- Email: [email protected]
Objective monitoring of physical activity in children: considerations for instrument selection
Available online at www.sciencedirect.com
Journal of Science and Medicine in Sport 12 (2009) 526–533
Opinion piece
Objective monitoring of physical activity in children: considerations for instrument selection James J. McClain a,∗ , Catrine Tudor-Locke b a
Cancer Prevention Fellowship Program, Office of Preventive Oncology, National Cancer Institute, National Institutes of Health, United States b Walking Behavior Laboratory, Pennington Biomedical Research Center, United States Received 14 May 2008; received in revised form 26 August 2008; accepted 11 September 2008
Abstract There has been a rapid recent increase in both the number and type of objective physical activity (PA) assessment instruments which are commercially available to researchers, practitioners, and consumers. Although this has provided improved capacity for PA assessment, it also presents a somewhat bewildering range of options related to instrument selection for users of these technologies. The purpose of this review is to provide a primer to guide selection of instruments for the objective monitoring of children’s PA. In an effort to inform without overwhelming, it is not intended to be exhaustive in terms of all available instruments. A general overview is provided of two primary categories of objective monitors: pedometers and accelerometers. Within each category we focus on distinctly relevant options and features important to consider during instrument selection. In general, the desired outcome measure will determine the specific instrument category, options, and features from which the ultimate instrument choice is made. Other considerations include evidence of validity and reliability, cost, computer interface and download options, memory capacity, data aggregation and storage methods, and general participant and researcher burden associated with instrument use. There is no single objective PA assessment instrument that is appropriate for all situations, populations, and research questions. Further, we can anticipate that the commercial nature of these instruments will drive an even greater range of features and options in the future, increasing both the opportunity and the challenge for objectively assessing PA in children. © 2008 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved. Keywords: Measurement; Physical activity; Children; Adolescent; Pedometer; Accelerometer
1. Introduction Over the past 20 years, improved awareness of the health benefits of physical activity (PA)17 has pressured development, validation, and application of new tools to objectively monitor this behaviour for the purpose of surveillance, intervention, or program evaluation. As a result, there has been a rapid increase in both the number and type of objective PA assessment instruments, including pedometers and accelerometers, which are commercially available to researchers, practitioners, and consumers.5,19 The end users of these technologies can be overwhelmed and confused when attempting to select from the array of instruments, options, and features offered. Instrument selection is further ∗
Corresponding author. E-mail address: [email protected] (J.J. McClain).
complicated for those who study children’s PA due to: (1) the challenge associated with detecting the typically short and sporadic nature of children’s PA patterns1 ; (2) the diversity of developmental maturity/age among potential participants (i.e., from infants and toddlers to adolescents); and, (3) children’s inherent curiosity regarding wearable technologies and the associated potential for reactivity to monitoring. As a result, researchers and practitioners scramble to make, at times, under-informed choices with regard to instrument selection. Therefore, the purpose of this review is to provide a primer to guide selection of instruments for the objective monitoring of children’s PA. In an effort to inform without overwhelming, it is not intended to be exhaustive in terms of all available instruments. We begin with an overview of the two primary categories of objective monitors: pedometers and accelerometers. Within each category we focus on
1440-2440/$ – see front matter © 2008 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.jsams.2008.09.012
Table 1 Summary of instrument categories, instrument specific data outputs, function, cost, and select validity and reliability references. Data outputs
