The use of objective methods to assess physical activity is increasingly prevalent (4). Accelerometers are particularly useful as they record intensity, frequency, and duration aspects of activity as well as the total volume of activity accumulated. Movement is assessed directly, with minimal hindrance to the user, making accelerometers ideal for use with children as well as adults (12,17). Recognized limitations include the inability to discern increases in energy cost due to walking/running up an incline, static work, upper-body movements, and carrying loads (1,4,16,17).
Both uniaxial and triaxial accelerometers are commercially available. However, research has indicated that the Tritrac triaxial accelerometer (Reining International, WI) is superior to a uniaxial accelerometer when assessing a variety of typical children’s activities against the criterions of oxygen consumption (6) and heart rate (12). Activity data can be expressed as minutes spent in varying intensities of physical activity for the Tritrac-R3D accelerometer (6,18) using published count cut-off values. The extent to which these cut-off values can be used in the field depends on how representative of normal physical activity the laboratory activities are. This highlights the importance of including a variety of activities when validating accelerometers (e.g., 3,6,7,20) rather than assessing treadmill activities alone. Additionally, accelerometer output is affected by anthropometric measures, for example, leg length and height (22). Therefore, cut-off values need to be population specific. This highlights the need to validate accelerometers in both children and adults.
Recently, Stayhealthy Inc. (Monrovia, CA) purchased the technology and rights to the Tritrac from Reining International. They have developed a smaller triaxial accelerometer called the RT3. This unit has been shown to be reliable (14), although studies have indicated that the anteroposterior vector reads higher than the vertical and mediolateral vectors for a given frequency and amplitude of movement (13). The manufacturers claim the output (counts recorded) from the RT3 is comparable to the output from the Tritrac, thereby allowing comparisons between studies using the different monitors (www.stayhealthy.com). However, to date, there are no published studies concerning the validity of the RT3 and the comparability of the output with the Tritrac. Therefore, the aims of this study were to assess and compare the validity of the RT3 accelerometer for the assessment of physical activity in boys and men, to compare RT3 and Tritrac accelerometry counts (across all three vectors and vector magnitude) across a variety of activities in boys, and to determine count cut-off values for moderate (≥3 < 6 METs) and vigorous (≥6 METs) for boys and men.
METHODS
Participants
The participants were 19 boys (mean age ± SD = 9.5 ± 0.8 yr, mass 33.5 ± 5.4 kg, height 137.8 ± 7.1 cm) and 15 men (age 20.7 ± 1.4 yr, mass 70.1 ± 6.6 kg, height 176.7 ± 5.4 cm). The boys were recruited from two primary schools in North Wales, and the men were students at the University of Wales, Bangor. Written informed consent was obtained from all participants and, in addition, from parents/guardians of the boys. The ethics review board of the School of Sport, Health and Exercise Sciences, University of Wales, Bangor granted approval for this study.
Procedure
The relationship between accelerometry counts and oxygen consumption was assessed during four treadmill speeds (4, 6, 8, and 10 km·h−1) and three nonregulated activities (hopscotch, kicking a ball to and fro, and sitting quietly). All activities lasted 4 min, except the seated activity, which lasted 10 min.
Before testing, each participant was habituated to the treadmill (Powerjog JM200, Birmingham, UK) for 5 min. After habituation, each participant rested to allow heart rate to return to baseline. The participant then sat quietly for 10 min while playing a keyboard computer game (boys) or completing a crossword (men). Participants then walked at 4 km·h−1 and 6 km·h−1 and ran at 8 km·h−1 and 10 km·h−1 on the motor driven treadmill. After a rest to allow heart rate to return to baseline, the participant kicked a soccer ball to and fro with one of the investigators (distance: boys = 2.4 m, men = 3 m). After the return of heart rate to baseline, participants alternately hopped and jumped on a hopscotch grid. During the latter two activities, participants established their own pace and this pace was maintained throughout the 4-min period.
Measures
RT3 triaxial accelerometer.
