Journal of Physical Activity and Health, 2011, 8(Suppl 2), S295 -S305 2011 Human Kinetics, Inc. Convergent Validity of Pedometer and Accelerometer Estimates of Moderate-to-Vigorous Physical Activity of Youth Michael William Beets, Charles F. Morgan, Jorge A. Banda, Daniel Bornstein, Won Byun, Jonathan Mitchell, Lance Munselle, Laura Rooney, Aaron Beighle, and Heather Erwin Background: Pedometer step-frequency thresholds (120 steps min -1, SPM) corresponding to moderate-tovigorous intensity physical activity (MVPA) have been proposed for youth. Pedometers now have internal mechanisms to record time spent at or above a user-specified SPM. If pedometers provide comparable MVPA (P-MVPA) estimates to those from accelerometry, this would have broad application for research and the general public. The purpose of this study was to examine the convergent validity of P-MVPA to accelerometer-mvpa for youth. Methods: Youth (N = 149, average 8.6 years, range 5 to 14 years, 60 girls) wore an accelerometer (5-sec epochs) and a pedometer for an average of 5.7 ± 0.8 hours day -1. The following accelerometer cutpoints were used to compare P-MVPA: Treuth (TR), Mattocks (MT), Evenson (EV), Puyau (PU), and Freedson (FR) child equation. Comparisons between MVPA estimates were performed using Bland-Altman plots and paired t tests. Results: Overall, P-MVPA was 24.6 min ± 16.7 vs. TR 25.2 min ± 16.2, MT 18.8 min ± 13.3, EV 36.9 min ± 21.0, PU 22.7 min ± 15.1, and FR 50.4 min ± 25.5. Age-specific comparisons indicated for 10 to 14 year-olds MT, PU, and TR were not significantly different from P-MVPA; for the younger children (5 8 year- olds) P-MVPA consistently underestimated MVPA. Conclusions: Pedometer-determined MVPA provided comparable estimates of MVPA for older children (10 14 year-olds). Additional work is required to establish age appropriate SPM thresholds for younger children. Keywords: children, adolescents, measurement During the past decade, pedometers have increasingly been used in physical activity surveillance, 1 and as an outcome measure in clinical 2 and behavioral interventions. 3 The popularity of pedometers can be attributed to their ease-of-use, low cost and demonstrated convergent (ie, related to other measures of physical activity) and construct (ie, related to other theoretically-related constructs age, gender, health status) validity. 4,5 Numerous studies across diverse populations indicate pedometers exhibit a high degree of accuracy during moderate and fast walking (ie, registered vs. observed steps), 6 10 and higher steps day -1 are associated with known health benefits (eg, cardiovascular fitness). 11 Further, total daily steps are moderate-to-highly correlated to other objective measures of physical activity (eg, accelerometers, heart Beets, Banda, Bornstein, Byun, Mitchell, and Rooney are with the Dept of Exercise Science, Arnold School of Public Health, University of South Carolina, Columbia, SC. Morgan and Munselle are with the Dept of Kinesiology and Rehabilitation Science, University of Hawaii, Honolulu, HI. Beighle and Erwin are with the Dept of Kinesiology and Health Promotion, University of Kentucky, Lexington, KY. rate monitor, observation). 12 16 Hence, pedometers exhibit many of the requisite characteristics for widespread acceptance as a reliable and valid measurement device to monitor physical activity. One of the major criticisms of pedometers is their inability to provide an estimate of moderate-to-vigorous physical activity (MVPA). Recently, step frequency (steps min -1 ) guidelines for youth 17 have been proposed that are related to the lower threshold for moderate intensity physical activity. This is based on the notion that step frequency is related to the amount of energy expended. For example, walking at 50 steps min -1 would have a lower metabolic equivalent value than walking at 150 steps min -1. For youth, an estimate of 120 steps min -1 has been proposed to be the lower limit of moderate intensity physical activity. 17 While this threshold is viewed as preliminary, it serves as a natural starting point to determine if this threshold can provide a valid assessment of MVPA in free-living settings. Technological advancements now allow pedometers to record the amount of time (hrs:min:sec) the wearer spends at or above a prespecified steps min -1 frequency in addition to the common pedometer output of total daily steps and physical activity time. 10 Thus, every time S295