Data aggregation and epoch lengths
Memory capacity
Instrument and accessories price ($US, as of 5/1/2008)
Select Refs validity/reliability
Yamax SW-200 (spring lever pedometer) www.new-lifestyles.com www.healthmg.com.au www.optimalhealthproducts.com Yamax PW-611 (piezo-electric pedometer) www.healthmg.com.au www.optimalhealthproducts.com
(1) Steps
Data aggregated since last manual reset
NA
Instrument: $17–30
8,27,33,34
(1) Steps
Data aggregated daily
7 days (on-instrument)
Instrument: $48–70
NA
(1) Steps
Data aggregated daily
7 days (on-instrument)
Instrument: $20–45
NA
(2) Aerobic steps (i.e., accumulated at >60 steps/min for >10 consecutive minutes) (1) Steps
Data aggregated daily
7 days (on-instrument)
HJ-150: $10 HJ-151: $13
NA
(2) Moderate steps (HJ-151) (3) MVPA (minutes) (HJ-151) (1) Steps
Data aggregated daily
7 days (on-instrument)
Instrument: $60
7,8,19,20,35
(1) Steps
Data aggregated daily
7 days (on-instrument)
Instrument: $43
NA
(1) Steps
Data aggregated daily (epoch length: 4 s)
7 days (on-instrument)
Instrument: $48
NA
Data aggregated daily (epoch length: 4 s)
7 days (on-instrument)
Instrument: $57–80
NA
(2) Movement time (3) Distance (4) Walking speed (5) PAEE Omron HJ-112/113 (piezo-electric pedometer) www.healthmg.com.au www.onlinefitness.com
Omron HJ-150 & 151 (piezo-electric pedometer) www.onlinefitness.com
NL-2000 (piezo-electric pedometer) www.new-lifestyles.com
(2) PAEE (3) TEE NL-800 (piezo-electric pedometer) www.new-lifestyles.com NL-1000 (accelerometer) www.new-lifestyles.com
(2) MVPA (min:s) (3) Distance Kenz Lifecorder e-Step (accelerometer) www.healthmg.com.au www.pedometerswithattitude.com
(1) Steps
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533
Instrument/selected distributor websites
(2) MVPA (min:s) (3) PAEE
527
9,23,37–41
42–44
Instrument: $335 USB cable and software: $349
Instrument: $450 infrared reader port and software: $500
2. Pedometers
Note: PAEE, physical activity energy expenditure; TDEE, total daily energy expenditure.
(2) Counts (3) PAEE
Data aggregated by epoch (epoch range: 15 s to 1 min) (1) Steps Actical (accelerometer) www.minimitter.com
(2) Counts (3) PAEE
Data aggregated by epoch (epoch range: 1 s to 5 min) (1) Steps ActiGraph (accelerometer) www.theActiGraph.com
(2) Frequency counts for epochs classified as sedentary, microactivity, and 9 PA intensity levels (3) PAEE (4) TDEE
distinctly relevant options and features important to consider during instrument selection. Table 1 summarises the primary instrument categories, options, and features covered herein. Finally, we discuss both participant and researcher burdens associated with objective PA assessment, specifically as it pertains to children.
5–22 days for 15 s to 1 min epochs (downloadable)
13,35,36
Instrument: $245–300 USB cable and software: $350 (1) Steps Kenz Lifecorder EX (accelerometer) www.healthmg.com.au www.new-lifestyles.com
Data aggregated daily (epoch length: 4 s)
7 days (on-instrument) 200 days (downloadable)
5–345 days for 1 s to 1 min epochs (downloadable)
Select Refs validity/reliability Memory capacity Data aggregation and epoch lengths Data outputs Instrument/selected distributor websites
Table 1 (Continued)
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533
Instrument and accessories price ($US, as of 5/1/2008)
528
As a category of objective PA assessment instruments, pedometers offer a typically simple and low cost estimate of total volume of PA which is outputted as the number of steps taken. The majority of pedometer instruments currently available detect steps by using a horizontal, spring-suspended lever arm which moves up and down with vertical accelerations of the hip.3 An event (i.e., step) is recorded when a sufficiently forceful (i.e., above the sensitivity threshold of the specific pedometer) vertical hip acceleration deflects the lever arm to complete an electronic circuit.3 Sensitivity thresholds for detecting steps can vary between available instruments.20 In general, the electronic circuitry within a pedometer accumulates steps and displays this information on a digital screen.3 A number of brand to brand comparisons have been conducted in adults10,14,19 but not in paediatric populations. Thirty of the 34 articles identified in a recent pedometry methods review article reported using the Yamax pedometer (Yamax Corporation, Tokyo, Japan) to determine free-living PA in youth.27 In adults the Yamax pedometer