The increased prevalence of childhood obesity (9,23) is driving an expanding interest in objective measurement of physical activity (PA) using pedometers and accelerometers in children. Pedometers are simple, low-cost devices that are well accepted as a valid (5,19,25) and reliable (2,4,20) measure of total daily PA (i.e., steps) in both children and adults. Pedometers do not provide estimates of time in different intensities of PA (sedentary, light, moderate, and vigorous), an output that speaks directly to public health recommendations (31). In contrast, accelerometer-based activity measurement technology is widely accepted as a valid means of assessing frequency, intensity, and duration of ambulatory PA in children (10,17,24,28) and adults (7,15,22,32). Specifically, accelerometers are being increasingly used to estimate time in moderate to vigorous PA (MVPA), a parameter considered indicative of health-promoting physical activity (31). However, widespread use of accelerometers for applications such as screening, surveillance, program evaluation, and self-monitoring is limited by their cost and operational logistics, including the time necessary for data collection and management, as well as technical expertise.
A newer accelerometer (released for sale in the United States in 2005), the Kenz Lifecorder EX (LC; Suzuken Co. Ltd, Nagoya, Japan), offers promise as an alternative monitor appropriate for assessing children's PA. The LC is slightly less expensive ($245 vs $325 as of 2/27/07) than an accelerometer with a similar function such as the AG ActiGraph (AG Health Services, Fort Walton Beach, FL). The LC also has a user-friendly software interface (Physical Activity Analysis Software, Version 1.0) and is simple to initialize and download. Specifically, the data outputs (11 progressively increasing activity-intensity categories, steps, and energy-expenditure estimates) are easily exported from the software into an Excel spreadsheet producing one line of data per participant for each day of monitoring (see Table 1 for an example of the printout). This data-output format allows the user to apply simple formulas to quickly calculate time in a given intensity category. With many other accelerometers, data interpretation (i.e., converting raw data outputs into derived variables such as time in PA intensity) is a time-intensive process without the implementation of uniquely defined, user-designed macro syntax. The LC can potentially simplify the data-interpretation process by reducing the time spent and the need for advanced technical expertise or software programs.
One limitation of many accelerometers is the inability to differentiate between light-intensity ambulatory movement versus other forms of nonambulatory postural changes or movement (such as jostling during vehicle travel). As a result, many researchers discard accelerometer data representing light-intensity PA and focus solely on MVPA. This approach contrasts pedometer-based methods, which have focused on an "every step counts" approach to PA assessment. The LC may improve our ability to detect light-intensity ambulatory movement (while discarding nonambulatory noise) through the use of a proprietary filtering process (which considers both frequency and magnitude of accelerations) for the determination of PA intensity.
Specifically, the maximum pulse (i.e., the g value of an acceleration) detected by the LC during the manufacturer-determined 4-s sampling interval is taken as the acceleration value, and each interval is categorized into one of 11 activity-intensity levels. A description of LC and AG raw outputs is provided in Table 2. The LC categorizes 4-s intervals with < 0.06g of detected accelerations as sedentary (level 0) (11). The AG has a similar minimal threshold for detection of movement (< 0.05g = 0 activity counts) (7). However the LC's micro activity category (level 0.5) provides a unique method of filtering ambulatory and nonambulatory accelerations that is not currently replicated by the AG. Classification of movement data into the LC micro activity category indicates that an acceleration signal was detected above the minimum threshold but that fewer than three pulses (vertical accelerations) were detected during the 4-s sampling interval (detected accelerations were not indicative of continuous ambulatory movement). In effect, the LC classifies movement data accumulated at a step frequency of fewer than 45 steps per minute (regardless of intensity) into the micro activity category considered to be nonambulatory movement. The LC's ability to detect all intensities (light, moderate, and vigorous) of ambulatory movement while filtering nonambulatory acceleration signals may allow for improved ability to examine the role of light-intensity PA with regard to outcomes such as weight maintenance or health status.
