There is growing evidence that physical activity is important to the short and long term health of children and adolescents (23). Among youth, physical activity is inversely associated with a number of cardiovascular disease risk factors, including elevated blood lipids(3,25), hypertension (2), obesity (4,26), and cigarette smoking(11), while positively associated with physical fitness(15), HDL cholesterol (1), bone mass (5), and psychological well-being(9). On the basis of this evidence, an expert panel recently recommended that adolescents be physically active on a daily or near daily basis and complete at least three bouts of continuous moderate to vigorous physical activity on a weekly basis (23). These recommendations are consistent with the national physical activity objectives for children and adolescents as outlined in Healthy People 2000(21) and the CDC/ACSM recommendation for physical activity and public health (20).
From a public health perspective, it is desirable to establish what proportion of children and adolescents are meeting the above guidelines and what proportion are in need of intervention programs to increase habitual physical activity (19,24). Unfortunately, the quality of existing descriptive information is limited because previous surveys have relied on self-report methods which are subject to recall problems and do not provide adequate description of the quantity and quality of physical activity (12,18). One solution to this problem is to use more burdensome objective measures of physical activity such as accelerometers or heart rate monitors in population-representative samples of children and adolescents. However, until recently the feasibility of such an approach has been limited by the excessive cost and obtrusiveness associated with these methods.
The Computer Science and Applications (CSA) activity monitor (WAM 7164) is a small (5.1 × 3.8 × 1.5 cm), light weight (45 g) uniaxial accelerometer which can be worn at the hip, ankle, or wrist. The small size, relatively low cost, and robust design features of the CSA activity monitor make this instrument highly suitable for use in moderately large, population-representative samples of children and adolescents. The CSA activity monitor can be easily initialized and downloaded on any IBM compatible personal computer and, in contrast with the more widely use Caltrac activity monitors, can store data continuously for up to 6 wk. Other features that make the CSA activity monitor a preferred instrument in epidemiological research include the ability to specify start and stop times and an internal real-time clock that allows data to be analyzed over intervals as short as 1 s.
To date, few studies have examined the validity of the CSA activity monitor. Janz (14) assessed the validity of the CSA monitor as a field measure of physical activity in children aged 7 to 15. Correlation coefficients between CSA counts and various indices of heart rate, including average net heart rate, number of minutes with heart rate greater than or equal to 60% of heart rate reserve, and number of minutes with heart rate greater than or equal to 150 beats·min-1, were statistically significant and ranged from 0.50 to 0.74. More recently, Melanson and Freedson (16) demonstrated the CSA monitor to be a valid measure of treadmill walking and running in adults using energy expenditure assessed via indirect calorimetry as the criterion measure. While both these studies support the validity of the CSA monitor as an objective measure of physical activity, the relationship between CSA activity counts and energy expenditure has yet to be examined in children. Hence, the purpose of this study was to evaluate the concurrent validity of the CSA activity monitor as a measure of children's physical activity using energy expenditure determined by indirect calorimetry as a criterion measure. A secondary purpose was to derive an energy expenditure prediction equation using CSA counts and other demographic and anthropometric variables.
Subjects. Subjects for this study were 30 children (19 boys and 11 girls) ranging in age from 10 to 14 yr. Descriptive data for these subjects are summarized in Table 1. Before participation in the study, written informed consent was obtained from each participant and his or her primary guardian. Protocol requirements established by the University of South Carolina School of Public Health Ethics Committee were satisfied before data collection.