is considered a criterion pedometer, against which others may be compared.19 Although different models of the Yamax have been used, they do not differ on their internal step counting mechanism, just on their offerings of additional outputs (e.g., distance travel, energy expended). These additional outputs are produced through the imputation of an internal microprocessor that considers additional information (e.g., stride length) inputted beforehand by the user. From a measurement perspective, the process of manipulating the raw step data to derive estimates of distance or energy expenditure introduces additional layers of error.28 Pedometers are generally not designed to detect time in specific intensity categories (e.g., time in moderate-tovigorous PA; MVPA) and therefore are not an appropriate choice for end users whose specific research questions are focused on these parameters. That being said, in a more delimited activity time, for example, recess or physical education (PE) class time, the pedometer output of steps taken can be manipulated to derive steps/min.27 To this end, Scruggs et al.21,22 have worked to provide preliminary steps/min cut points potentially useful for interpreting children’s time in MVPA during PE class specifically. Pedometer manufacturers are also now beginning to offer additional features claiming estimates of activity time (e.g., accumulated time of stepping) and also time in MVPA (e.g., time accumulated above a specified stepping cadence). However, these new offerings
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533
have not been well validated in absolute terms and the degree to which such estimates agree with each other across different instruments is as yet unknown, especially in paediatric populations. Determination of MVPA from children’s stepping cadence (e.g., steps/min) during free-living is particularly challenging, due to the nature of their typical PA behaviours which consist of shorter duration and higher frequency movement bouts.1,2 Researchers and practitioners must remain in a “buyer beware” stance as there is no appointed authority to establish or regulate measurement standards. Despite these cautionary statements, pedometry as a whole continues to advance, especially with regards to the measurement of children’s PA. As noted above, young peoples’ activity patterns have been described as intermittent or sporadic, characterised by brief bursts of intense movement intermingled with bouts of light and sedentary activity.31 Given these aspects of young people’s movement behaviours, and the youth-appropriate public health focus on accumulation of daily PA,15 it follows that a cumulative record of steps taken over the course of the day is a suitable marker to track in youth. Further, pedometers are relatively inexpensive (∼ =$10– 50 per instrument, with a few exceptions) and are therefore feasible for widespread use by a range of researchers and practitioners working directly with young people. Recently, piezoelectric pedometers have emerged within the commercial market. In fact, many well-known pedometer manufacturers now offer at least one piezoelectric model, including the Yamax PW-611; New-Lifestyles NL-2000 and NL-800; and Omron models HJ-112, HJ-150, and HJ-151. These pedometers use a basic accelerometer-type mechanism to detect steps. Briefly, the mechanism consists of a horizontal suspended beam and a piezoelectric crystal which directly measures vertical accelerations, recording a step if detected an acceleration is above manufacturer-defined sensitivity thresholds.20 From a mechanical perspective, piezoelectric pedometers are similar to most other accelerometers (i.e., described in detail below). However, they are classified as pedometers since instrument outputs include only steps and variables derived from detected steps (e.g., estimates of PAEE, distance, or walking speed calculated from total steps and user inputs such as stride length; and intensity based on steps/min thresholds). Far less research is available for piezoelectric pedometers, due primarily to their recent market emergence. Initial research in adults7,14 and at least one study in children,8 have demonstrated improved precision in step counting with these new piezoelectric pedometers (i.e., versus spring-suspended lever arm models) at slower walking speeds and when tilted at increased angles from vertical (i.e., such as on participants with increased abdominal adiposity).
