate PAEE for flat walking, and the Actiheart HR model predicted PAEE with no significant difference from the measured criterion value for graded walking. The Actiheart HR, the combined Actiheart, and the Actical models predicted PAEE with no significant difference from the measured values during running. All other predictions were significantly different from the measured values for all activity intensities.
TABLE 3:
PAEE (J·kg−1·min−1) assessed by indirect calorimetry and predicted from six different models during three different phases of treadmill activity in children (derivation subsample (N = 24), cross-validation subsample (N = 15)).
DISCUSSION
The aim of this study was to validate the Actiheart combined HR and movement sensor to predict PAEE during treadmill walking and running in children. The predictions of PAEE from the Actiheart were compared with those from a hip- and an ankle-mounted MTI Actigraph and a hip-mounted Actical accelerometer, against the criterion measure of PAEE by indirect calorimetry. The combined HR and body movement model provided the most accurate prediction of PAEE. The HR model was the most accurate of the single-measure models, followed by the body movement model from Actiheart, body movement model from the Actical, body movement model from the hip-mounted MTI Actigraph, and lastly the ankle-mounted MTI Actigraph. All monitors except for the combined Actiheart model significantly overestimated PAEE during flat walking, with overestimates ranging from 33 J·kg−1·min−1 for the Actiheart HR model to 104 J·kg−1·min−1 for the ankle MTI model.
Our measure of REE compares well with the estimates of Harrell et al. (14). We used a progressive treadmill protocol in this study, which has major advantages over a classic steady-state protocol for developing the regression line between HR, movement counts, and PAEE. When using a continuous protocol, each data point contributes to the regression line, which then makes this line much more robust compared to averaging a few data points for each exercise intensity from a steady-state protocol. Due to the very small increases in workload in this protocol, any possible time shift may only arise from differences in PAEE, HR, and activity kinetics, which in the case of PAEE and HR will be virtually zero, and in the case of PAEE and activity (believed to respond immediately to a new workload) negligible. Thus, we believe our measure of PAEE is in practice a true reflection of the underlying workload.
This study was conducted during walking and running on a treadmill in a controlled laboratory setting. Therefore, the developed PAEE prediction equations may only apply to treadmill walking and running, and should not be used without caution for free-living prediction of PAEE. However, the controlled setting, the use of the same monitors for all children, and the use of a strict protocol provided the optimal conditions for between-model and monitor comparisons, which are easily reproduced. Therefore, we have no reason to believe that our results are due to bias or chance. As walking and running are two of the most frequent types of children’s physical activity, the results from this study are likely to generalize to a relatively large proportion of children’s physical activities.
The results are, however, limited to healthy, nonobese children of the same age. Further studies would be required to establish the validity of the Actiheart in obese children and other age groups, in which movement economy and style could well differ from our sample (17,23). This study may not fully highlight possible advantages of a combined HR and movement sensor over other methods, as all of the activities performed were in the moderate- to vigorous-intensity category. The benefits of accelerometry to a combined model would probably be greatest during low-intensity activities, below and around the HR flex point, which is normally used to discriminate between rest and physical activity when assessing PAEE solely by HR monitoring but with the use of individual calibration (36). Low-intensity physical activities were not included in this study. It is plausible that a free-living validation study would better illustrate the advantage of combined sensing, enabling determination of whether a specific HR increase is due to an increase in energy expenditure or is due to other factors such as stress. For free-living scenarios, it is likely that more complex modeling techniques are needed to most accurately assess PAEE (7), as a starting point perhaps using the two single-measure Actiheart HR and Actiheart activity models from the present study. In this regard, it is important to note the biomechanical limitations of vertical accelerometry as a measure of PAEE during running, load-carrying, swimming, and cycling activities (4,5,9,16). Some of these limitations were also evident in this study, where accelerometry equations derived on flat walking data would underestimate uphill walking PAEE and vice versa, and where estimation error for running PAEE increases with running speed.