CSA instrumentation. The CSA activity monitor (WAM 7164) is designed to detect acceleration ranging in magnitude from 0.05 to 2.00 G with frequency response from 0.25 to 2.50 Hz. These parameters allow for the detection of normal human motion and will reject high frequency motion encountered in activities such as operation of a lawn mower. The filtered acceleration signal is digitized and the magnitude is summed over a user-specified epoch interval. At the end of each epoch, the summed value is stored in memory and the integrator is reset. This process repeats itself until memory is filled or the instrument is reset. The hardware used in the WAM 7164 includes an 8-bit microcontroller with an 8-bit analog to digital converter, 64 kilobytes of nonvolatile RAM, a low power operational amplifier, and a piezoelectric motion sensor with analog signal conditioners and filters. Power is supplied by a readily available 2430 coin cell lithium battery which typically lasts 4 to 6 months. Communications with the WAM 7164 is achieved with a coded infrared beam of light via a Reader Interface Unit (RIU) connected to a serial port. Smart terminal emulation software is supplied with the RIU to support communications with the activity monitor(10).
Protocol. Before each testing session the activity monitors were initialized according to the manufacturers specifications and were synchronized to an external time piece. For the present study, the epoch duration or sampling period was set at 1 min and output was expressed as counts per minute. To evaluate inter-instrument reliability, subjects wore two CSA activity monitors. Both monitors were housed in small nylon pouches provided by the manufacturer and secured directly over the right and left hip using an adjustable elastic belt. Following a 5- to 7-min familiarization period with the treadmill, subjects performed three 5-min exercise bouts which consisted of (a) walking at 3 mph, (b) walking at 4 mph, and (c) jogging at 6 mph. Because previous studies have shown vertical axis accelerometers to be generally nonresponsive to changes in grade(13,16), all bouts were performed at 0% grade. Each exercise bout was separated by a 3-min rest period during which the subject rested quietly in a seated position. At the end of the testing session, the CSA activity monitors were immediately removed and downloaded on a personal computer.
Indirect calorimetry. Expired air for the determination of pulmonary ventilation (˙VE), oxygen consumption (˙VO2), and carbon dioxide production (˙VO2) was analyzed in 15-s intervals using an on-line computer-based acquisition system (Rayfield Equipment, Waitsfield, VT) consisting of an Applied Electrochemistry S-3A oxygen analyzer(Ametek, Pittsburgh, PA), a Beckman LB-2 carbon dioxide analyzer (Ventura, CA), and a Parkinson-Cowan (London) gasometer. Analyzers were calibrated before each testing session using verified calibration gasses. Steady-state energy expenditure (kcal·min-1) during each exercise bout was determined by multiplying mean ˙VO2 (L·min-1) over minutes 3 to 5 by the caloric equivalent of the corresponding mean respiratory exchange ratio. Exercise heart rate was monitored using a Polar Vantage XL heart rate monitor (Polar CIC Inc., Port Washington, NY) and was recorded as the average of minutes 3 to 5 of each exercise bout.
Statistical analysis. All statistical analyses were performed using SAS version 6.08. Differences in the CSA monitor output at different speeds and across units were tested using two-way ANOVA with repeated measures. Tukey's post-hoc analysis was used to determine the location of significant pairwise differences. For both of the CSA activity monitors worn, associations between CSA counts per minute and energy expenditure, ˙VO2, heart rate, and treadmill speed were assessed using Pearson product-moment correlation coefficients. To evaluate the validity of the CSA monitor in estimating energy expenditure, 20 subjects were randomly assigned to a developmental group, with the remaining 10 subjects serving as a validation group. An equation for predicting energy expenditure(kcal·min-1) was developed from developmental group data using forward stepwise multiple regression. Independent variables included CSA counts per minute, body mass, gender, age, and height. The resultant prediction equation was cross-validated with validation group data by computing the correlation and SEE between predicted and actual energy expenditure values. For the correlation and regression analyses, data from the three treadmill speeds were pooled and treated as independent observations. Data from the CSA monitor worn on the right hip was used to develop and cross-validate the energy expenditure prediction equation. For all analyses, statistical significance was set at an alpha level of 0.05.
Mean and SD for the physiological dependent variables are shown inTable 2. As expected, mean values for each variable increased in proportion to treadmill speed. Mean activity counts for the two activity monitors are shown in Figure 1. For both CSA monitors, activity counts increased significantly with treadmill speed (Speed main effect: F2,114 = 575.5, P < 0.001). Within each level of treadmill speed, mean activity counts for the two CSA monitors were not significantly different (Unit × Speed interaction:F2,114 = 0.03, P = 0.97). The inter-instrument reliability coefficient (intraclass R) for CSA activity counts averaged over the three treadmill speeds was 0.87.