3. Accelerometers As noted above, many PA research questions and behaviour change goals may only require a simple indicator of total volume of daily PA participation, an output avail-
529
able from a standard pedometer. However, other questions or goals may be related to more specific components of PA such as the frequency, intensity and duration of daily PA. Intensity and duration of PA are of particular interest within surveillance research24 due to their relationship to current PA public health guidelines.17 A wide range of accelerometers are available which allow objective assessment of PA intensity and duration among children. A detailed discussion of the instrumentation and mechanical properties of accelerometer sensors is provided by Chen and Bassett.5 A brief overview is also provided below. Body worn accelerometers detect accelerations (measured in gravitational acceleration units (g; 1 g = 9.8 ms−2 ) at specific attachment points (such as the midaxillary line of the hip) on the body.30 Typically, a single sensor is positioned in line with the vertical axis of the body.5,30 However, some basic accelerometers use multiple sensors (e.g., contained within the same casing and positioned on the vertical, longitudinal, and/or transverse axis) to detect accelerations in multiple axes.5,6 The sensor used by most instruments consists of a piezoelectric element and a seismic mass within an enclosed casing.5 When the sensor is exposed to an acceleration, the seismic mass (which places force on the piezoelectric element) causes the piezoelectric element to deform (i.e., bend or compress depending on the structure of the particular sensor).5 This deformation produces a displaced and detectable electrical charge (positive or negative) to build up on one side of the sensor, generating a variable output voltage signal that is proportional to the applied acceleration.5 Voltage outputs are then converted into unitless numerical values typically called counts.5 Counts are a linear reflection of the sum of the voltage amplitude (i.e., a scalar measure of a wave signal’s magnitude of oscillation) detected. Counts are summed and stored (i.e., for most instruments) over a relatively brief length of time (typically ranging from 1 s up to 1 min) called an epoch or sampling interval.5 Additionally, the absolute number of combined positive and negative accelerations detected by raw voltage signal outputs is interpreted as number of steps taken and this distinct output can also be stored on an epoch-by-epoch basis by many accelerometers.5 The resulting epoch-by-epoch outputs of counts can be utilised in their raw form as a measure of activity volume (i.e., total counts) or activity rate (i.e., counts/min). They can also be transformed and/or re-coded to derive frequency, intensity and duration of PA, or PA energy expenditure (PAEE) estimates based on validated prediction models or count cutpoints.9,29 Details about this process are given below. However, interpreting raw data outputs (i.e., counts) as a primary indicator of PA participation can minimise error associated with data manipulations aimed at producing derived estimates of PAEE or intensity.30 Most accelerometer researchers have used count cutpoints based on validation research to derive time in intensity variables from raw count data.9,12 Using these cutpoints, epochs can then be categorised and the number of epochs in a specific intensity category (i.e., sedentary, light, moderate, or
530
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533
vigorous) can be multiplied by the epoch length to estimate accumulated time in specific intensities of PA. For example, if data were collected over a single day for a child using 30 s epochs, and counts recorded in 60 distinct epoch periods exceeded an established moderate count cutpoint, then we could estimate that child’s accumulated time in MVPA would be 1800 s (i.e., 30 s × 60 = 1800 s) or 30 min. Validation research has established children’s count cutpoints for moderate and vigorous PA for most accelerometers.9 The extent, however, that these studies provide appropriate cutpoints across the spectrum of paediatric populations from preschool to elementary to adolescent age groups varies by instrument.9 With the above caveat, sedentary and light intensity count cutpoints have also been established for some paediatric populations among specific instruments.9,23 When selecting an accelerometer to assess PA intensity and duration, it is important to determine if available count cutpoints have been validated in a similar population. Published reviews of child-specific accelerometer activity count cutpoints can also be used as a resource to inform instrument selection.9 The most widely used and extensively validated9 accelerometer for assessment of PA among children is the ActiGraph (ActiGraph LLC, Pensacola, FL, USA). The widespread use of the ActiGraph, in comparison to competing instruments such as the Actical (Mini Mitter, Bend, OR, USA), has propagated largely due to the above-mentioned disproportionate volume of available validation research. As a result, selection and usage of the ActiGraph has perhaps provided researchers the ability to collect data on a wider range of populations, due to the availability of validated count cutpoints necessary for producing derived PA intensity outcomes. However, there is relatively little functional difference between the ActiGraph and the Actical, in terms of their internal piezoelectric sensors’ ability to quantify accelerations associated with children’s PA. Typically, accelerometers range in price from $50 to over $400 per unit. The large discrepancy in costs between brands and/or models of accelerometers are primarily associated with differences in features such as personal computer (PC) interface and download options, memory capacity, and data aggregation and storage methods. A principle difference between many accelerometers is their ability to download data and their mode of interface with a PC. Most accelerometers models released in the 1990s, including the Actical and a previous version of the ActiGraph (model #7164), download to a PC using infrared reader ports or docking stations, with unit prices for these ancillary hardware requirements as high as $500. More recently released (or updated) downloadable accelerometers, such as the ActiGraph (model GT1 M) and the Lifecorder EX, have on-instrument Universal Service Bus (USB) ports for PC connectivity and generally have minimal additional costs for connection hardware (e.g., USB cable). In addition, instrument software prices for downloadable accelerometers typically range from $200 to $400.