S296 Beets et al the wearer is at or surpasses the designated steps min -1 threshold, the time spent at this step frequency is recorded on an MVPA timer (analogous to the activity timer evaluated in previous studies). 10 The information derived for time in MVPA is instantly available to the wearer via a digital display. The utility of this function is that individuals will immediately know their accumulated MVPA time and this could be used to determine accumulation of recommended daily MVPA levels. If the suggested threshold of 120 steps min -1 for youth provides a comparable estimate of MVPA in relation to accelerometry ( gold standard ), then this would have broad application for both researchers and the general public. The purpose of this study was therefore to examine the convergent validity of pedometer-derived MVPA in relation to MVPA estimates from accelerometry. Participants Methods Youth attending summer camp in 3 locations South Carolina, Kentucky, and Hawaii were recruited to participate in the study. Parents and children were informed through summer camp staffers and summer camp information sessions about the purpose of the study and invited to participate. Summer camps were physical activitybased, with youth participating in a variety of structured (eg, team sports), semistructured (eg, capture the flag), and free-play (eg, playground) activities throughout the day. This setting was selected due to the wide range of activities of varying intensity the youth would participate on a daily basis. Children were not eligible to participate if they had a physical and/or health impairment (eg, physical disability, respiratory disorder) that limited their ability to engage in physical activity. Participation was voluntary and based on a first come, first served basis. Participants self-reported their date of birth and race/ ethnicity. All procedures were approved from each site s Institutional Review Board and summer camp directors. Written parental consent and verbal assent were obtained from each participant before data collection. Anthropometry Height and weight of each participant were measured using a portable stadiometer (nearest 0.1 cm) and digital scale (nearest 0.01 lbs, both measures taken without shoes). Units were transformed and body mass index computed to derive BMI classifications according to age-sex specific CDC percentiles. Stride length was collected to control for the effects of step frequency due to leg length. Stride length is inversely related to the number of steps accrued during a given time frame (ie, 1 minute) hence, the longer the stride, the less steps taken during a minute. Stride was collected by counting the number of steps each participant took while crossing a 10 m distance. Participant characteristics are presented in Table 1. Instrumentation The Walk4Life MVPA pedometer (WL, Plainfield, IL) was used to measure MVPA via steps min -1 (referred to as P-MVPA). This pedometer is similar to previously validated Walk4Life models and includes a separate timer, called the MVPA timer which records movement at or above a prespecified step frequency. 6,10 For all units, the steps min -1 threshold was set at 120 steps min -1, in accordance with previous research with this age group. 17 The MVPA timer registers time (hr:min:sec) based on a step frequency ratio, in this case 120 steps min -1. Thus, when the wearer moves for any duration of time at or above the ratio, the time is recorded in the MVPA timer. For instance, if a person took 20 steps in 10 seconds, Table 1 Participant Characteristics Total (N = 149) Girls (n = 60) Boys (n = 89) Mean SD Mean SD Mean SD Age (yrs) 8.6 ±1.7 8.7 ±1.8 8.5 ±1.6 Height (cm) 131.8 ±12.6 132.7 ±14.2 131.2 ±11.4 Weight (kg) 33.3 ±12.2 34.8 ±14.0 32.3 ±10.7 BMI 18.6 ±4.0 19.0 ±4.5 18.4 ±3.6 Stride a 17.7 ±2.6 17.7 ±2.7 17.7 ±2.5 Total accelerometer wear time (min) 341.0 ±48.7 340.3 ±57.3 341.5 ±42.2 Overweight/obese b 34.2% 35.0% 33.7% African American 32.8% 33.3% 32.6% White non-hispanic 32.2% 38.3% 27.0% a Number of steps taken across 10 meters. b At or above the 85th age-sex specific BMI percentile. Abbreviations: SD, standard deviation.