PA intensity categories for the LC have been validated in adults (1-3 = light, 4-6 = moderate, 7-9 = vigorous) (11). Step outputs from a previous model of the LC (Kenz Lifecorder) have been compared in adults against observed steps with an accuracy of ± 1-3% (5,20), and no differences were observed in comparison with concurrent free-living steps detected by the Yamax pedometer (DW; Yamax Digiwalker SW-200, Yamasa, Tokyo, Japan) (19). However, no studies have evaluated the LC on any parameter in children.
Taken together, the functional properties of the LC (user-friendly interface, reduced time and technical demands, and filtering of nonambulatory movement at all intensities) may make it a useful alternative to the more commonly used AG accelerometer for practitioners or researchers assessing PA in children. Currently, the AG is the most commonly used accelerometer for measurement of PA intensity (25), and the DW is widely accepted as a valid measure of steps taken in a free-living environment (19). The utility of the LC would be further strengthened if it provided similar estimates of MVPA and steps counts compared with the respective outputs from the AG accelerometer and DW pedometer. Therefore, the purpose of this study was to evaluate comparable outputs (specifically, steps and time in light and MVPA) from these three objective monitors in free-living children.
A sample of 31 fifth-grade boys (N = 13, height = 1.4 ± 0.1 m, weight = 38.6 ± 8.8 kg) and girls (N = 18, height = 1.4 ± 0.1 m, weight = 41.9 ± 7.4 kg) without physical impairments that affected ambulation were recruited from a single school located in an urban school district in the southwestern United States. All aspects of this study were approved by the Arizona State University institutional review board and the community school district. Parents provided informed consent, and children provided written assent for study participation.
The LC is 7.0 × 4.0 × 2.5 cm in size and attaches to the waist line or belt by an attached belt clip at the midline of either thigh. Twenty LC were used for data collection. The LC uses a single-axis accelerometer that measures vertical accelerations at the hip. The accelerometer samples at 32 Hz and assesses values ranging from 0.06 to 1.94g. The signal is filtered by an analog bandpass filter and is digitized. Additionally, the LC assesses the number of cycles present in the acceleration signal and outputs this value as a measure of total activity (i.e., steps). LC were initialized (which requires setting start time and date and optionally includes inputting height, weight, age, and gender of the participant, and selecting display and keypad function for subsequent data collection) and downloaded according to manufacturer's instructions using Physical Activity Analysis software (Version 1.0, Suzuken Co. Ltd, Nagoya, Japan).
The AG is 5.3 × 5.0 × 2.0 cm in size and is worn on an elastic belt or in a clip pouch at the midaxillary line of either hip. Twenty AG were used for data collection. Detailed technical specifications for the AG are provided elsewhere (7). Accelerations are converted into activity count values, which increase linearly with magnitude of accelerations. Activity counts are summed (and recorded) for a user-specified time-sampling interval (epoch) ranging from 5 s to 1 min. Additionally, the number of cycles present in the signal within an epoch (which can be interpreted as the number of steps taken) can be stored. Resulting raw data outputs are activity counts and steps. The sum of activity counts in an epoch is linearly related to activity intensity (i.e., with increasing quantity or speed of ambulatory movement, we can infer that intensity is increasing) and can be classified into intensities of PA based on validated established activity-count cut points. AG were initialized (which requires setting start time and date, epoch, and data file name) for 30-s epochs and downloaded according to manufacturer's instructions, using Actisoft Analysis software (Version 3.2.5, AG Health Services). Although an epoch length as low as 5 s could have been selected for the MTI, a 30-s epoch length was used in this study to provide a more accurate comparison of LC and AG intensity outputs under typical monitoring conditions for children (17,24,26). Before initialization and after completion of data collection, AG calibration was assessed with an AG calibrator (Model CAL71, AG Health Services). A calibration gain factor of 0.60 represents the factory calibration, and a gain factor range of 0.57-0.63 (± 5%) is considered acceptable. All monitors were in calibration before initialization (meangain factor = 0.60 ± 0.01), and no monitors were out of calibration at posttesting calibration (mean gain factor = 0.60 ± 0.01).