The RT3 accelerometer (Stayhealthy Inc.) measures activity in three dimensions. With the RT3 worn on the hip, the vectors are as follows: vertical (x), anteroposterior (y), and mediolateral (z). Size and mass (including battery) of the RT3 are 7.1 × 5.6 × 2.8 cm, 65.2 g. The RT3 was attached to a belt, using its integral belt clip, and worn on the right hip of the participants. The epoch interval was set at one min and output was expressed as mean counts per minute for each activity. The same RT3 accelerometer was used by all participants.
Tritrac triaxial accelerometer.
The Tritrac-R3D accelerometer (model T303A, Reining International Professional Products) measures activity in three dimensions. With the Tritrac worn on the hip, the vectors are as follows: mediolateral (x), vertical (y), and anteroposterior (z). Size and mass (including battery) of the Tritrac are 10.8 × 6.8 × 3.3 cm, 170.4 g. Tape was used to firmly secure the Tritrac unit to the belt of the boys, above the left hip. Tape was necessary to prevent any extraneous movement. The epoch interval was set at 1 min, and output was expressed as mean counts per minute for each activity. The same Tritrac accelerometer was used by all participants. The men did not wear a Tritrac accelerometer.
Heart rate.
Heart rate was measured using the Polar Sport Tester (Electro Ltd., Kempele, Finland). The transmitter belt was worn around the chest and the receiver was held by the researcher. Heart rate was recorded at 20, 40, and 60 s during the final minute of each activity. The mean of the three readings was recorded. The three readings were also used to confirm steady state had been reached.
Gas analysis.
Expired air was collected using Douglas Bags (Harvard Apparatus Ltd., Edenbridge, Kent, UK) and analyzed using a CO2/O2 analyser (Hitech-GIR250, Luton, UK) and a dry gas meter (Harvard Apparatus Ltd.). Boys wore a pediatric face mask (Hans Rudolph pediatric mask, Kansas City, MO), and men used a standard mouthpiece. The mask or mouthpiece was connected to a Douglas Bag by plastic tubing (length 1.52 m, internal diameter 2.5 cm). The mask/mouthpiece was worn throughout each activity; however, expired air was only collected during the final minute of the activity, when the participant was at steady state. Recordings of all measures were referenced to the same watch to allow the retrieved data to be matched temporally.
Data Analysis
Steady state oxygen consumption and heart rate data were used in all analyses. To account for the differences in body size, oxygen consumption was expressed relative to body mass raised to the power of 0.75 (S V̇O2). This exponent was recommended by Rogers et al. (15) for use when comparing prepubertal children with adults performing similar activities. Nonsignificant correlations between scaled oxygen consumption and mass for boys alone (r = 0.026), men alone (r = −0.034), and the whole group (r = −0.052) confirmed that mass had been factored out by the scaling.
Output from the RT3 and the Tritrac was analyzed for each vector individually (vertical, anteroposterior and mediolateral) and for the vector magnitude (calculated as VM = (x2 + y2 + z2)0.5). This allowed comparison of individual vectors between units, and also investigation of whether the vector magnitude from the RT3 accelerometer is superior to the vertical vector (VERT) alone, as appears to be the case for the Tritrac (6).
Descriptive statistics were calculated for all output measures. Shapiro-Wilk tests of normality showed all variables were normally distributed. The concurrent validity of the RT3 and the Tritrac (boys only) with the criterion measure of S V̇O2 was investigated using Pearson’s product moment correlation coefficients for boys alone, men alone, and the whole group. Planned comparisons of correlation coefficients were carried out for the following pairs of correlations using the method described by Meng et al. (10). This method of comparison accounts for the correlation between the two related correlation coefficients.
Boys
- RT3VERT and S V̇O2, RT3VM and S V̇O2
- TritracVERT and S V̇O2, TritracVM and S V̇O2
- RT3VM and S V̇O2, TritracVM and S V̇O2.
Men
- RT3VERT and S V̇O2, RT3VM and S V̇O2.