Pedometer MVPA S297 the corresponding time would be recorded on the MVPA timer (120 steps divided by 60 seconds equals 2 steps sec -1, which translates to 20 steps within 10 seconds to be at the prespecified threshold). The step filter was set at 0 (ie, counts all movement). Before data collection, the shake test 18 was performed to ensure the pedometers were calibrated and operating accurately. The ActiGraph GT1M accelerometer (Shalimar, FL) was used to assess physical activity intensity levels and was designated as the gold-standard measure of MVPA in the current study. The accelerometers were initialized to collect data using 5-second epoch lengths to account for the sporadic, intermittent nature of children s physical activity. 19 Five commonly used accelerometry-based physical activity intensity cutpoints were used to derive accelerometer MVPA time; these cutpoints are specific to children and adolescents. All 5 were chosen because there is considerable population specificity of accelerometer cutpoints when categorizing physical activity intensity levels of youth. 20,21 The selected cutpoints were Evenson (EV) for ages 5 to 8 years, 22 Freedson child equation (FR) for ages 6 to 18 years, 23 Mattocks (MT) for age 12 years, 24 Puyau (PU) for ages 6 to 16 years, 25 and Treuth (TR) for ages 13 to 14 years. 26 Since the cutpoints were established using different epochs (eg, 15 sec, 60 sec), the cutpoints were divided by the appropriate denominator to derive a 5-second epoch cutpoint equivalent. This procedure has been used extensively to accommodate differing cutpoint epoch vs. epoch of measurement. 27 29 Protocol Upon arrival at the summer camp, participants height and weight were measured by trained research staff. After completion, an accelerometer and pedometer (preset to the specifications above and reset to 0) were affixed to each participant s waist, at the right hip, with an elastic belt. Each pedometer was sealed with a plastic zip-tie to reduce tampering. To minimize movement of the devices around the waist due to belt slippage and to mitigate tilting of the devices, fasteners were used to stabilize the belt on the participants waistband of their pants/ shorts. Concurrently with placing the devices on the waist, a research staffer recorded the time of attachment (hr:min:sec). After this was recorded, the participants were free to take part in their regular scheduled activities. At the end of each day, the research staff removed the belts from the participants and immediately recorded the time of detachment and the pedometer steps, activity time (hr:min:sec) and, P-MVPA (hr:min:sec). Data Reduction Initially, accelerometer counts were uploaded into a userwritten macro to remove any nonwear time strings of 10 minutes or longer of 0 counts. The morning attachment time and afternoon detachment time (see above) were used to clean the accelerometer data of any aberrant activity counts before and after monitor attachment/ detachment. Counts were then transformed into intensity levels corresponding to the 5 studies 22 26 and summarized into total MVPA time. Statistical Analysis Comparison Between MVPA Estimates All statistical analyses were conducted using Stata (v.10.1, College Station, Texas). Bland-Altman plots 30 were constructed that contrasted P-MVPA with MVPA estimated from each of the 5 accelerometer cutpoints. Accelerometer MVPA estimates were subtracted from P-MVPA so that positive numbers on the y-axis indicated a higher estimate by the P-MVPA, whereas a negative number on the y-axis indicated a higher estimate by the accelerometer MVPA. Pitman s Test of differences in the variance (ie, testing the difference between the mean between the 2 measuring devices and the slope of the difference) were computed for each of the 5 Bland- Altman plots. Paired sample t tests were computed to examine the correspondence of P-MVPA vs. each of the 5 accelerometer estimates of MVPA. To address the issue of age-specificity of the accelerometer cutpoints, separate Bland Altman plots and t test analyses were conducted on the EV cutpoints with the 5 to 8 year-olds in addition to the FR and PU cutpoints, and Mattocks and Treuth cutpoints with the 10 to 14 year-olds, in addition to the FR and PU cutpoints. Predictors of Difference Follow-up ordinary least squares regression analyses were performed with the entire sample (5 14 years) to examine predictors of the absolute value of the difference between P-MVPA and each of the 5 accelerometer cutpoints. 