The DW is 5.2 × 3.9 × 1.9 cm in size and attaches to the waist line or belt by an attached belt clip at the midline of either thigh. Twenty DW were used for data collection. The DW is an electronic pedometer with a horizontal, spring-suspended lever arm that moves up and down with vertical accelerations of the hip (3). When accelerations are ≥ 0.35g, the lever arm makes an electrical contact, and one event (step) is recorded (30). The DW model used here does not have a memory function; therefore, pedometers were unsealed to allow participants to self-record steps on a provided log.
On the morning of the scheduled monitoring period, participants received visual, verbal, and written instructions and feedback on monitor placement and fitting, appropriate use of the LC, AG, and DW monitors, and an explanation of the monitoring protocol. Participants were fitted with all three monitors. The LC was the only monitor positioned in a location other than according to the manufacturer-recommended placement. However, the LC displayed small, nonsignificant differences between right- and left-hip monitor outputs and high ICC values of 0.99 for steps and 0.98 for MVPA in previous evaluations in adults within our lab (McClain et al., unpublished results, 2006). Therefore, placement of the LC was not expected to impact resulting monitor outputs.
Participants were instructed to wear the monitors from the time of the initial attachment at school (which took place in the morning at about 7:45 a.m.) and during all waking hours (other than water activities) for the remainder of the day. Participants were instructed to engage in their normal PA habits during the monitoring period. At bed time, children removed the monitors and recorded DW steps on a provided log. On the subsequent day, researchers collected monitors from participants on their arrival at school. Participants completed a brief exit interview to assess compliance with the monitoring protocol. This particular group of children was accustomed to monitoring their PA using pedometers on other occasions during physical education (PE) classes, so reactivity was not anticipated, nor would it have affected the specific research questions under study. A single day of data collection was sufficient for this study because the purpose was to compare monitor outputs during concurrent wear (12,19), rather than to determine participants' usual PA patterns, which would have required additional days of monitoring (27).
Daily activity summary files (summary of steps and the number of 4-s sampling intervals measured in each of 11 activity categories) from the LC were downloaded into a master dataset. In the absence of validated LC intensity cut points for children, we first used a validated adult intensity classification (LC_4; activity level 1-3 = light, 4-9 = MVPA) for comparison with the AG (11). Validation research for the AG in children has shown that a higher moderate-intensity threshold is needed for children (17,24) than for adults (7). Therefore, we made an a priori decision to also evaluate an alternative intensity category within this study (LC_5; activity level 1-4 = light, 5-9 = MVPA). Measured time in light-intensity activity and MVPA were calculated using LC_4 and LC_5 derivations.
AG data management requires creation of separate pedometer (steps) and activity-count files and variables. Individual files were created for processing. A number of validated activity-count categories are available for interpreting children's AG data. Cut points established by Treuth et al. (24) (AG_Treuth; 0-50 counts per 30 s = sedentary, 51-1499 counts per 30 s = light, ≥ 1500 counts per 30 s = MVPA) and Puyau et al. (17) (AG_Puyau; 0-399 counts per 30 s = sedentary, 400-1599 counts per 30 s = light, ≥ 1600 counts per 30 s = MVPA) were recommended in a 2005 review of children's accelerometer-calibration literature (6) for use within the age range of our participants. In addition, the Freedson et al. (8) children's age-specific MET predication equation (METs = 2.757 + (0.0015 × counts per minute) - (0.08957 × age) − (0.000038 × counts per minute × age)) was used to determined MVPA cut points for 3 METs (8). The Freedson children's equation does not provide validated methods to differentiate time in sedentary and light-intensity activity. Therefore, only MVPA estimates are provided from the Freedson intensity derivation (AG_Freedson) with resulting MVPA cut points in 10- and 11-yr-olds, respectively, of 509 counts per 30 s and 568 counts per 30 s.
