Over the past decade, accelerometers have increased in popularity as an objective measure of physical activity to be used in free-living studies. Accelerometers measure the acceleration and the deceleration of body movement and provide a direct assessment of the frequency, intensity, and duration of physical activity (26). Accelerometers use rapid sampling of accelerometer counts over a preset sampling period (e.g., 5 s) called an epoch.
Much of the research conducted with children and adolescents has used 60-s epochs (3). This is mainly because in the past, accelerometers were only capable of storing data collected using epoch lengths of <60 s for a limited number of days, and ≥7 d has been suggested as the most appropriate length for measuring physical activity objectively in children and adolescents (24). However, researchers have described children's physical activity as spontaneous and intermittent (5), with the majority of physical activity bouts lasting between 3 and 22 s (4,5). The use of 60-s epochs may therefore be inappropriate when measuring young people's physical activity and may result in an underestimation of moderate and vigorous physical activities (MPA and VPA, respectively) (25). This is possible because if a child alternates between vigorous and light physical activities (VPA and LPA, respectively) within a given minute, the accumulation of counts for that minute will only reflect the average activity level during that period resulting in a smoothing effect (22). Total volume of activity accumulated per day is not affected by epoch length; epoch length only becomes an issue when physical activity intensity is the outcome of interest.
Recent advances in accelerometer storage capacity have meant that researchers can select an epoch length of <60 s while still being able to measure physical activity for ≥7 d. Therefore, recent research measuring young people's physical activity is using a variety of different epoch lengths (16,19,21), and it is unclear how these various epoch lengths affect the recorded time spent in physical activity. Other decisions that researchers have to make regarding the number of hours that constitute a valid day, defining nonwearing time, deciding on the minimum number of days to be used in the analysis, and extracting moderate-to-vigorous physical activity (MVPA) by applying cut points to the data have also varied across studies and some have been shown to effect accelerometer output (12,14,15,18), which limits the ability to compare results across studies. These decisions can be collectively named postdata collection processing rules because these are choices that are made once physical activity data have been collected; however, selecting an epoch length is a choice that has to be made before data collection, and therefore it can perhaps be argued that selecting the most appropriate epoch length is the first critical decision to be made.
A small number of studies (17,18,20) have begun to investigate the effect of epoch length on physical activity intensities using small samples of children. Nilsson et al. (17) found that time spent in vigorous and very vigorous physical activities was inversely associated with epoch length in 16 children age 7 yr. Using the same methods, Reilly et al. (18), using data from 32 children age 5 and 6 yr, reported significant differences for MVPA between epoch settings but noted that these differences were small. In another study of twenty-five 7- to 11-yr-old children, Rowlands et al. (20) reported a similar finding for very vigorous physical activity but interestingly found that a 60-s epoch resulted in a significantly greater number of minutes of MPA and VPA. The empirical evidence on this topic is limited in children, and to date, no studies have examined the effect of epoch length on the measurement of adolescents' physical activity. Furthermore, with the use of different epoch lengths to measure young people's physical activity, it is not known if studies using different epoch lengths are comparable, and the most appropriate epoch length remains unclear.
The purpose of this study was to investigate the effect of different epoch lengths (5, 15, 30, and 60 s) on derived levels of physical activity in both a child and an adolescent sample to determine which epoch length would be most appropriate for use with young people. To do this, postdata collection processing rules will be held constant to concentrate on the effect of a decision that is made before data collection. A secondary aim is to investigate whether studies using different epoch lengths are comparable.
Participants consisted of two groups of young people: The first set of data were collected as baseline data for a larger study (11) involving eight primary schools throughout England. In total, 589 children age 7-11 yr provided consent from parents to participate, of which 311 children (49% male) wore an accelerometer. The second set of data were collected from adolescents recruited from three secondary schools in the East Midlands region of England. In total, 363 adolescents age 12-16 yr provided consent from parents to participate, of which 234 adolescents (56% male) wore an accelerometer. Staff at participating schools selected a subset of their classes for participation, and all pupils were eligible for participation. Before participation, informed consent was obtained from the parent or the guardian of all participants. Study procedures were approved by the Ethical Advisory Committee of Loughborough University.