24 Predictors were selected based on physiological associations (eg, increased stride associated with decreased step frequency) and review of the Bland-Altman plots. The predictors in the model were: age, classification as obese ( 95th age-sex specific BMI percentile), stride, and physical activity level. Physical activity level was categorized into 3 equal tertiles based on total daily accelerometer counts. The moderate (n = 50) and high (n = 50) active categories were dummy coded (0/1) and included in the model with the low (n = 49) active category serving as the reference group. Results For the overall sample (5 14 years), the participants physical activity was assessed for an average of 341.0 min day -1 (see Table 1). Estimates of P-MVPA and MVPA derived from the 5 accelerometer cutpoints are presented in Table 2. For the total sample, P-MVPA was 24.6 min ± 16.7 in comparison with EV 36.9 min ± 21, FR 50.4 ± 25.5, MT 18.8 ± 13.3, PU 22.7 ± 15.1, and TR 25.2 ± 16.2. Paired t tests indicated a significant difference between
S298 Beets et al P-MVPA and MVPA estimated from EV (t = 9.16, P <.01), MT (t = 5.33, P <.01), and FR (t = 16.43, P <.01). Conversely, no significant differences were observed for PU (t = 1.76, P =.08) and TR (t =.51, P =.61). The Bland- Altman plots indicated that across the 5 comparisons, as MVPA increased, the difference between P-MVPA and accelerometer-derived MVPA also increased. This was supported by the Pitman s Test of differences in variance which indicated this was significant for EV (r =.286, P =.001), FR (r =.498, P <.01), and MT (r =.287, P <.01) comparisons. No significant differences in variance were observed for comparisons with PU (r =.135, P =.10) and TR (r =.04, P =.63). The mean difference (95% confidence intervals, 95% CI) and limits of agreement for were EV 12.3min (95% CI 14.9 to 9.6) and 45.1 to 20.5 min, FR 25.8 min (95% CI 28.9 to 22.7) and 64.1 to 12.5 min, MT 5.9 min (95% CI 3.7 to 8.0) and 20.9 to 32.6 min, PU 2.0 min (95% CI 0.2 to 4.2) and 25.5 to 29.4 min, and TR 0.6min (95% CI 2.9 to 1.7) and 28.7 to 27.6 min. For the old youth (10 14yr olds), paired t tests indicated no significant difference between P-MVPA and MT (t = 1.78, P =.08), PU (t = 0.53, P =.60), and TR (t = 1.86, P =.07). The Bland Altman plots showed similar patterns to those from the overall sample, with greater error observed with increasing MVPA (plots available upon request). The mean difference (95% CI) and limits of agreement were FR 21.6 min (95% CI 26.9 to 16.2) and 57.13 to 14.0, MT 2.9 min (95% CI 0.4 to 6.2) and 19.1 to 24.9 min, PU 0.9min (95% CI 4.41 to 2.58) and 24.17 to 22.3, and TR 3.4 min (95% CI 7.1 to 0.3) and 28.1 to 21.2 min. Pitman s Test were FR (r =.698, P =.001), MT (r =.03, P =.84), PU (r =.207, P =.191) and TR (r =.3, P =.06). Table 2 Pedometer-Determined MVPA vs. Accelerometer-Determined MVPA Using 5 Cutpoints for Youth Total Girls Boys Age group MVPA estimate (minutes) Mean SD Mean SD Mean SD 5 14 year-olds Sample size 149 60 89 Pedometer-determined MVPA (120 steps min -1 ) 24.6 ±16.7 18.7 ±11.4 28.7 ±18.6 Accelerometer-determined MVPA (total sample) Evenson et al 22 36.9 a ±21.0 28.7 ±16.8 42.5 ±21.8 Freedson et al 23 50.4 a ±25.5 39.6 ±20.0 57.7 ±26.3 Mattocks et al 24 18.8 a ±13.3 14.7 ±11.0 21.5 ±14.0 Puyau et al 25 22.7 ±15.1 17.6 ±12.4 26.1 ±15.8 Treuth et al 26 25.2 ±16.2 19.6 ±13.4 29.0 ±17.0 10 14 year-olds Sample size 45 20 25 Pedometer-determined MVPA (120 steps min -1 ) 19.9 ±11.3 18.3 ±10.7 21.3 ±11.8 Accelerometer-determined MVPA Freedson et al 23 41.5 b ±22.7 39.0 ±19.9 43.5 ±24.9 Mattocks et al 24 17.0 ±11.6 17.2 ±12.6 16.9 ±11.1 Puyau et al 25 20.9 ±13.5 20.4 ±13.9 21.2 ±13.3 Treuth et al 26 23.4 ±14.6 22.6 ±14.9 24.0 ±14.7 5 8 year-olds Sample size 72 26 46 Pedometer-determined MVPA (120 steps min -1 ) 27.3 ±18.7 19.4 ±11.8 31.7 ±20.5 Accelerometer-determined MVPA Evenson et al 22 37.4 c ±22.5 24.8 ±16.9 44.5 ±22.3 Freedson et al 23 54.9 c ±26.3 38.9 ±20.7 63.9 ±24.9 Puyau et al 25 23.7 c ±16.5 14.9 ±12.2 28.2 ±16.8 Abbreviations: SD, standard deviation. a Indicates significantly different MVPA estimate compared with pedometer derived MVPA for entire sample. b Indicates significantly different MVPA estimate compared with pedometer derived MVPA for youth ages 10 14 years. c Indicates significantly different MVPA estimate compared with pedometer derived MVPA for youth ages 5 8 years.