Additionally, the Trost et al. (29) predication equation for energy expenditure (EE) in kilocalories per minute (kilocalories per minute = −2.23 + (0.0008 × counts per minute) + (0.08 × body mass in kilograms)) was used to derived estimates (AG_Trost) of activity energy expenditure (AEE) (29). To use the above equation with the 30-s epoch length used in this study, activity count outputs were multiplied by 2 and entered into the equation. Resulting outputs in kilocalories per minute were then divided by 2 to provide a measure of EE in kilocalories per 30 s. Resting metabolic rate (RMR) was predicted from each participant's gender, age, height, and weight using Schofield's equation for children aged 10-18 yr and converted to RMR per 30 s (21). Predicted EE from the Trost equation in kilocalories per 30-s values were converted into activity energy expenditure (AEE; kcal·kg−1·min−1) by dividing kilocalories per 30 s by body mass (kg) and subtracting predicted RMR per 30 s. Resulting AEE values were classified into intensity categories using a 30-s conversion (dividing by 2) of an AEE intensity-classification scheme described by Puyau et al. (18). Thirty-second AEE intensity classifications were defined as sedentary (≤0.0075 kcal·kg−1·min−1), light (≥ 0.0075 kcal·kg−1·min−1, and < 0.02 kcal·kg−1·min−1), and MVPA (≥ 0.02 kcal·kg−1·min−1). Measured time in light-intensity activity and MVPA were calculated using AG_Treuth, AG_Puyau, AG_Freedson, and AG_Trost intensity derivations.
Delta values (absolute difference determined by subtraction) by individual between DW versus LC and DW versus AG steps comparisons revealed one participant with an unusually extreme set of values (Δ = 8933 and 8951 steps, respectively) that was more than three standard deviations from the mean delta values. This outlier was interpreted as a probable case in which the DW pedometer was either malfunctioning, inadvertently reset during data collection, or taken off while the participant continued to wear the LC and AG, resulting in comparatively much lower pedometer versus accelerometer steps. This one case was removed before further analyses. Additionally, two individuals failed to return pedometer logs and one AG monitor failed to initialize. The final analyses sample included 27 participants with complete steps data (from all three monitors) and 30 participants with complete intensity data (from the two accelerometers).
Means and standard deviations were calculated for the DW, LC, and AG raw step data and derived intensity variables (Table 3). Intraclass correlation coefficients (ICC) were calculated for steps from all monitors and minutes in light-intensity activity and MVPA (for each of the different derivations of these data) for the LC and AG. Repeated-measures ANOVA was used to compare variable means between monitor outputs. SPSS version 13.0 was used to complete all analyses, and an alpha level of 0.05 was interpreted for significance.
Monitor-wearing time for participants was calculated on the basis of AG outputs and averaged 12.8 ± 1.7 h (range = 8.9-14.5 h). ICC values for LC, AG, and DW steps ranged from 0.87 to 0.96. ICC values for two LC and four AG intensity derivations for minutes in light-intensity activity and MVPA, respectively, ranged from 0.38 to 0.99 and from 0.67 to 0.99.
Data for monitor step outputs are provided in Table 4. Repeated-measures ANOVA displayed a significant overall difference in steps between monitors (F = 5.1, P = 0.021). Post hoc pairwise analyses indicated that significant differences were limited to the LC and AG steps comparison (Δ = 801 steps, P = 001), with higher steps recorded for the AG. A nonsignificant trend was observed in the pairwise comparison between DW and AG steps (P = 0.058), with higher steps recorded for the AG. Compared with the DW, mean differences and effect size (ES) of the difference in steps for the LC and AG, respectively, were 200 ± 364 (ES = 0.06) and 1001 ± 401 (ES = 0.29) steps.
Data for LC and AG intensity derivations for light and MVPA minutes are displayed in Table 5. There were significant overall differences for light-intensity time between the two LC and three AG intensity derivations (F = 26.5, P < 0.001). Post hoc pairwise analyses indicated that AG_Trost was not significantly different from the remaining light-intensity derivation. All pairings of LC_4, LC_5, AG_Treuth, and AG_Puyau intensity derivations of light-intensity activity minutes were significantly different. The lowest estimate of light-intensity activity minutes was produced by LC_4 (75.6 ± 18.4 min), whereas the highest estimate of light-intensity activity minutes was produced by AG_Treuth (309 ± 69.2 min).
There were significant overall differences in MVPA between the two LC and four AG intensity derivations (F = 21.5, P < 0.001). Post hoc pairwise analyses indicated significant differences in MVPA for LC_4, AG_Trost, and AG_Freedson with all possible pairings of MVPA derivations. MVPA derivations from AG_Treuth and AG_Puyau were significantly different from each other. No differences were statistically significant between LC_5 and either AG_Treuth or AG_Puyau. The LC_5 detected 4.9 (or 18.9%) and 1.7 (or 5.4%) fewer minutes in MVPA compared with AG_Treuth and AG_Puyau, respectively.