Whole group
- RT3VERT and S V̇O2, RT3VM and S V̇O2.
Planned cross-group comparisons were carried out for the following pairs of correlation coefficients, using Fisher z transformations.
- RT3VM and S V̇O2 boys, RT3VM and S V̇O2 men
- RT3VERT and S V̇O2 boys, RT3VERT and S V̇O2 men.
Differences in RT3 counts by group and by activity were investigated using a two-way, activity (7) by group (2) mixed model ANOVA. Differences between Tritrac and RT3 counts (vertical, anteroposterior, and mediolateral vectors) when assessing the boys’ activities were explored using a three-way fully repeated vector (3) by accelerometer (2) by activity (7) ANOVA. Where necessary, the Greenhouse Geisser correction was used to correct for violations of the assumption of sphericity. Significant effects were followed up using a Tukey’s test adapted for mixed model or repeated measures ANOVA, as appropriate.
As body size variables may affect the relationship between S V̇O2 and RT3 accelerometer counts multiple regression analysis was used to investigate whether height or mass explained any additional variance in S V̇O2 to that explained by RT3 accelerometer counts. This was carried out on all activity data, treadmill activity data and unregulated activity data for the whole group.
Alpha was set at 0.05; where stated, alpha was adjusted to account for multiple tests and control Type 1 error.
RESULTS
Correlational analyses.
The descriptive statistics, broken down by group and activity, are shown in Table 1. RT3 accelerometer counts correlated positively with S V̇O2 (P < 0.01) across all activities for boys alone, men alone, and the whole group (Table 2). Similarly, Tritrac counts correlated positively with S V̇O2 (P < 0.01) for boys alone (Table 2). When treadmill activities and unregulated activities were considered separately, the relationships remained significant (Figs. 1 and 2).
;)
TABLE 1:
Descriptive statistics by group and by activity.
TABLE 2:
Correlations between accelerometer counts and S V̇O2.
FIGURE 1:
Relationship between RT3 counts and S V̇O2 during walking and running in boys and men.
FIGURE 2:
Relationship between RT3 counts and S V̇O2 during unregulated activities in boys and men.
When examining all activities combined, correlations between S V̇O2 and RT3VM or RT3VERT did not differ significantly between men and boys (Figs. 1 and 2). The RT3VM was a significantly better predictor of S V̇O2 than the RT3VERT when boys were assessed alone (P < 0.05). However, this was not the case when assessing men or the whole group. The Tritrac vector magnitude was not superior to the Tritrac vertical vector for boys alone. Correlations between RT3 and S V̇O2 did not differ from correlations between the Tritrac and S V̇O2 (Table 2).
Group differences.
ANOVA revealed there was an activity by group interaction for RT3 counts (F3.0, 90.4 = 35.4, P < 0.01). Posthoc analyses showed that boys had higher RT3 counts than men for all treadmill activities, but there were no significant differences between groups for the unregulated activities (Table 1). All activities were significantly different from each other.
Multiple regression analysis indicated that body size (either height or mass) added a small but significant amount to the variance explained in S V̇O2 when all activities (R2 change = 0.006 (height), 0.008 (mass), P < 0.05), or treadmill activities alone (R2 change = 0.024 (height), 0.028 (mass), P < 0.01), were considered, explaining the group differences detected here. However, body size did not explain any further variance when assessing unregulated activities (Table 3). Group differences for activity cut-off points corresponding to 3 METs and 6 METs (thresholds commonly used to classify moderate and vigorous activity) were negligible when all activities or unregulated activities alone were considered. However, when treadmill activities were analyzed alone, activity cut-off points were approximately 400 counts·min−1 lower in men relative to boys (Table 4, Fig. 1).
TABLE 3:
Regression qnalyses: Prediction of S V̇O2 from RT3 accelerometry counts and body size.
TABLE 4:
Threshold RT3 counts (counts·min−1) relating to moderate (3 METs) and vigorous (6 METs) intensity activity.