Pedometer MVPA S299 For the younger youth (5 8yr olds), the paired t test indicated a significant difference between P-MVPA and EV (t = 5.35, P <.001), FR (t = 12.45, P = <.001), and PU (t = 2.40, P =.019). The Bland Altman plot indicated the pedometer underestimated minutes spent in MVPA at higher levels of MVPA (plots available upon request). The mean difference was EV 10.1 min (95% CI 13.8 to 6.3) and 42.0 to 21.9 min, FR 27.6 min (95% CI 31.9 to 23.16) and 65.2 to 10.0), and PU 3.9 min (95% CI 0.7 to 7.2) and 23.7 to 31.6. Pitman s Tests were EV (r =.258, P =.04), FR (r =.436, P =.001), and PU (r =.173, P =.145). Regression analyses, using the absolute value of the difference between MVPA estimates indicated that, in accordance with the Bland-Altman plots (Figure 1), more active youth exhibited a larger difference between P-MVPA and accelerometer-derived MVPA in comparison with the least active youth (see Table 3). Moderately active youth (middle tertile) had an associated difference ranging from 4.19 mins (PU) to 8.59 mins (FR), while the most active youth (highest tertile) had an associated difference between MVPA estimates ranging from 8.33 mins (MT) to 30.14 mins (FR) in comparison with the least active youth. Classification as obese ( 95th age-sex specific percentile) was associated with a larger difference between estimates for MT (7.44 mins), PU (7.70 mins), and TR (7.28 mins) in comparison with youth < 95th percentile. In addition, as age (in years) increased, the difference between estimates decreased for FR ( 2.75 mins) and MT ( 0.98 mins). Discussion This is the first study to examine the convergent validity of pedometer and accelerometer derived MVPA in youth. The results of this study suggest that MVPA via pedometery, based on a 120 steps min -1 threshold, provides a reasonable estimate when compared with MVPA from accelerometers using established cutpoints. Importantly, the age-specific analyses indicated that for the older youth (10 14 years), pedometer MVPA was more closely related to MVPA estimated from accelerometers using the MT, PU, and TR cutpoints. It is important to note these accelerometer cutpoints have been used widely in youth physical activity research. Most prominently the Treuth et al 26 cutpoints were used in the large-scale intervention Trial of Activity for Adolescent Girls (TAAG). 31,32 Because costs associated with measuring MVPA on a large-scale can be cost prohibitive, and that costs for MVPA pedometers are approximately 90% less per unit in relation to accelerometers (eg, ActiGraph $335/ unit), pedometer-determined MVPA may serve as a costeffective alternative for measuring MVPA for old youth (10 14 years). Despite these findings, there are several considerations before wholesale adoption. Estimates of MVPA from the pedometer deviated significantly from the accelerometer estimates for the younger children (5 8 years) and the more active and obese youth (see Table 2 and 3 and Figure 1). The difference for the younger youth may be due to the age of youth the 120 steps min -1 step frequency guideline was established 10 to 12 years olds. 17 The difference may be largely associated with the differences in leg length between younger and older children. Leg length increases with height, height increases with age, and leg length plays a prominent role in step frequency (longer legs require fewer steps for a given amount of time). Thus, 120 steps min -1 for one age range may not directly translate to the same threshold for older/younger age range. The findings presented herein suggest this may be the case. Hence, additional work is required to determine age-specific step frequency guidelines for younger youth. For the overall sample, examination of the Bland- Altman plots suggest that at the higher levels of MVPA (x-axis), the difference between pedometer and accelerometer MVPA was not systematic, indicating the difference between estimates fell both positively (indicating the pedometer estimate was higher) and negatively (indicating the accelerometer estimate was higher). Hence, 1 device was not systematically over/underestimating MVPA in relation to the other, except in the Freedson et al 23 comparison. Thus, more active youth and youth classified as obese have P-MVPA estimates that may not accurately reflect their activity level. Future research is required to examine the mechanisms for these differences. Although low-cost and ease of use are benefits to using pedometers, this comes at the expense of other features researchers use with accelerometer-derived MVPA. For instance, pedometers that incorporate the step frequency technology, at this time, do not have the ability to measure consecutive days. Thus, as with prior pedometer research, a diary of daily pedometer activity or accumulation and division across the number of days worn is required. Moreover, one cannot determine nonwear time with pedometers, except for use with a daily recall. Research examining the segments of the day where youth are active is more complicated with pedometers, in relation to accelerometers, 33 and requires on-sight recording of values or the exchange of pedometers during transitional parts of the day (eg, school to home or with the school day). 