A new class of low cost accelerometers (i.e., under $100) has recently emerged onto the commercial market including the New Lifestyles NL-1000 and the Kenz Lifecorder e-Step. These low cost accelerometers utilise the raw voltage output signal from their internal sensor to specifically classify PA intensity during each epoch using a modifiable intensity threshold, similar to higher cost instruments. This function differentiates the NL-1000 and e-Step from above mentioned piezoelectric pedometers (i.e., one of which, the NL-800, shares the same internal sensor), which simply record an event or step each time the raw voltage output signal exceeds a manufacturer-defined sensitivity thresholds. Both of these instruments output steps and time in MVPA as well as other outputs such as estimated distance (NL-1000) or PAEE (eStep). However, these low cost instruments do not have the ability to interface with or download to a PC. Instead, both have a LCD display screen similar to a pedometer, by which they display available outputs for the current day, or data stored in memory from the previous 7 days. For example, accumulated time in MVPA for each of the previous 7 days could be accessed on the display screen by cycling through the memory mode using the on-screen buttons. As mentioned above, the NL-1000 and e-Step both also allow the user/researcher to modify the manufacturer’s default settings for the MVPA intensity range, although this specific function is “hidden” and must be accessed using a specific button sequence on the keypad. This process reduces risk of accidental resetting by wearers. On the one hand, the ability to modify thresholds for MVPA intensity is of potential utility, allowing for population specific (e.g., preschool, elementary, adolescent, and even adult age groups) validation of optimal ranges. On the other hand, lack of a standardised usage of a single MVPA range can possibly impair direct comparisons between populations and between studies. Perhaps of greater concern is the lack of a verifiable record of the MVPA range used during data collection; such a record would be typical of data which is downloaded and processed in a post hoc manner. Little validation research is available on these emerging instruments in children, although both the NL-1000 and e-Step use the same internal sensor and intensity classification scheme as the Lifecorder EX, which has displayed convergent validity with MVPA outputs from several commonly used children’s ActiGraph count cutpoints.13 As mentioned above, nearly all pedometers are equipped with a digital screen display of instrument outputs (e.g., steps). In contrast, accelerometers have primarily been utilised as a blinded research instrument and few were capable of providing feedback to the user prior to downloading and post processing of data. Lifecorder EX (released in 2004) has a multi-function display which can be programmed during instrument initialisation to display a blank screen, or one, or all of its available outputs (i.e., steps, activity intensity, PAEE, TEE, and time). The optional screen displays on the Lifecorder EX, which allows the researcher to control feedback to the wearer, makes it well suited for
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533
use in both surveillance (by providing no direct feedback and thereby reducing potential for reactivity) and behaviour change interventions (by providing immediate feedback to the user). There is conflicting evidence of reactivity (i.e., increasing PA as a result of being monitored) in children’s PA assessment literature.4 Ultimately, however, the specific research question will dictate whether or not such feedback is preferable or not, and this in turn will inform instrument selection. Another important consideration for instrument selection is specific data aggregation, storage and memory features of an accelerometer. Data aggregation refers to a method of continually summing sampled data before it is stored in a permanent memory. Three principal methods and a range of data