Monitor differences by vector.
Comparison of the RT3 and Tritrac accelerometer counts across activities in the boys resulted in a significant vector by monitor by activity interaction (F2, 32 = 328.8, P < 0.01). This interaction was followed up with a series of three two-way repeated measures monitor (2) by activity (7) ANOVA, one for each vector. A significant monitor by activity interaction was found for the vertical vector (F2.8, 44.8 = 6.5, P < 0.001, Fig. 3A) and the anteroposterior vector (F2.4, 37.6 = 64.2, P < 0.001, Fig. 3B). On the mediolateral vector, there was a significant main effect for activity (F2.0, 32.0 = 245.2, P < 0.001, Fig. 3C) but not for monitor (F1, 16 = 1.00, P = 0.333), and there was no monitor by activity interaction (F2.3, 36.6 = 2.8, P = 0.066).
FIGURE 3:
A. Accelerometer counts (vertical vector) by monitor and by activity. B. Accelerometer counts (anteroposterior vector) by monitor and by activity. C. Accelerometer counts (mediolateral vector) by monitor and by activity.
Posthoc analyses revealed that on the vertical axis, RT3 counts were significantly higher than Tritrac counts during walking at 6 km·h−1, running at 10 km·h−1, and hopscotch (P < 0.05) but not during the remaining activities (Fig. 3A). On the anteroposterior axis, RT3 counts were significantly higher than Tritrac counts for all activities (P < 0.05), except sitting and kicking a ball to and fro (Fig. 3B). Regardless of the monitor used, accelerometry counts differed significantly between each activity for all three vectors (Figs. 3A–C).
DISCUSSION
Main findings.
The RT3 is a relatively new triaxial accelerometer that has replaced the Tritrac. This study tested the validity of the RT3 across a variety of treadmill and unregulated activities against the criterion of oxygen consumption. Our data demonstrate that the RT3 appears to provide an equally valid assessment of activity in both boys and men. This was true for treadmill activities alone, unregulated activities alone, or all activities combined. However, results indicate that, for any given activity, output from the RT3 is higher than output from the Tritrac.
Relationship between accelerometer counts and SV̇O2.
The strength of the relationship between oxygen consumption and Tritrac or RT3 counts was similar to that identified by previous research with triaxial accelerometers (3,6,7) and uniaxial accelerometers (9,11). There is some evidence that the triaxial nature of the RT3 offers an advantage to the vertical vector alone in boys, as identified for the Tritrac by Eston et al. (6) and the Tracmor by Bouten et al. (3). However, this was not the case for the Tritrac in the current study or the RT3 when used with men.
Considering the nature of the unregulated activities, the correlations between accelerometer counts and S V̇O2 were relatively high. Accelerometers have been shown to overestimate the energy cost of sedentary activities (19) and underestimate the energy cost of high intensity activities (8). Closer examination of the scatterplot for unregulated activities reveals some clumping of activities at the low intensity range. This was caused by very low counts per minute output for the sedentary activity. In similar studies (6,7), counts for seated activities and catch had been clumped at the lower end of the scale. In the current study, kicking a ball was included to try and spread the scores, but a gap is still evident. This will have inflated the correlation coefficients. However, we feel that inclusion of sedentary and light activities is very important when validating accelerometers, as the majority of the waking day is spent in sedentary activities in both children and adults.
Some activities with similar oxygen or heart rate demands elicited very different accelerometer outputs. Hopscotch resulted in significantly lower RT3 counts than running at 8 and 10 km·h−1, although oxygen consumption for hopscotch was comparable to that from the running activities. Similarly, previous studies have reported accelerometer counts that are relatively low, considering the metabolic demand, for basketball and hopscotch activities (6,7,12). This is most likely related to the strenuous jumping nature of these activities requiring more energy per movement than running. Accelerometers are unable to assess this type of activity accurately, along with upper-body movement, carrying loads, and traveling up an incline (1,9,16,17,19).
Group differences.