Average wearing time (12.8 ± 1.7 h) was above the minimum threshold frequently used to constitute a full day of monitoring (10 h) (14). One participant's wearing time fell below 10 h and may not be representative of a full day of monitoring. However, wearing time would not affect our specific research question or current analyses; therefore, no participants were excluded on the basis of wearing time. PA monitoring was purposefully scheduled for days on which participants attended a 30-min PE class to maximize participation in MVPA. Surprisingly, average participation in MVPA (including PE) ranged from only 25.9 to 37.1 min among the three derivation methods, with nonsignificant differences (LC_5, AG_Treuth, AG_Puyau) for our sample of fifth-grade males and females. Examination of individual values for MVPA revealed that for AG_Treuth, three participants accumulated ≥ 60 min of MVPA during the monitoring day. According to LC_4, LC_5, and AG_Puyau intensity derivations, no students accumulated ≥ 60 min of MVPA. Admittedly, we did not collect PA data early in the day, before monitors were worn. Regardless, according to LC_5, AG_Treuth, and AG_Puyau MVPA intensity derivations for this single day of data collection, the vast majority of our sample would be considered insufficiently active on the basis of the revised NASPE recommendation that calls for children between the ages of 5 and 12 yr to accumulate 60 or more minutes of developmentally appropriate PA on all or most days of the week (16). Once again, however, this unusual finding would not have direct bearing on our primary research question. In contrast, interpretation of individual data on the basis of AG_Trost and AG_Freedson intensity derivations respectively resulted in 66.6 and 96.6% of participants achieving at least 60 min of MVPA for this single day.
A principal finding within this study was the comparison of step outputs from the DW, LC, and AG monitors. To date, no study has compared steps from the LC and AG with that of the DW pedometer in children. Research in adults has indicated that the AG is more susceptible to counting nonambulatory movement (such as jostling during vehicle travel) as steps (13) and records significantly higher steps (up to 19%) compared with the DW in adults during free living (30). There was a nonsignificant trend towards higher steps in the AG versus DW, with the AG detecting 1001 greater steps on average compared with the DW. It is possible that the nonsignificant difference between AG and DW steps shown here, compared with the significant differences noted above in adults, is a result of factors such as potentially less motor vehicle commuting time in children. The monitors used here require differential minimum vertical accelerations (measured as a g value) at the hip to register movement as a step. The DW is reported to have the most conservative threshold, requiring a force of 0.35g to register a step (30). Both the AG and the LC monitors register steps at a lower g threshold (AG = 0.30g, LC = proprietary value purportedly lower than both the AG and DW), suggesting an increased sensitivity for step detection compared with the DW (12). A small mean difference was observed between DW and LC steps (200 steps). A probable explanation for the smaller relative difference between DW and LC compared with DW and AG is the proprietary data-filtering process (which considers both frequency and intensity of vertical accelerations in the determination of steps) for the LC steps output. Thus, the LC may be less susceptible to registering random nonambulatory vibrations or postural changes as steps compared with the AG.
The large variability between LC and AG derivations of light-intensity activity minutes is not an unexpected result. Partitioning of sedentary and light-intensity behaviors according to accelerometer outputs is in the developmental stages. Early monitor-calibration studies in children (29) and adults (7) made no attempt to differentiate between sedentary and light-intensity behaviors. Recently, studies in children (17,24) and adults (30) have begun to used a light-intensity count cut point. Light-intensity activity minutes derived from AG_Treuth (309.0 min) and AG_Trost (184.5 min) were proportionally higher relative to all other derivations (range = 75.6-116.0 min). This may suggest that for the AG_Treuth derivation, the lower limit of the light-intensity category (50 counts per 30 s) should be interpreted as a threshold for any purposeful movement rather than continuous light-intensity ambulatory movement. The wide range of individual outputs for the AG_Trost light-intensity derivation for minutes (33.5-662.5 min) suggests that the conversion of EE data (kilocalories per 30 s) into categorical intensity data is susceptible to error in both the EE and RMR prediction equations. Future studies should consider using a behavioral criterion standard (e.g., direct observation) to evaluate monitor outputs of light-intensity activity minutes.