34 However, the number of studies that segment the day are relatively few, with the majority of the accelerometer research focused on average daily MVPA. The step frequency technology does not allow for the separation of moderate and vigorous physical activities and is unable to provide profiles of activity behavior. Finally, with P-MVPA one cannot use more complex modeling approaches to potentially identify type, intensity, and duration, such as Neural Network Analysis 35,36 and Hidden Markov Modeling. 37 An additional limitation is the use of accelerometer cutpoints that were developed for youth with a wide range of ages, the wide range of ages represented in the current study
(continued) Figure 1 Bland-Altman plots representing the differences between pedometer-determined MVPA and accelerometer-determined MVPA estimated from 5 cutpoints. Solid line represents the regression on the variance between the average MVPA (x-axis) and the difference (y-axis). Dashed lines represent the 95% limits of agreement. S300
(continued) Figure 1 (continued) S301
S302 Beets et al Figure 1 (continued) (140 youth were between 6 and 11 years old), and the discrepancies among accelerometer MVPA estimates. The accelerometer cutpoints were developed on samples that ranged from 5 to 8 years (EV), 6 to 18 years (FR), 6 to 16 years (PU), 13 to 14 years (TR), and 12 years (MT). Of interest is the comparability between the MT, PU, and TR estimates, for the overall sample and especially for the 10 to 14yr olds, in light of the different age groups included in the original calibration studies. Because of the wide range of values observed from comparison studies of the predictiveness of accelerometer cutpoints (population specificity), it remains unclear as to which cutpoints are the most appropriate for a given sample. 20,38 Another issue in interpreting MVPA estimates from these accelerometer cutpoints are findings that suggest they either over- or underestimate energy expenditure. 39,40 Thus, these issues need to be taken into account when interpreting the comparisons with P-MVPA. Before widespread use, there are several additional limitations that need to be addressed. Primarily, pedometer MVPA was estimated using 120 steps min -1 threshold. This guideline was based on observed step frequencies at 3.0, 3.5, and 4.0 mph in a sample of 10 to 12 year-olds. 17 No other physiologic measures, such as VO 2 or heart rate, were used to establish this threshold. Although this guideline is a significant step in establishing MVPA step frequency guidelines for youth, it may not generalize to older or younger youth, to those who are overweight or obese, or to those who are significantly shorter or taller than the children in the sample. Thus, additional work is needed to develop step frequency thresholds that account for these potential differences. This may lead to more consistent pedometer-based estimates of MVPA compared with accelerometry. In addition, the pedometer used in this study employed a spring-lever recording mechanism that may have unduly influenced the accuracy of the device 41,42 when tilted or at slow walking speeds. Pedometers are also less accurate at slower speeds, 10 therefore, when youth in the current study were moving at speeds where the pedometer may not register, a reduction in overall steps would occur. Future innovations in pedometers should incorporate an activity recording mechanism (eg, piezo-electric) similar to that currently used in accelerometers, which is more accurate that spring-lever pedometers while tilted and at slow walking speeds. 42 Despite these limitations, the current study provides preliminary evidence that a low-cost, easy-to-use pedometer can provide reasonable MVPA estimates compared with more expensive devices. For researchers, the ability to measure MVPA with a less expensive pedometer will address a major barrier to free-living physical activity assessment cost (eg, accelerometers ~$335 vs. MVPA pedometer ~$35). A less expensive, but similarly reliable