Correlations between counts per minute and each of the physiological criterion measures are shown in Table 3. For both CSA monitors, counts per minute were strongly correlated with energy expenditure,˙VO2, heart rate, and treadmill speed. Correlation coefficients ranged from 0.77 to 0.87, and all were statistically significant at the 0.001 level. Notably, the validity coefficients for the two CSA activity monitors were almost identical.
Stepwise multiple linear regression analyses indicated that energy expenditure could be predicted by CSA counts per minute and body mass. No other variables entered the regression model at 0.05 level of significance. The final prediction equation was:
The multiple R-squared for the prediction equation was 0.83 and the SEE was 0.97 kcal·min-1. The results of the cross-validation study are presented in Table 4. Averaged across all three treadmill speeds, the regression equation predicted energy expenditure within 0.01 kcal·min-1. The mean absolute difference between actual and predicted energy expenditure at 3, 4, and 6 mph was 0.47, 0.60, and 0.81 kcal·min-1, respectively. The correlation between actual and predicted mean energy expenditure was 0.93 (P < 0.001) and the SEE was 0.93 kcal·min-1. The correlations between actual and predicted energy expenditure at each treadmill speed were statistically significant (P < 0.01) and ranged from 0.62 to 0.85. The SEE increased with treadmill speed, ranging from 0.66 kcal·min-1 at 3 mph to 1.08 kcal·min-1 at 6 mph.
This study assessed the validity of the CSA activity monitor in quantifying physical activity in children and adolescents. Consistent with Melanson and Freedson's adult study (16), a strong and highly significant correlation was observed between CSA activity counts and energy expenditure assessed by indirect calorimetry. In addition, the CSA activity monitor exhibited a high degree of inter-instrument reliability. Validity coefficients for the two CSA monitors were almost identical(Table 3), and at each speed the average counts per minute recorded by the two CSA monitors were not significantly different. The intraclass reliability coefficient for the two CSA units was 0.87. From these data, it appears that the CSA WAM 7164 activity monitor is a valid and reliable tool for quantifying physical activity performed by children aged 10 to 14 yr.
It is important to note that the validity coefficients reported in the present study were obtained using laboratory-based treadmill exercise and may not be applicable to physical activity performed in field settings. The CSA activity monitor is a uniaxial accelerometer (vertical axis) and, as such, may not be able to detect the torsional accelerations produced by youngsters during normal play and other nonlocomotor activities. Data from early activity monitor studies suggest that uniaxial accelerometer devices maybe as effective as triaxial accelerometer devices in quantifying human physical activity. Webster et al. (27) examined output from three piezo-ceramic transducers mounted at perpendicular angles on the wrist. While small differences in amplitude were observed among the three axis, there were virtually no instances where movements were not recorded simultaneously by all three transducers. Redmond and Hegge (22) also examined movement detection by accelerometers oriented in three perpendicular planes on the wrist. Output from each axis was similar, indicating that movement in a single plane was accompanied by countermovements in other planes. More recently, Bouten et al. (8) compared uniaxial and triaxial accelerometer output during nonlocomotor sedentary activities and treadmill walking (3-7 km·h-1). In contrast to earlier studies, their accelerometer device was worn on the hip instead of the wrist. During sitting, writing, arm work (ironing clothes), and sitting and standing, the sum of output from all three axis was the most accurate predictor of energy expenditure. During treadmill walking, unidirectional acceleration recorded in the antero-posterior plane was the most accurate predictor of energy expenditure, despite the fact that the major acceleration component was in the vertical direction. Notably, for each activity examined, movement recorded in the antero-posterior and medio-lateral planes was accompanied by movement detection in the vertical axis.