Recent AG validation research indicates that a higher-intensity threshold is needed as a floor value for moderate-intensity activity in children (17,24) compared with adults (7). AG_Treuth and AG_Puyau count-per-minute thresholds (adjusted up from 30-s epoch values used here for comparison with a 1-min adult epoch) for moderate-intensity PA in children are 3000 and 3200 counts per minute, respectively. In comparison, a widely referenced AG threshold for moderate-intensity PA in adults is 1952 counts per minute (7). Validation research with the LC in adults has indicated that an intensity category of 4-9 should be used when assessing MVPA (11). Results of this study indicate that a higher LC MVPA intensity category of 5-9 (as opposed to 4-9) improves the comparison of MVPA with those of AG_Treuth and AG_Puyau in 10-yr-old children. A criterion comparison was not conducted between the AG and LC outputs. However, nonsignificant differences based on the intuitive LC_5 threshold (higher MVPA range for children than for adults) with those of AG_Treuth and AG_Puyau provide a convergent indicator of the validity of the LC_5 MVPA intensity category. However, additional criterion validation of the LC_5 MVPA intensity category is warranted to confirm these findings.
MVPA estimates from AG_Trost and AG_Freedson range from two- to fourfold higher than the remaining four MVPA derivations. Given that no criterion measure of intensity was available for this study, we cannot be certain which derivation is most accurate. Higher relative estimates of time in MVPA from AG_Freedson compared with other intensity derivations is a result of the lower MVPA cut points derived from the Freedson et al. age-specific equation (for 10- to 11-yr-olds). It is probable that categorical intensity estimates based on AEE from AG_Trost may be impacted (in either direction) by error in either the prediction of EE or RMR for an individual. Specifically, underestimation of RMR would result in overestimates of MVPA, and overestimates of RMR would result in underestimates of MVPA. Therefore, using AEE derived from AG_Trost EE and estimated RMR as a categorical indicator of PA intensity within an epoch (rather than a measure of total energy expenditure during an epoch) may result in misclassification of epoch intensity dependent on the accuracy of individual EE and RMR estimates. Additional criterion validation is necessary to better understand variability in intensity derivations for children's time in MVPA, both within and between established and emerging accelerometers.
Several design features of the LC monitor may also be well suited for PA assessment in children. The LC display function can be set using the software so that the wearer can view either the time of day, steps, recent activity-intensity levels (intensity of the eight previous 4-s sampling intervals), or no display. Compared with devices that have no display, such as the AG, the versatility of the LC display would allow for appropriate use in either self-monitoring and behavior change interventions (using immediate feedback on steps and intensity from the display) or PA surveillance and research (using either no display or the time display to minimize reactivity to the monitor). Also, the interactive display of the LC may positively affect children's compliance with wearing the monitor (children may feel that it is more fun or meaningful to wear a device that provides them some form of immediate feedback). Another critical consideration for children's PA monitors is that the device should be wearable. If a monitor is uncomfortable to wear or requires additional equipment such as an elastic belt to secure it to the body, children's compliance with wearing the device may be reduced. The LC is attached by an existing plastic clip to the wearer's waistband at the midline of the thigh (similar to a pedometer). Future research investigating actual and perceived barriers to prolonged wearing of these and other PA monitors may provide additional insight into improving compliance with PA monitoring protocols.