All participants were asked to wear an ActiGraph GT1M accelerometer (Fort Walton Beach, FL) for 7 d during waking hours. The ActiGraph was initialized with a start and end time and a 5-s epoch. The accelerometer was attached to a flexible belt that was fastened around the participants' waist. After 7 d, accelerometers were collected, and data were uploaded to a data reduction program (ActiGraph Analysis Tool, Hadleigh, UK). A valid day was classified as ≥9 h of monitoring per day (8), and mean monitor wear time was 12.3 ± 1.0 and 13.2 ± 1.1 h·d−1 for children and adolescents, respectively. Participants with less than 4 d (three weekdays and one weekend day) of complete monitoring were excluded from the analysis (3). Missing data were defined as ≥20 min of consecutive zero counts (1). To determine time spent in rest, LPA, MPA, VPA, and MVPA the age-specific count ranges were derived from the energy expenditure prediction equation developed by Freedson et al. (10). Data, originally collected in 5-s epochs, were then reintegrated into 15-, 30-, and 60-s epochs using a software feature within the ActiGraph Analysis Tool, and data were reprocessed. The count ranges used for a 5-s epoch were 0-8, 9-74, 75-288, and >289 for rest, LPA, MPA, and VPA, respectively, for the child sample and 0-8, 9-134, 135-397, and >398 for the adolescent sample. The count ranges were then multiplied to correspond to the outcome of 15-, 30-, and 60-s epochs.
All analyses were conducted using the Statistical Package for the Social Sciences version 16.0 (SPSS Inc., Chicago, IL). Descriptive statistics were used to examine mean and standard deviation for time spent in rest, LPA, MPA, VPA, and MVPA for the various epoch lengths. One-way repeated-measures ANOVA examined the differences in time spent in the different intensity levels for 5-, 15-, 30-, and 60-s epochs. Post hoc Tukey tests followed any significant ANOVA. In cases when Mauchley's test revealed that the assumption of sphericity had been violated (P < 0.05), the Greenhouse-Geisser procedure was applied to adjust the degrees of freedom. To determine the degree of agreement between epochs for different physical activity intensity levels, the Bland-Altman method (6) was used. The significance level was set at P < 0.05.
The mean time spent in different physical activity intensity levels between each epoch length for both the child and the adolescent data are shown in Figures 1A-E. Using a one-way repeated-measures ANOVA, a significant epoch effect was seen for time spent in MVPA (F(1.13, 0.38) = 455.91), MPA (F(1.13, 0.38) = 929.51), VPA (F(1,29, 0.43) = 2081.04), LPA (F(1.11, 0.37) = 6467.84), and rest (F(1.08, 0.36) = 8384.49) in the child sample (all P < 0.05). Post hoc tests revealed that for time spent in all of these physical activity intensities, all epoch lengths significantly differed from each other (P < 0.05). For the adolescent sample, a significant main epoch effect was seen for time spent in VPA (F(1.33, 0.44) = 30.78, P < 0.05), LPA (F(1.31, 0.44) = 470.75, P < 0.05), and rest (F(1.39, 0.46) = 113.95, P < 0.05). Post hoc tests revealed that for time spent in all of these physical activity intensities, all epoch lengths significantly differed from each other (P < 0.05). No significant epoch effect was observed for time spent in MVPA and MPA between epoch lengths (P > 0.05) in the adolescent sample. To examine whether the epoch effect was the same for both boys and girls, analyses were run separately by gender. For boys and girls in both the child and the adolescent samples, results for time spent in MVPA, MPA, VPA, LPA, and rest showed identical patterns to those reported above (results available on request).