Consequently, it appears that triaxial accelerometers may outperform uniaxial devices in predicting energy expenditure during sedentary activities such as writing, sitting, and standing. Nevertheless, uniaxial accelerometers may still provide useful estimates of nonlocomotor physical activity, since movement in the antero-posterior and medio-lateral planes is typically accompanied by movement in the vertical plane. In support of this notion, Welk and Corbin (28) found a uniaxial accelerometer (Caltrac) to be equally effective as a triaxial model in quantifying physical activity performed by children. Additionally, a number of investigators have shown the uniaxial Caltrac accelerometer to provide valid estimates of energy expenditure in children and adolescents(7,12,18). Hence, it appears reasonable to conclude that our findings may be applicable to physical activity performed in field settings. Clearly, future studies are required to assess the relationship between CSA counts and children's energy expenditure during free play and other non-laboratory-based activities.
The placement of the CSA monitor at the hip was selected to duplicate previous accelerometer studies (6,17); however, the CSA monitor can also be placed on the ankle or wrist. In Melanson and Freedson's adult study (16), activity monitors were placed on the wrist, hip, and ankle. Interestingly, the best energy expenditure prediction equation was obtained using output from all three CSA monitors (multiple R-squared = 0.92; SEE 0.85 Kcal·min-1), while the best prediction equation using a single monitor was obtained using output from the CSA unit worn at the wrist (multiple R-squared = 0.86; SEE = 1.05 Kcal·min-1). In the present study, we obtained a similar R-squared and a slightly smaller SEE (multiple R-squared = 0.83; SEE = 0.97 kcal·min-1) for the prediction of energy expenditure using a one monitor hip placement. Considering our study population of children, this represents an important finding, as activity monitors worn at the wrist may increase subject reactivity, increase the risk of tampering, and in field studies may inadvertently capture extraneous arm movements. With respect to the number of activity monitors worn, it appears that wearing three monitors may slightly improve the prediction of energy expenditure during treadmill walking/running. However, this small improvement in energy expenditure prediction does not appear to warrant the extra cost and subject burden associated with wearing multiple activity monitors. Thus, we recommend a single activity monitor be used among children; however, future validation studies should confirm the optimal location for the CSA monitor for measurement of habitual physical activity in youth.
Our regression equation for estimating energy expenditure from CSA counts and body weight predicted mean energy expenditure within 0.01 kcal·min-1, and the correlation between mean actual and predicted energy expenditure was 0.93 (P < 0.001). However, the SEE was quite large at approximately 1 kcal·min-1, and individual differences between actual and predicted energy expenditure ranged from -2.5 to 3.4 kcal·min-1. Consequently, our energy expenditure prediction equation appears to be appropriate for estimating the mean energy expenditure of a group but should not be used to predict energy expenditure in individuals. For the latter purpose, researchers should calibrate the CSA monitor to energy expenditure on an individual basis.
In summary, the CSA activity monitor appears to be a reliable and valid tool for quantifying treadmill walking and running in children aged 10 to 14 yr. However, larger cross validation studies are needed before more definitive conclusions can be made regarding the validity of the CSA monitor in estimating energy expenditure.
Address for correspondence: Stewart G. Trost, Department of Exercise Science, School of Public Health, University of South Carolina, Blatt P.E. Center, 1300 Wheat Street, Columbia, SC
1. Al-Hazza, M. A. Sulaiman, A. J. Al-Matar, and K. F. Al-Mobaireek. Cardiorespiratory fitness, physical activity
patterns and coronary risk factors in preadolescent boys. Int. J. Sports Med.
2. Alpert, B. S. and J. H. Wilmore. Physical activity
and blood pressure in adolescents. Pediatr. Exerc. Sci.
3. Armstrong, N. and B. Simons-Morton. Physical activity
and blood lipids in adolescents. Pediatr. Exerc. Sci.
4. Bar-Or, O. and T. Baranowski. Physical activity
, adiposity, and obesity among adolescents. Pediatr. Exerc. Sci.