The primary functional limitation of the LC versus the AG is the inability to extract LC data during specific time periods within a day. For example, AG data collected from a child during a school day could be segmented into distinct time-identified activity periods ultimately linked to each other, such as before or after school, or during lunch, recess, and PE. The LC, in contrast, is best suited for collecting complete days of data or short, defined periods of data (e.g., during a defined recess period only) and then immediately downloading the monitor. This limitation is a result of differences in raw data-storage processes between devices. The AG stores acceleration data from every epoch, and these data can be output in a time-synchronized file. The LC classifies a given time-sampling interval's intensity and totals it across 11 intensity categories during the monitoring period, but the monitor does not store real-time data. The LC does store the mode (most frequently occurring) intensity value during each 2-min time interval. However, this is of limited use beyond sustained periods of PA, because sedentary (coded as 0) or micro activity (coded as 0.5) behavior is likely to be the mode value recorded during intermittent activities that are more typical of PA participation in children (1). The general strengths of the LC, including reduced time cost and technical skills necessary for data processing, far outweigh this single limitation, which may not be relevant to many research or evaluation questions.
Future accelerometer-comparison studies in children should consider matching time-sampling intervals between monitors or using multiple monitors sampling at both typical and time-matched sampling intervals. This approach will be necessary to independently determine the extent to which a monitor's measurement properties (including intensity thresholds, data-filtering processes, and time-sampling intervals) contribute to differences in PA intensity outcome measures. The current study indicates that the LC_5 and either AG_Treuth or AG_Puyau provide similar mean estimates of MVPA during free-living activity in children. The small, nonsignificant differences seen between LC_5 and either AG_Treuth or AG_Puyau may be the results of differences in time-sampling intervals rather than intensity thresholds for MVPA. As new accelerometers emerge, it will be important to continually study the impact of such measurement properties' effects on interpretation of PA outcomes.
In summary, an interesting finding within this study was the relationship between DW, AG, and LC step counts. There were no significant differences between DW and LC or AG step counts. However, the DW registered a mean difference of 200 and 1001 fewer steps compared with the LC and AG, respectively. Additionally, LC and AG steps were significantly different with regard to steps taken. The two LC and three AG derivations of light-intensity activity resulted in a wide range of time estimates, suggesting that further criterion-validation research is needed to improve estimates of light-intensity PA in children. MVPA estimates from AG_Trost and AG_Freedson ranged from two- to fourfold higher than the remaining four MVPA derivations. Current results indicate that the LC_5 and either AG_Treuth or AG_Puyau provided similar mean estimates of MVPA during free-living activity in 10-yr-olds. In general, this comparison study demonstrates that the LC detects a similar number steps as the DW and significantly fewer steps than the AG, and the LC_5 intensity derivation provides similar estimates of MVPA to AG_Treuth and AG_Puyau estimates during free-living PA in 10-yr-old children.
1. Bailey, R. C., J. Olson, S. L. Pepper, J. Porszasz, T. J. Barstow, and D. M. Cooper. The level and tempo of children's physical activities: an observational study. Med. Sci. Sports Exerc.
2. Barfield, J. P., D. A. Rowe, and T. J. Michael. Interinstrument consistency of the Yamax Digi-Walker pedometer in elementary school-aged children. Meas. Phys. Educ. Exerc. Sci.
3. Bassett, D. R., Jr., B. E. Ainsworth, S. R. Leggett, et al. Accuracy of five electronic pedometers for measuring distance walked. Med. Sci. Sports Exerc.
4. Beets, M. W., M. M. Patton, and S. Edwards. The accuracy of pedometer steps and time during walking in children. Med. Sci. Sports Exerc.
5. Crouter, S. E., P. L. Schneider, M. Karabulut, and D. R. Bassett Jr. Validity of 10 electronic pedometers for measuring steps, distance, and energy cost. Med. Sci. Sports Exerc.
6. Freedson, P., D. Pober, and K. F. Janz. Calibration of accelerometer output for children. Med. Sci. Sports Exerc.
7. Freedson, P. S., E. Melanson, and J. Sirard. Calibration of the Computer Science and Applications, Inc. accelerometer. Med. Sci. Sports Exerc.
8. Freedson, P. S., J. Sirard, E. Debold, et al. Calibration of the Computer Science and Applications Inc. (CSA) accelerometer. Med. Sci. Sports Exerc.
9. Hedley, A. A., C. L. Ogden, C. L. Johnson, M. D. Carroll, L. R. Curtin, and K. M. Flegal. Prevalence of overweight and obesity among US children, adolescents, and adults, 1999-2002. JAMA
10. Janz, K. F., J. Witt, and L. T. Mahoney. The stability of children's physical activity as measured by accelerometry and self-report. Med. Sci. Sports Exerc.