Figures 2 and 3 present the bias and 95% limits of agreement for the children and adolescent sample, respectively. In the child sample, bias was close to zero, and 95% limits of agreement were small between 5- and 15-, 15- and 30-, and 30- and 60-s epoch lengths for MVPA, MPA, VPA, LPA, and rest. In the adolescent sample, bias was close to zero, and 95% limits of agreement were small between 5- and 15-, 5- and 30-, and 15- and 30-s epoch lengths for MVPA, MPA, VPA, LPA, and rest.
The aim of this study was to determine the effect of integrating a 5-s epoch into a 15-, 30-, and 60-s epoch on different intensities of physical activity to determine which epoch length would be most appropriate for use with young people. A secondary aim was to investigate whether studies using different epoch lengths are comparable. This was done in both a child sample (7-11 yr) and an adolescent sample (12-16 yr) to investigate whether an epoch effect would be present across age groups.
In the child sample, a significant epoch effect was observed for MVPA, MPA, VPA, LPA, and rest. A shorter epoch was associated with less time in MVPA, MPA, and LPA. In contrast, for VPA and rest, a shorter epoch was associated with more time being spent in that intensity. Further analysis using the Bland-Altman method showed reasonable agreement between 5- and 15-, 15- and 30-, and 30- and 60-s epoch lengths for MVPA, MPA, VPA, LPA, and rest. For these epoch lengths, bias was closer to zero, indicating that the epoch lengths are producing similar results, and 95% limits of agreement were small, suggesting that comparison of physical activity levels between activity prevalence studies using these different epochs lengths could be made.
Previous research using child samples reported similar epoch effects for time spent in MVPA (18) and VPA (17,20). However, for other intensities, previous studies have found either no significant epoch effect or that a short epoch was associated with more time being spent in a given physical activity intensity, which contrasts with the current study's findings. The inconsistencies could be attributed to the differences in epoch lengths examined between the present study and the previous research. For example, Rowlands et al. (20) compared a 1- and 60-s epoch, Nilsson et al. (17) reintegrated a 5-s epoch into 10-, 20-, 40-, and 60-s epochs, and Reilly et al. (18) reintegrated a 15-s epoch into 30-, 45-, and 60-s epochs. Another contributing factor may be the large differences in sample sizes between studies.
It is recommended for children to achieve at least 60 min of MVPA daily, and although significant differences were observed in MVPA between epochs in the child sample, time spent in this intensity did not vary considerably. Using cut points on the basis of an age-specific equation (10), average time spent in MVPA ranged from 122.67 to 139.92 using a 5- to 60-s epoch. Therefore, the recommended guidelines are achieved regardless of the epoch length used, which is an agreement with previous research (18). This finding suggests that if the aim of a study is to measure MVPA, then choice of epoch length is less important. It is important to note, however, that the achievement of recommended guidelines is dependent on the cut points used to determine time spent in MVPA (15). As a result of using different published cut points, average time spent in MVPA may not have reached the recommended levels.
In the adolescent sample, a significant epoch effect was observed for time spent in VPA, LPA, and rest. For VPA and rest, a shorter epoch was associated with more time being spent in that intensity. In contrast, for LPA, a shorter epoch was associated with less time being spent in that intensity. Further analysis using the Bland-Altman method showed that considerable agreement was observed between 5- and 15-, 5- and 30-, and 15- and 30-s epoch lengths for MVPA, MPA, VPA, LPA, and rest. For these epoch lengths, bias was closer to zero, indicating that the epoch lengths are producing similar results, and 95% limits of agreement were small, suggesting that comparison of physical activity levels between activity prevalence studies using these different epochs lengths could be made.
Although significant differences in estimates of time spent in VPA, LPA, and rest in both the child and the adolescent samples and MVPA and MPA in just the child sample with different epoch lengths were found, the biological significance of these differences in accelerometer output is unclear. Further research examining whether these observed differences in estimates of time spent in different intensities of physical activity affect indices of health is needed.