The Actiheart movement sensor explained the most variance in PAEE of any accelerometer model in this study. Despite significant correlations between the activity counts from the different monitors and criterion-measured PAEE, the differences in explained variance between models were substantial. There are at least two reasons for this. Firstly, there are between-monitor type differences in measuring acceleration (5,8). The Actiheart measures vertical acceleration in the longitudinal axis of the trunk for frequencies between 1 and 7 Hz. This frequency range is different from that reported for the Actical (0.5–3.0 Hz) and from the MTI Actigraph (0.25–2.5 Hz). Furthermore, the Actical is mounted slightly differently within its case (“omnidirectional”). Although not a factor in the present study, between-unit within-monitor type differences would be an additional factor in a large study where multiple units are needed of a particular type (4,8,25,37). Secondly, the placement of the monitor may contribute to the between-monitor differences observed (38). The explained variance in PAEE was most similar from the Actiheart and the Actical, despite different placements, indicating the accelerometer output from these monitors are comparable when mounted as suggested by the manufacturer. The effect of placement was also evident when comparing the hip- and ankle-mounted MTI Actigraph. The hip-mounted MTI Actigraph explained significantly more of the variance in PAEE compared with the ankle-mounted monitor, indicating the influence of placement when assessing PAEE from accelerometry in a controlled setting. The difference between the Actical and the MTI Actigraph (both placed on the hip) is likely due to differences in the sensitivity of the sensors within the two monitors to quantify acceleration. For example, the MTI Actigraph displays a movement frequency–dependent response to acceleration (4), which inflates between-subject error in predicting work rate (5,6). In contrast, the accelerometer sensors used in the Actiheart and Actical are similar and highly linear with acceleration, regardless of movement frequency (8) (Fig. 1).
Our results are comparable to previous studies examining the validity of activity monitors based on accelerometry in controlled settings in children (12,26,35). However, all accelerometer models used in the present study display a similar pattern of inherent limitations, indicating a systematic error for prediction of lower intensity levels (Fig. 2). This limitation is likely to be due to the inability of accelerometers to accurately estimate the rate of energy expenditure during walking on a gradient, as previously observed for the MTI Actigraph (24). Our treadmill protocol included a combination of increasing treadmill speed and incline, and it is evident that none of the tested accelerometers accurately estimated the elevated energy expenditure due to incline and high running speed. It is possible that segmental linear modeling may improve estimation accuracy. This limitation of body movement measurement by accelerometry may translate to larger errors when estimating free-living PAEE as compared with a combined HR and body movement monitor.
The HR model was the most accurate single-measure model for the prediction of PAEE in the children in this study. Previous studies (12) have found HR monitoring to provide a relatively accurate estimation of PAEE in children (R2 = 0.80), especially for higher work rates, to which these results are comparable. At lower intensities, predictions from HR tend to be less accurate (19,31). However, the moderate-to-high intensity levels in this treadmill protocol probably contributed to the accuracy of HR monitoring for prediction of PAEE.
The combined Actiheart prediction equation explained most of the variance in this study. To the best of our knowledge, no study has previously used a single-piece combined HR and movement sensor to assess PAEE in children. However, the validation coefficient for the combined model was similar to those reported from other studies validating the combined use of HR and movement data for predicting PAEE during various activities in children (12). In that study, a combination of body movement from triaxial accelerometry and HR explained 85% of the variation in PAEE. However, these authors argued that the added explained variance by combination of multiple monitors is offset by the burden for the children. The Actiheart combines body movement registration and HR in a single-piece instrument, which is lightweight and waterproof. The one-piece design of the Actiheart offers advantages over simultaneous accelerometry and HR monitoring with separate monitors, including lower investigator and participant burden and intrinsic time synchronization. It remains to be demonstrated whether the combined predictions from the Actiheart are significantly more accurate than single-objective methods for the prediction of free-living PAEE in children.
Although the output from the Actiheart explained most of the variation in intensity (PAEE·kg–1·min–1), the cross-validation indicated that all prediction models were suitable for assessing the total PAEE for the treadmill activity performed in this study. Further validation during free-living and other specific activities is required to determine whether the Actiheart offers further advantages over other objective measures of PAEE.
In conclusion, our results suggest that the combined HR and activity monitor Actiheart is valid for estimating PAEE in children during treadmill walking and running. The combination of HR and activity counts provided the most accurate estimate of PAEE as compared with accelerometry measures alone.