RT3 accelerometer counts correlated strongly and significantly with S V̇O2 in both boys and men. However, the counts corresponding to any given S V̇O2 were higher for boys during treadmill activities than for men. This is due to anthropometric differences between the groups (22). Body size (height or mass) added significantly to the variance explained in S V̇O2 by RT3 counts for locomotor activities, reflecting the greater stride frequency in boys compared with men. For any given speed, stride frequency will be higher in boys, due to relatively shorter legs (21), but oxygen consumption (independent of body size) will be the same (5). This was reflected in the lack of differences in S V̇O2 between boys and men, yet higher RT3 counts for all locomotor activities. Boys’ heart rate, S V̇O2, and Tritrac data for the various activities compares well with previous research utilizing similar procedures (6,7).
Moderate and vigorous intensity cut-off values.
To our knowledge, these are the first published cut-off values available for the RT3 accelerometer. Values are presented for boys and men; these do not differ by group for all activities combined or unregulated activities alone. However, the relatively lower accelerometer counts at a given S V̇O2 for boys compared with men when assessing locomotor activities is reflected in the count cut-off values identified for moderate and vigorous intensity. Therefore, it is important for researchers’ to consider what types of activity are going to be assessed as well as using population specific cut-off values. Additionally, research indicates that the relationship between metabolic demand and accelerometer counts differs when assessed in the laboratory and the field, even when only simple activities like walking are assessed (11). Several accelerometer cut-off points are available for the CSA accelerometer. Strath et al. (19) demonstrated that estimates of times spent in different intensities of physical activity varied according to the cut-off values used.
Comparability of RT3 and Tritrac.
The manufacturers of the RT3 claim it is more precise and accurate than the Tritrac due to the use of an integrated triaxial accelerometer as opposed to three separate accelerometers. On their website (www.stayhealthy.com), they provide comparative data for the Tritrac and the RT3. Their data show the RT3 counts are consistently higher than the Tritrac counts (approximately 130 counts difference), but the manufacturers stated that this would be corrected before release. The results of this study indicate that this was not the case. Therefore, output from the two monitors is not comparable. An investigation of the Tritrac and RT3 results by axis indicates that the differences are largely due to the anteroposterior vector. Powell et al. (13) also reported that the anteroposterior vector read consistently higher than the vertical and mediolateral vector when all three vectors in turn were subjected to a series of identical vibration frequencies in a test jig. When conducting any study using multiple RT3, their inter-unit variability, ideally across each axis, should be assessed.
CONCLUSIONS
The RT3 is a valid tool for the assessment of physical activity in men and boys. The smaller size and more user-friendly software of the RT3, in comparison with the Tritrac, make it an attractive option for researchers requiring a triaxial accelerometer. It is as good a predictor of oxygen consumption as the Tritrac, although the resulting activity counts from the two monitors are not comparable. When assessing locomotor activities, body size adds significantly to the variance explained in oxygen consumption by activity counts. Hence, for these activities moderate and vigorous intensity cut-off values differ between men and boys. Further research is required regarding the identification and efficacy of cut-off values corresponding to moderate and vigorous intensity activities and how these differ by population, environment, and dominant activity.
REFERENCES
1. Bassett, D. R. Jr. Validity and reliability issues in objective monitoring of physical activity. Res. Q. Exerc. Sport. 71: 30–36, 2000.
2. Bassett, D. R. Jr., B. E. Ainsworth, A. M. Swartz, S. J. Strath, W. L. O’Brien, and G. A. King. Validity of four motion sensors in measuring moderate intensity, physical activity. Med. Sci. Sports Exerc., 32: S471–S480, 2000.
3. 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. 26: 1516–1523, 1994.
4. Dale, D., G. J. Welk, and C. E. Matthews. Methods for assessing physical activity and challenges for research. In: Physical Activity Assessments for Health-Related Research, G. J. Welk (Ed.). Champaign, IL: Human Kinetics, 2002, pp. 19–34.