5. Bailey, D. A. and A. D. Martin. Physical activity
and skeletal health in adolescents. Pediatr. Exerc. Sci.
6. Bray, M. S., J. R. Morrow, J. M. Pirvarnik, and J. T. Bricker. Caltrac validity for estimating caloric expenditure with children.Pediatr. Exerc. Sci.
7. Bray, M. S., W. W. Wong, J. R. Morrow, N. F. Butte, and J. M. Pivarnik. Caltrac versus calorimeter determination of 24-h energy expenditure
in female children and adolescents. Med. Sci. Sports Exerc.
8. Bouten, C. V., K. R. Westerterp, M. Verduin, and J. D. Janssen. Assessment of energy expenditure
for physical activity
using a triaxial accelerometer
. Med. Sci. Sports Exerc.
9. Calfas, K. J., and W. C. Taylor. Effects of physical activity
on psychological variables in adolescents. Pediatr. Exerc. Sci.
10. Computer Science, and Applications, Inc. Wrist Activity Monitor Users Manual Model 7164
11. Escobedo, L. G., S. E. Marcus, E. Holtzman, and G. A. Giovino. Sports participation, age at smoking initiation, and the risk of smoking among US high school students. J. A. M. A.
12. Freedson, P. S. Electronic motion sensors and heart rate as measures of physical activity
in children. J. School Health
13. Haskell, W. L., M. C. Lee, A. Evans, and P. J. Irby. Simultaneous measurement of heart rate and body motion to quantitative physical activity
. Med. Sci. Sports Exerc.
14. Janz, K. F. Validation of the CSA accelerometer
for assessing children's physical activity
. Med. Sci. Sports Exerc.
15. Malina, R. M. Physical activity
and fitness of children and youth: Questions and implications. Med. Exerc. Nutr. Health
16. Melanson, E. L. and P. S. Freedson. Validity of the Computer Science and Applications, Inc. (CSA) activity monitor. Med. Sci. Sports Exerc.
17. Mukeshi, M. B., B. Gutin, W. Anderson, P. Zybert, and C. Basch. Validation of the Caltrac movement sensor using direct observation in young children. Pediatr. Exerc. Sci.
18. Pate, R. R. Physical activity
assessment in children and adolescents. Crit. Rev. Food Sci. Nutr.
19. Pate, R. R., B. J. Long, and G. W. Heath. Descriptive epidemiology of physical activity
in adolescents. Pediatr. Exerc. Sci.
20. Pate, R. R., M. Pratt, S. N. Blair, et al Physical activity
and public health. A joint recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine.J. A. M. A.
21. Public Health Service. Healthy People 2000: National Health Promotion and Disease Prevention Objectives
(DHHS Publication No. PHS
91-50212): Washington, DC: U.S. Department of Health and Human Services, 1990.
22. Redmond, D. P. and F. W. Hegge. Observations on the design and specification of a wrist-worn human activity monitoring system.Behav. Res. Methods Instrum. Comp.
23. Sallis, J. F. and K. Patrick. Physical activity
guidelines for adolescents: Consensus statement. Pediatr. Exerc. Sci.
24. Simons-Morton, B. G., N. M. O'Hara. D. G. Simons-Morton, and G. S. Parcel. Children and fitness: a public health perspective. Res. Q. Exerc. Sport
25. Thorland, W. G. and T. B. Gilliam. Comparison of serum lipids between habitually high and low active pre-adolescent males.Med. Sci. Sports Exerc.
26. Ward, D. S. and R. Evans. Physical activity
, aerobic fitness, and obesity in children. Med. Exerc. Nutr. Health
27. Webster, J. B., S. Messin, D. J. Mullaney, and D. F. Kripke. Transducer design and placement for activity recording. Med. Biol. Eng. Comp.
28. Welk, G. J. and C. B. Corbin. The validity of the Tritrac-R3D activity monitor for the assessment of physical activity
in children. Res. Q. Exerc. Sport