REFERENCES
1. Bailey, R. C., J. Olsen, S. L. Pepper, P. Porszasz, T. J. Barstow, and D. M. Cooper. The level and tempo of children’s physical activities: an observational study.
Med. Sci. Sports Exerc. 27(7):1033–1041, 1995.
2. Bassett, D. Validity and reliability issues in objective monitoring of physical activity.
Res. Quart. Exerc. Sport. 71(2 SuppI.):S30–36, 2000.
3. Bland, J. M., and D. G. Altman. Statistical methods for assessing agreement between two methods of clinical measurement.
Lancet 8:307–310, 1986.
4. Brage, S., N. Brage, N. Wedderkopp, and K. Froberg. Reliability and validity of the Computer Science and Applications Accelerometer in a Mechanical Setting.
Meas. Phy. Edu. Exerc. Sci. 7(2):101–119, 2003.
5. Brage, S., N. Wedderkopp, P. W. Franks, L. B. Andersen, and K. Froberg. Reexamination of Validity and Reliability of the CSA monitor in walking and running.
Med. Sci. Sports Exerc. 35(8):1447–1454, 2003.
6. Brage, S., N. Wedderkopp, L. B. Andersen, and K. Froberg. Influence of Step Frequency on Movement Intensity Predictions with the CSA Accelerometer: A Field Validation Study in Children.
Ped. Exerc. Sci. 15, 2003.
7. Brage, S., N. Brage, P. W. Franks, et al. Branched equation modeling of simultaneous accelerometry and heart rate monitoring improves estimate of directly measured physical activity energy expenditure.
J. Appl. Physiol. 96:343–51, 2004.
8. Brage, S., N. Brage, P. W. Franks, U. Ekelund, and N. J. Wareham. Reliability and validity of the combined heart rate and movement sensor. Actiheart.
Eur. J. Clin. Nutr. 59:561–70, 2005.
9. Cavagna, G. A., H. Thys, and A. Zamboni. The sources of external work in level walking and running.
J. Physiol. 262:639–57, 1976.
10. de Weir, J. B. New methods for calculating metabolic rate with special reference to protein metabolism.
J. Physiol. 109:1–9, 1949.
11. Ekelund, U., M. Sjostrom, A. Yngve, et al. Physical activity assessed by activity monitor and doubly labelled water in children.
Med. Sci. Sports and Exerc. 33:275–281, 2001.
12. Eston, R. G., A. V. Rowlands, and D. K. Ingledew. Validity of heart rate, pedometry, and accelerometry for predicting the energy cost of children’s activities.
J. Appl. Physiol. 84:362–371, 1998.
13. Freedson, P., E. Melanson, and J. Sirard. Calibration of the Computer Science and Applications, Inc. accelerometer.
Med. Sci. Sports Exerc. 30:777–781, 1998.
14. Harrell, J. S., R. G. McMurray, C. D. Baggett, M. I. Pennell, P. F. Pearce, and S. I. Bangdiwala. Energy costs of physical activities in children and adolescents.
Med. Sci. Sports Exerc. 37: 329–336, 2005.
15. Haskell, W. L., M. E. Yee, A. Evans, and P. J. Irby. Simultaneous measurement of heart rate and body motion to quantitate physical activity.
Med. Sci. Sports Exerc. 25:109–115, 1993.
16. Haymes, E. M., and W. E. Byrnes. Walking and running energy expenditure estimated by caltrac and indirect calorimetry.
Med. Sci. Sports Exerc. 25:1365–1369, 1993.
17. Hills, A., and A. Parker. Locomotor characteristics of obese children.
Child Care Health Dev. 18:29–34, 1992.
18. Lamonte, M., and B. Ainsworth. Quantifying energy expenditure and physical activity in the context of dose response.
Med. Sci. Sports Exerc. 33(6 Suppl):S370–378, 2001.
19. Li, R., P. Deurenberg, and J. Hautvast. A critical evaluation of heart rate monitoring to assess energy expenditure in individuals.
Am. J. Clin. Nutr. 58:602–607, 1993.