5. Eston, R. G., S. Robson, and E. Winter. A comparison of oxygen uptake during running in children and adults. In: Kinanthropometry IV, W. Duquetand J. Day (Eds.). London: E and FN Spon, 1993, pp. 236–241.
6. 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.
7. Louie, L., R. G. Eston, A. V. Rowlands, K. K. Tong, D. K. Ingledew, and F. H. Fu. Validity of heart rate, pedometry and accelerometry for estimating the energy cost of activity in Hong Kong Chinese boys. Pediatr. Exerc. Sci. 11: 229–239, 1999.
8. Meijer, G. A., K. R. Westerterp, H. Koper, and F. ten Hoor. Assessment of energy expenditure by recording heart rate and body acceleration. Med. Sci. Sports Exerc. 221: 343–347, 1989.
9. Melanson, E. L., and P. S. Freedson. Validity of the Computer Science and Applications, Inc. (CSA) activity monitor. Med. Sci. Sports Exerc. 27: 1071–1078, 1995.
10. Meng, X. L., R. Rosenthal, and D. B. Rubin. Comparing correlated correlation coefficients. Psychol. Bull. 111: 172–175, 1992.
11. Nichols, J. F., C. G. Morgan, L. E. Chabot, J. F. Sallis, and K. J. Calfas. Assessment of physical activity with the computer science and applications, Inc. accelerometer: laboratory versus field validation. Res. Q. Exerc. Sport 71: 36–43, 2000.
12. Ott, A. E., R. R. Pate, S. G. Trost, D. S. Ward, and R. Saunders. The use of uniaxial and triaxial accelerometers to measure children’s free play physical activity. Pediatr. Exerc. Sci. 12: 360–370, 2000.
13. Powell, S. M. P., D. I. Jones, and A. V. Rowlands. Technical variability of the RT3 accelerometer. Med. Sci. Sports Exerc. 35: 1773–1778, 2003.
14. Powell, S. M. P., and A. V. Rowlands. Inter-monitor variability of the RT3 accelerometer during a variety of physical activities. Med. Sci. Sports Exerc. 36: 324–330, 2004.
15. Rogers, D. M., B. L. Olson, and J. H. Wilmore. Scaling for the VO
2-to-body size relationship among children and adults. J. Appl. Physiol. 79: 958–967, 1995.
16. Rowlands, A. V. Field measures of assessing physical activity and energy balance. In: Exercise Physiology and Kinanthropometry Laboratory Manual: Tests, Procedures and Data, Vol. 1. Anthropometry, R. G. Estonand T. Reilly (Eds.). London: Routledge, 2001, pp. 151–170.
17. Rowlands, A. V., R. G. Eston, and D. K. Ingledew. Measurement of physical activity in children with particular reference to the use of heart rate and pedometry. Sports Med. 24: 258–272, 1997.
18. Rowlands, A. V., R. G. Eston, and D. K. Ingledew. The relationship between activity levels, body fat and aerobic fitness in 8–10 year old children. J. Appl. Physiol. 86: 1428–1435, 1999.
19. Strath, S. J., D. R. Bassett Jr., and A. M. Swartz. Comparison of MTI accelerometer cut-points for predicting time spent in activity. Int. J. Sports Med. 24: 298–303, 2003.
20. Swartz, A. M., S. J. Strath, D. J. Bassett Jr., W. L. O’Brien, G. A. King, and B. E. Ainsworth. Estimation of energy expenditure using CSA accelerometers at hip and wrist sites. Med. Sci. Sports Exerc. 32: S450–S456, 2000.
21. Unnithan, V. B., and R. G. Eston. Stride frequency and submaximal treadmill running economy in adults and children. Perdiatr. Exerc. Sci. 2: 149–155, 1990.
22. Welk, G. J. Use of accelerometry-based activity monitors to assess physical activity. In: Physical Activity Assessments for Health-Related Research, G. J. Welk (Ed.). Champaign, IL: Human Kinetics, 2002, pp. 125–141.