Studies examining patterns of physical activity among children have concluded that children's physical activity is intermittent and characterized by rapid changes from rest to VPA. Bailey et al. (2) reported a mean duration of 6 s for LPA and MPA and 3 s for VPA. In agreement with this, Baquet et al. (4) found a mean duration of 9 s for LPA and MPA and 3 s for VPA, with 80% of MVPA bouts and 93% of VPA bouts lasting less than 10 s. The results of the current study demonstrate that using a 5-s epoch would be more effective than using longer epochs in detecting these short bouts of physical activity and would therefore enable the "real" patterns of children's physical activity to be measured. Further evidence to demonstrate that a 5-s epoch may be most appropriate for use with young people can be seen in bone health research. High intensities of strain to the musculoskeletal system appear to be more important than the volume of activity to bone development, and therefore short periods of intense activity would be particularly important (20). It is important therefore that these short intense periods are captured during physical activity measurement, and this current study demonstrates that a 5-s epoch would be the most appropriate epoch length to detect these types of physical activity. Furthermore, McClain et al. (13) found that a 5-s epoch along with applying the cut points of Freedson et al. (10) yielded similar mean estimates of MVPA compared with a direct observation criterion standard. Although they found that all Freedson epochs (5, 10, 15, 20, 30, and 60 s) yielded similar estimates of MVPA, a 5-s epoch yielded the lowest root mean squared error. It was concluded that short epoch lengths should be used to minimize error among individual estimates.
The unique contribution of the present study is the large sample size of 311 children and 234 adolescents. Previous research in this area included sample sizes ranging from 16 to 32 participants. Furthermore, to our knowledge, this is the first study to examine the effect of epoch length in an adolescent sample. However, as with previous research that has examined the effect of epoch length on physical activity intensity, the results of this study should be interpreted with some caution. This is because studies have used 60-s epochs when determining cut points for different physical activity intensities, and at present there is no evidence to support the validity of epoch-adjusted cut points (13). An additional limitation of this study was the lack of a criterion measure of physical activity intensity to allow for determination of which epoch length produced the most accurate estimate of physical activity.
Overall, the results of the present study suggest that if MVPA and the recommended guidelines for physical activity are the outcome of interest, then the choice of epoch length is less important. However, if time spent in individual intensities of physical activity and/or physical activity periods such as school break times and free play, in which physical activity is intermittent, are the main interest, then a short epoch would be recommended. This would provide researchers with a "real" picture of children's and adolescents' physical activity behavior and prevent accumulation of counts reflecting the average activity level when longer epochs are used. For example, if young people participate in short bouts of vigorous physical activity followed by longer bouts of LPA, the accumulation of counts for that minute will only reflect the average activity level during that period as a result of a smoothing effect (22). In addition, activity prevalence studies measuring physical activity at population levels that use epoch lengths of 5 and 60 s in a child or an adolescent sample should not be compared nor should 15- and 60-s epochs and 30- and 60-s epochs in an adolescent sample.
Several issues on epoch length still remain unclear. First, it is not known which epoch length produces the most accurate estimate of actual physical activity performed. To address this issue, comparison of time spent in rest, LPA, MPA, VPA, and MVPA against a criterion measure such as direct observation needs to be examined. Second, although this current study demonstrates that a 5-s epoch would be the most appropriate epoch length to detect short periods of intense physical activity, even shorter epoch lengths (i.e., 1- or 2-s epoch lengths) may be more appropriate and therefore require further investigation. Third, if epoch lengths other than 60 s are going to be used and consequently different physical activity intensities are determined, then research into the validity of epoch-adjusted cut points needs to be conducted. Finally, as noted by Cliff et al. (7), clarification of the biological significance of differences in estimates of MVPA, MPA, VPA, LPA, and rest according to differing epoch lengths is warranted.
The child phase of this research was financially supported by the Great Run and the Coca-Cola Company.