storage rates or epochs are utilised by objective assessment instruments. The three methods include aggregation of data: (1) since last manual reset, (2) daily, and (3) by epoch. Pedometers with no memory capacity are the only instruments which aggregate data exclusively since the last manual reset. Pedometers with memory (e.g., Yamax CW-600 or NL-800 for example) and at least three accelerometers (e.g., Lifecorder EX, NL-1000, and e-Step) continuously aggregate data and store it to permanent memory once daily. The principal limitation of this process of data aggregation and storage is that the ability to examine data from specific time segments of the day (e.g., during PE or recess for children) is lost unless data are self-recorded (if possible) at the time of transitions between activities.26 Data aggregation by epoch is the most common method used by accelerometers. This method allows data to be interpreted (i.e., using validated cutpoints or prediction equations) for short duration intervals and allows daily data to be segmented as desired (e.g., segmenting before school, in class, PE, recess, lunch and after school periods) to determine the contribution of specific periods of the day to total daily PA. Instrument epoch lengths are either manufacturerdetermined (i.e., unmodifiable) or researcher-selected (i.e., from a range of available epoch lengths). The ActiGraph has a minimum epoch length of 1 s, while the Actical has a 15 s minimum sampling interval. With adults, use of a 1 min epoch is typical for accelerometer assessment of PA.25 However, research suggests that due to children’s specific movement patterns (which tend to consist of shorter and more frequent bouts of PA)1 , use of epochs shorter than 1 min may be required16,25 . Therefore, the ability to use an epoch (user/researcher-selected or manufacturerdetermined) shorter than 1 min is a critical consideration in instrument selection for PA assessment in children.25 In addition, exploratory data processing methods under development are currently investigating the validity of high frequency sampling (i.e., using 1 s or 2 s epochs),2 and raw acceleration signal processing (i.e., extracting additional features such as median and peak acceleration from the raw signal collected at a rate of 32 Hz or samples/second, and coefficient of variation over a brief period such as 10 s)18 as methods for improving validity of PA outcomes.
531
A range of memory features and capacities are available for accelerometers. The only accelerometers which offer an on-instrument memory are the NL-1000 and e-Step (i.e., past 7 days and summed past week). All other accelerometers offer a downloadable memory of varying lengths. Accelerometer downloadable memory capacities are largely dependent on the data aggregation methods (e.g., daily or by epoch) and epoch length (i.e., shorter epoch lengths result in few days of memory capacity due to finite memory capacity). For the ActiGraph and Actical, data can be collected over 5 days at the minimum epoch lengths of 1 s and 15 s, respectively. Using 1 min epochs, the ActiGraph and Actical can collect data for over 1 year and 22 days, respectively. In comparison, the Lifecorder EX, which stores data daily, can collect data for over 200 days using its manufacturer-determined 4 s epochs. For instruments that have adopted (e.g., ActiGraph) or that choose to convert to using a rechargeable lithium ion battery as a power source (i.e., versus the typical coin cell battery used in most instruments), data collection is more likely to be limited by battery life than by memory capacity. For example, battery life for an ActiGraph with a full charge is approximately 14 days during data collection. However, these types of batteries can be recharged during non-wear periods (e.g., overnight) using a USB connection to a computer. However, this feature then demands that the wearer have ready access to a computer and be compliant with such a regimen, concerns that may limit this manner of use in the assessment of children’s PA.