Table 3 Comparison of Participant Characteristics to the Absolute Difference Between Pedometer-Determined MVPA and Each of the 5 Accelerometer-Determined Estimates of MVPA (N = 149) Age (yrs) Obese a Stride b Physical activity tertile c Moderate (tertile 2) High (tertile 3) Accelerometer-determined MVPA Coef. (95%CI) Coef. (95%CI) Coef. (95%CI) Coef. (95%CI) Coef. (95%CI) Evenson et al 22 0.25 ( 0.87, 1.38) 2.28 ( 4.29, 8.86) 0.35 ( 1.23, 0.52) 5.10 (2.38, 7.81) 21.02 (16.01, 26.02) Freedson et al 23 2.75 ( 4.08, 1.43) 0.72 ( 8.03, 9.46) 0.40 ( 1.47, 0.68) 8.59 (5.31, 11.87) 30.14 (24.04, 36.23) Mattocks et al 24 0.98 ( 1.93, 0.04) 7.44 (2.84, 12.04) 0.30 ( 0.91, 0.31) 4.45 (1.48, 7.42) 8.33 (5.15, 11.50) Puyau et al 25 0.87 ( 1.74, 0.00) 7.70 (3.30, 12.10) 0.36 ( 0.96, 0.23) 4.19 (1.57, 6.82) 9.10 (5.73, 12.47) Treuth et a 26 0.73 ( 1.62, 0.17) 7.28 (2.79, 11.76) 0.28 ( 0.90, 0.34) 4.30 (1.77, 6.83) 10.63 (7.19, 14.07) Abbreviations: Coef., coefficient. Note. 5 models were estimated, 1 for each accelerometer cutpoint. Predictors were all entered simultaneously into the model. a Obese defined as at or above the 95th percentile for age-sex. b Stride estimated by number of steps taken over 10 m. c Tertile based on total daily accelerometer counts. Reference group lowest tertile (0 33% of distribution). S303
S304 Beets et al and valid device would mitigate concerns of unit loss or damage which burden current efforts at assessing free-living activity and MVPA. For the general public, the ability to conveniently and inexpensively document MVPA levels and track progress toward MVPA recommendations may help promote greater time spent at physical activity levels associated with health benefits. 43 Translating the output of this technology into practical information that children, adolescents, parents, practitioners, and researchers can use, however, will require additional development of step frequency guidelines. References 1. Cameron C, Wolfe R, Craig CL. Physical activity and sport: encouraging children to be active. Ottawa, ON: Canadian Fitness and Lifestyle Research Institute; 2007:213. 2. Bravata DM, Smith-Spangler C, Sundaram V, et al. Using pedometers to increase physical activity and improve health: a systematic review. JAMA. 2007;298(19):2296 2304. 3. Lubans DR, Morgan PJ, Tudor-Locke C. A systematic review of studies using pedometers to promote physical activity among youth. Prev Med. 2009;48(4):307 315. 4. Tudor-Locke C, Williams JE, Reis JP, Pluto D. Utility of pedometers for assessing physical activity: convergent validity. Sports Med. 2002;32(12):795 808. 5. Tudor-Locke C, Williams JE, Reis JP, Pluto D. Utility of pedometers for assessing physical activity: construct validity. Sports Med. 2004;34(5):281 291. 6. Crouter SE, Schneider PL, Karabulut M, Bassett DR. Validity of 10 electronic pedometers for measuring steps, distance, and energy cost. Med Sci Sports Exerc. 2003;35(8):1455 1460. 7. Schneider PL, Crouter SE, Bassett DR. Pedometer measures of free-living physical activity: comparison of 13 models. Med Sci Sports Exerc. 2004;36(2):331 335. 8. Beets MW, Foley JT, Lieberman LJ, Tindall DWS. Accuracy of voice-announcement pedometers for youth with visual impairments. Adap Phys Act Qu. 2007;24(3):218 227. 9. Beets MW, Combs C, Pitetti KH, Morgan M, Bryan RR, Foley JT. Accuracy of pedometer steps and time for youth with disabilities. Adap Phys Act Qu. 2007;24(3):228 244. 10. Beets MW, Patton MM, Edwards S. The accuracy of pedometer steps and time during walking in children. Med Sci Sports Exerc. 2005;37(3):513 520. 11. Le Masurier GC, Corbin CB. Steps counts among middle school students vary with aerobic fitness level. Res Q Exerc Sport. 2006;77(1):14 22. 12. Cardon G, De Bourdeaudhuij I. Comparison of pedometer and accelerometer measures of physical activity in preschool children. Pediatr Exerc Sci. 2007;19(2):205 214. 13. Jago R, Watson K, Baranowski T, et al. Pedometer reliability, validity and daily activity targets among 10- to 15-year-old boys. J Sports Sci. 2006;24(3):241 251. 14. Rowlands AV, Eston RG. Comparison of accelerometer and pedometer measures of physical activity in boys and girls, ages 8-10 years. Res Q Exerc Sport. 2005;76(3):251 257. 15. Eston RG, Rowland AV, Ingledew DK. Validity of heart rate, pedometry, and accelerometry for predicting the energy cost of children s activities. J Appl Physiol. 1998;84(1):362 371. 16. Oliver M, Schofield GM, Kolt GS, Schluter PJ. Pedometer accuracy in physical activity assessment of preschool children. Journal of Science and Medicine in Sport / Sports Medicine Australia. 2007;10(5):303 310. 17. Graser SV, Pangrazi RP, Vincent WJ. Steps it up: activity intensity using pedometers. J Phys Educ, Recreat Dance. 