The authors acknowledge the assistance provided by Dr. Katherine Brooke-Wavell and all of the young people who took part in this study.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Anderson CB, Hagstromer M, Yngve A. Validation of the PDPAR as an adolescent diary: effect of accelerometer cut points. Med Sci Sports Exerc
2. Bailey RC, Olson J, Peper SL, Porszasz J, Barstow TJ, Cooper DM. The level and tempo of children's physical activities: an observation study. Med Sci Sports Exerc
3. Bagley S, Salmon J, Crawford D. Family structure and children's television viewing and physical activity. Med Sci Sports Exerc
4. Baquet G, Stratton G, Van Praagh E, Berthoin S. Improving physical activity assessment in prepubertal children with high-frequency accelerometer monitoring: a methodological issue. Prev Med
5. Berman N, Bailey R, Barstow TJ, Cooper DM. Spectral and bout detection analysis of physical activity patterns in healthy, prepubertal boys and girls. Am J Hum Biol
6. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet
7. Cliff DP, Reilly JJ, Okley AD. Methodological considerations in using accelerometers to assess habitual physical activity in children aged 0-5 years. J Sci Med Sport
8. Cradock AL, Wiecha JL, Peterson KE, Sobol AM, Colditz GA, Gortmaker SL. Youth recall and tritrac accelerometer estimates of physical activity levels. Med Sci Sports Exerc
10. Freedson PS, Sirard J, Debold E. Calibration of the Computer Science and Applications, Inc (CSA) accelerometer. Med Sci Sports Exerc
11. Gorely T, Nevill ME, Morris JG, Stensel DJ, Nevill A. Effect of a school-based intervention to promote healthy lifestyles in 7-11 year old children. Int J Behav Nutr Phys Act
12. Guinhouya CB, Hubert H, Soubrier S, Vilhelm C, Lemdani M, Durocher A. Moderate-to-vigorous physical activity among children: discrepancies in accelerometry-based cut-off points. Obesity
13. McClain JJ, Abraham TL, Brusseau TA, Tudor-Locke C. Epoch length and accelerometer outputs in children: comparison to direct observation. Med Sci Sports Exerc
14. Mattocks C, Ness A, Leary S, et al. Use of accelerometers in large field-based study of children: protocols, design issues, and effects on precision. J Phys Act Health
15. Mota J, Valente M, Aires L, Silva P, Santos MP, Ribeiro JC. Accelerometer cut-points and youth physical activity prevalence. Eur Phy Educ Rev
16. Ness AR, Leary SD, Mattcocks C, et al. Objectively measured physical activity and fat mass in a large cohort of children. PLoS Med
17. Nilsson A, Ekelund E, Yngve A, Sjöström M. Assessing physical activity among children with accelerometers using different time sampling intervals and placements. Pediatr Exerc Sci
18. Reilly JJ, Penpraze V, Hislop J, Davies G, Grant S, Paton JY. Objective measurement
of physical activity and sedentary behaviour: a review with new data. Arch Dis Child
19. Ridgers ND, Stratton G, Fairclough SJ, Twisk JWR. Children's physical activity levels during school recess: a quasi-experimental intervention study. Int J Behav Nutr Phys Act
20. Rowlands AV, Powell SM, Humpreys R, Eston RG. The effect of accelerometer epoch on physical activity output measures. J Exerc Sci Fit
21. Treuth MS, Catellier DJ, Schmitz KH, et al. Weekend and weekday patterns of physical activity in overweight and normal-weight adolescent girls. Obesity
22. Trost SG, Mciver KL, Pate RR. Conducting accelerometer-based activity assessment in field-based research. Med Sci Sports Exerc
24. Trost SG, Pate RR, Freedson PS, Sallis JF, Taylor WC. Using objective physical activity measures with youth: how many days of monitoring are needed? Med Sci Sports Exerc
25. Ward DS, Evenson KR, Vaughan A, Rodgers AB, Troiano RP. Accelerometer use in physical activity: best practices and research recommendations. Med Sci Sports Exerc
. 2005;37(11 suppl):S582-8.
26. Welk GJ, Almeida J, Morss G. Laboratory calibration and validation of the biotrainer and actitrac activity monitors. Med Sci Sports Exerc