20. Livingstone, M. B., W. A. Coward, A. M. Prentice, et al. Daily energy expenditure in free-living children: comparison of heart-rate monitoring with the doubly labelled water (2H2(l8)O) method.
Am. J. Clin. Nutr. 56:343-352, 1992.
21. Livingstone, M. B., P. J. Robson, and M. Totton. Energy expenditure by heart rate in children: an evaluation of calibration techniques.
Med. Sci. Sports Exerc. 32:1513-1519, 2000.
22. Lohman, T., A. Roche, and R. Martorell.
Anthropometric Standardization Reference Manual. Champaign: Human Kinetics; pp. 1–90, 1991.
23. McCann, D. and W. Adams. The size-independant cost of running.
Med. Sci. Sports Exerc. 35:1049–1056, 2003.
24. Melanson, E., and P. Freedson. Validity of the Computer Science and Application, Inc. (CSA) activity monitor.
Med. Sci. Sports Exerc. 27:934, 1995.
25. Metcalf, B., L. S. Curnow, C. Evans, L. D. Voss, and T. L. Wilkin. Technical reliability of the CSA activity monitor: The Early Bird Study.
Med. Sci. Sports Exerc. 34:1533–1537, 2002.
26. Puyau, M., A. L. Adolph, P. A. Vohra, and N. P. Butte. Validation and calibration of physical activity monitors in children.
Obes. Res. 10:150–157, 2002.
27. Rennie, K. L., T. Rowsell, S. A. Jebb, D. Rolburn, and N. J. Wareham. A combined heart rate and movement sensor: proof of concept and preliminary testing study.
Eur. J. Clin. Nutr. 54: 409–414, 2000.
28. Sallis, J. P. Self-report measures of children’s physical activity.
J. School Health. 61:215–219, 1991.
29. Simpson, J. D., N. Lobo, J. A. Anderson, et al. Body water compartment measurements: a comparison of bioelectrical impedance analysis with tritium and sodium bromide dilution techniques.
Clin. Nutr. 20:339–343, 2001.
30. Shanholtzer, B. and S. Patterson. Use of bioelectrical impedance in hydration status assessment: reliability of a new tool in psychophysiology research.
Int. J. Psychophysiol. 49:217–226, 2003.
31. Strath, S., A. M. Swartz, D. R. Bassett, W. L. O’Brien, G. A. King, and B. E. Ainsworth. Evaluation of heart rate as a method for assessing moderate intensity physical activity.
Med. Sci. Sports Exerc. 32:465–470, 2000.
32. Sun, S., W. C. Chumlea, S. B. Reymsfield, et al. Development of bioelectrical impedance analysis prediction equations for body composition with the use of a multicomponent model for use in epidemiological surveys.
Am. J. Clin. Nutr. 77:331–340, 2003.
33. Tanaka, R., K. Monahan, and D. Seals. Age-predicted maximal heart rate L. J revisited.
J. Am. Call. Cardia I. 37:153–156, 2001.
34. Treuth, M. S., A. L. Adolph, and N. F. Butte. Energy expenditure in children predicted from heart rate and activity calibrated against respiration calorimetry.
Am. J. Physiol. 275:EI2-EI8, 1998.
35. Trost, S., D. S. Ward, S. M. Moorehead, P. D. Watson, W. Riner, and J. R. Burke. Validity of the computer and science and applications (CSA) activity monitor in children.
Med. Sci. Sports Exerc. 30:629–633, 1998.
36. Wareham, N. J., S. J. Hennings, C. D. Byrne, C. N. Hales, A. M. Prentice, and N. E. Day. A quantitative analysis of the relationship between habitual energy expenditure, fitness and the metabolic syndrome.
Br. J. Nutr. 80:235–241, 1998.
37. Welk, G., J. Schaben, and J. Morrow. Reliability of accelerometry-based activity monitors: a generalizability study.
Med. Sci. Sports Exerc. 36:1637–1345, 2004.
38. Yngve, A., A. Nilsson, M. Sjostrom, and U. Ekelund. Effect of monitor placement and of activity setting on the MTI accelerometer output.
Med. Sci. Sports Exerc. 35:320–326, 2003.