2009;80(1):22 24. 18. Vincent SD, Sidman CL. Determining measurement error in digital pedometers. Meas Phys Educ Exerc Sci. 2003;7(1):19 24. 19. Bailey RC, Olson J, Pepper SL, Porszaz J, Barstow TJ, Cooper DM. The level and tempo of children s physical activities: an observational study. Med Sci Sports Exerc. 1995;27:1033 1041. 20. Nilsson A, Brage S, Riddoch C, et al. Comparison of equations for predicting energy expenditure from accelerometer counts in children. Scand J Med Sci Sports. 2008;18(5):643 650. 21. Pate RR, Almeida MJ, McIver KL, Pfeiffer KA, Dowda M. Validation and calibration of an accelerometer in preschool children. Obesity (Silver Spring, MD). 2006;14(11):2000 2006. 22. Evenson KR, Catellier DJ, Gill K, Ondrak KS, McMurray RG. Calibration of two objective measures of physical activity for children. J Sports Sci. 2008;26(14):1557-65.. 23. Freedson P, Pober D, Janz KF. Calibration of accelerometer output for children. Med Sci Sports Exerc. 2005;37(11, Suppl):S523 S530. 24. Mattocks C, Leary S, Ness A, et al. Calibration of an accelerometer during free-living activities in children. Int J Pediatr Obes. 2007;2(4):218 226. 25. Puyau MR, Adolph AL, Vohra FA, Butte NF. Validation and calibration of physical activity monitors in children. Obes Res. 2002;10(3):150 157. 26. Treuth MS, Schmitz K, Catellier DJ, et al. Defining accelerometer thresholds for activity intensities in adolescent girls. Med Sci Sports Exerc. 2004;36(7):1259 1266. 27. Trost SG, Rosenkranz RR, Dzewaltowski D. Physical activity levels among children attending after-school programs. Med Sci Sports Exerc. 2008;40(4):622 629. 28. Baquet G, Stratton G, Van Praagh E, Berthoin S. Improving physical activity assessment in prepubertal children with high-frequency accelerometry monitoring: a methodological issue. Prev Med. 2007;44(2):143 147. 29. Nilsson A, Ekelund U, Yngve A, Sjostrom M. Assessing physical activity among children with accelerometers using different time sampling intervals and placements. Pediatr Exerc Sci. 2002;14:87 96. 30. Bland JM, Altman DG. Measuring agreement in method comparison studies. Stat Methods Med Res. 1999;8(2):135 160. 31. Webber LS, Catellier DJ, Lytle LA, et al. Promoting physical activity in middle school girls: Trial of Activity for Adolescent Girls. Am J Prev Med. 2008;34(3):173 184. 32. Stevens J, Murray DM, Baggett CD, et al. Objectively assessed associations between physical activity and body composition in middle-school girls: the Trial of Activity for Adolescent Girls. Am J Epidemiol. 2007;166(11):1298 1305.
Pedometer MVPA S305 33. Guinhouya CB, Soubrier S, Vilhelm C, et al. Physical activity and sedentary lifestyle in children as time-limited functions: usefulness of the principal component analysis method. Behav Res Methods. 2007;39(3):682 688. 34. Tudor-Locke C, Lee SM, Morgan CF, Beighle A, Pangrazi RP. Children s pedometer-determined physical activity during the segmented school day. Med Sci Sports Exerc. 2006;38(10):1732 1738. 35. Karantonis DM, Narayanan MR, Mathie M, Lovell NH, Celler BG. Implementation of a real-time human movement classifier using a triaxial accelerometer for ambulatory monitoring. IEEE Trans Inf Technol Biomed. 2006;10(1):156 167. 36. Rothney MP, Neumann M, Beziat A, Chen KY. An artificial neural network model of energy expenditure using nonintegrated acceleration signals. J Appl Physiol. 2007;103(4):1419 1427. 37. Pober DM, Staudenmayer J, Raphael C, Freedson PS. Development of novel techniques to classify physical activity mode using accelerometers. Med Sci Sports Exerc. 2006;38(9):1626 1634. 38. Guinhouya CB, Hubert H, Soubrier S, Vilhelm C, Lemdani M, Durocher A. Moderate-to-vigorous physical activity among children: discrepancies in accelerometry-based cutoff points. Obesity (Silver Spring, MD). 2006;14(5):774 777. 39. Trost SG, Way R, Okely AD. Predictive validity of three ActiGraph energy expenditure equations for children. Med Sci Sports Exerc. 2006;38(2):380 387. 40. Wickel EE, Eisenmann JC, Welk GJ. Predictive validity of an age-specific MET equation among youth of varying body size. Eur J Appl Physiol. 2007;101(5):555 563. 41. Pitetti KH, Beets MW, Flaming J. Accuracy of pedometer steps and time for youth with developmental disabilities during dynamic movements. Adap Phys Act Qu. 2009:26:1 16. 42. Crouter SE, Schneider PL, Bassett DR. Springlevered versus piezo-electric pedometer accuracy in overweight and obese adults. Med Sci Sports Exerc. 2005;37(10):1673 1679. 43. U.S. Department of Health and Human Services. 2008 physical activity guidelines for Americans. Washington, D.C.: U.S. Department of Health and Human Services; 2008:61.