11. Kumahara, H., Y. Schutz, M. Ayabe, et al. The use of uniaxial accelerometry for the assessment of physical-activity-related energy expenditure: a validation study against whole-body indirect calorimetry. Br. J. Nutr.
12. Le Masurier, G. C., S. M. Lee, and C. Tudor-Locke. Motion sensor accuracy under controlled and free-living conditions. Med. Sci. Sports Exerc.
13. Le Masurier, G. C., and C. Tudor-Locke. Comparison of pedometer and accelerometer accuracy under controlled conditions. Med. Sci. Sports Exerc.
14. Masse, L. C., B. F. Fuemmeler, C. B. Anderson, et al. Accelerometer data reduction: a comparison of four reduction algorithms on select outcome variables. Med. Sci. Sports Exerc.
15. Matthews, C. E. Calibration of accelerometer output for adults. Med. Sci. Sports Exerc.
16. National Association for Sport and Physical Education. Guidelines for Appropriate Physical Activity for Children
. 2003 update. Reston, VA: NASPE, 2003.
17. Puyau, M. R., A. L. Adolph, F. A. Vohra, and N. F. Butte. Validation and calibration of physical activity monitors in children. Obes. Res.
18. Puyau, M. R., A. L. Adolph, F. A. Vohra, I. Zakeri, and N. F. Butte. Prediction of activity energy expenditure using accelerometers in children. Med. Sci. Sports Exerc.
19. Schneider, P. L., S. E. Crouter, and D. R. Bassett. Pedometer measures of free-living physical activity: comparison of 13 models. Med. Sci. Sports Exerc.
20. Schneider, P. L., S. E. Crouter, O. Lukajic, and D. R. Bassett Jr. Accuracy and reliability of 10 pedometers for measuring steps over a 400-m walk. Med. Sci. Sports Exerc.
21. Schofield, W. N. Predicting basal metabolic rate, new standards and review of previous work. Hum. Nutr. Clin. Nutr.
39(Suppl 1):5-41, 1985.
22. Strath, S. J., D. R. Bassett, Jr, and A. M. Swartz. Comparison of MTI accelerometer cut-points for predicting time spent in physical activity. Int. J. Sports Med.
23. Strauss, R. S., and H. A. Pollack. Epidemic increase in childhood overweight, 1986-1998. JAMA
24. Treuth, M. S., K. Schmitz, D. J. Catellier, et al. Defining accelerometer thresholds for activity intensities in adolescent girls. Med. Sci. Sports Exerc.
25. Treuth, M. S., N. E. Sherwood, N. F. Butte, et al. Validity and reliability of activity measures in African-American girls for GEMS. Med. Sci. Sports Exerc.
26. Trost, S. G., K. L. McIver, and R. R. Pate. Conducting accelerometer-based activity assessments in field-based research. Med. Sci. Sports Exerc.
27. Trost, S. G., R. R. Pate, P. S. Freedson, J. F. Sallis, and W. C. Taylor. Using objective physical activity measures with youth: how many days of monitoring are needed? Med. Sci. Sports Exerc.
28. Trost, S. G., R. R. Pate, J. F. Sallis, et al. Age and gender differences in objectively measured physical activity in youth. Med. Sci. Sports Exerc.
29. Trost, S. G., D. S. Ward, S. M. Moorehead, P. D. Watson, W. Riner, and J. R. Burke. Validity of the computer science and applications (CSA) activity monitor in children. Med. Sci. Sports Exerc.
30. Tudor-Locke, C., B. E. Ainsworth, R. W. Thompson, and C. E. Matthews. Comparison of pedometer and accelerometer measures of free-living physical activity. Med. Sci. Sports Exerc.
31. United States Department of Health and Human Services. Physical Activity and Health: A Report of the Surgeon General
. Washington, DC: U.S. Government Printing Office, 1996, pp. 1-265.
32. Welk, G. J., S. N. Blair, K. Wood, S. Jones, and R. W. Thompson. A comparative evaluation of three accelerometry-based physical activity monitors. Med. Sci. Sports Exerc.