4. Participant and researcher burden Participant burden during objective PA assessment is comprised of two primary factors: (1) ease and comfort of instrument wear, and (2) data recording requirements. With regard to the wearing issues, the majority of instruments (including most pedometers and many simple accelerometers) are mounted at the waist, are mostly unobtrusive, and cause relatively minimal participant burden or discomfort. Most waist-mounted pedometers and some accelerometers have pre-molded or manufactured plastic or metal belt clips which allow them to be easily attached to the waist line of clothing or to a belt. However, some instruments require the use of an additional elastic belt to properly fit the instrument to the wearer. Use of a separate elastic belt (in contrast to a manufactured on-instrument clip) may be considered as an additional burden to some children (i.e., due to comfort, inconvenience, or fashion). However, both an instrument and elastic belt could easily be hidden from view of peers by simply wearing it under an un-tucked shirt, or other outer layers of clothing, in most cases. Affixing instruments to elastic belts may actually allow children to more quickly and independently prepare, attach, and wear the instrument (e.g., in settings such a relatively brief PE class); and may reduce the chance of a child inadvertently losing the instrument.
532
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533
In general, objective assessment instruments reduce participant recording burden by aggregating PA data electronically during the day. In contrast, paper based PA recording logs (i.e., such as the Bouchard 3-day PA record) requires the participant to record PA behaviours on a continuous basis throughout the monitored day.32 However, as mentioned above, some objective instruments have either no memory or limited on-instrument memory. In these cases, participants may still be required to self-record instrument outputs (e.g., steps, time in MVPA, etc.) on a PA log.25 This is typically done at the end of the day, however, specific research questions may require more frequent recording (e.g., before and after recess, etc.). Instruments which require such participant attention may decrease compliance (and therefore require a prompting schedule) or compromise the integrity of collected data (due to the potential for factiously recorded data).25 Such efforts may be particularly challenging for younger children. Previous research in paediatric populations overcame the risk of factiously recorded data by sealing pedometers when distributed to children in schools, and having study personnel return to the school daily to unseal children’s pedometers, record data, and reseal them prior to returning to each child. However, the act of recording behaviour is a theoretically based element of a successful behaviour change intervention, if such a feature is required. While participant burden is decreased by using downloadable instruments, researcher burden is often greatly increased by the use of such instruments. This is caused by the immense volume of data outputted by many instruments (e.g., up to thousands of data points per day) and the need to use sophisticated, researcher-designed post-processing programs (e.g., using database or SAS macro-coding programs) to screen data for factors such as minimum wear time, off-body periods, and data collection errors.11 Therefore, post-data collection processing (i.e., in order to obtain desired outcome measures) is perhaps the largest source of researcher burden with regards to use of downloadable objective PA assessment instruments. Additional time costs to the researcher associated with instrument initialisation, downloading, and data processing are important practical factors to consider when selecting an instrument. However, improvements in manufacturers’ software and/or sharing of processing macros and codes among researchers have the potential to greatly reduce this burden. A researcher’s available time, technical expertise, and ability to conduct the above mentioned data post-processing must be considered when selecting an instrument.
5. Summary The rapid increase in both the number and type of objective PA assessment instruments has provided improved capacity for PA assessment and increased options for instrument selection for users of these technologies. However, it also presents a sometimes confusing array of options with regard to instrument selection for research and applied uses of PA
assessment instruments. Perhaps the most important factor with regards to instrument selection is a researcher or practitioner’s required or desired PA outcome measure. In general, this desired outcome measure will determine the specific instrument category, options, and features from which the ultimate instrument choice is made. Other considerations include evidence of validity and reliability, cost, CP interface and download options, memory capacity, data aggregation and storage methods, and general participant and researcher burden associated with instrument use. There is no single objective PA assessment instrument that is appropriate for all situations, populations, and research questions. Further, we can anticipate that the commercial nature of these instruments will drive an even greater range of features and options in the future, increasing both the opportunity and the challenge of objectively assessing PA in children.
Practical implications • This review provides an overview of two categories of objective physical activity assessment instruments commonly used in children: pedometers and accelerometers. • Options and features important to consider prior to instrument selection for children’s physical activity assessment are discussed, providing a primer to guide researchers and practitioners in selection of instruments.
Disclosure statement The information contained within this review does not constitute an endorsement or recommendation on the behalf of the authors, of any product, manufacturer, or distributor discussed herein. Authors have no financial or other interest in the products or distributors of the products reviewed. This work was not supported by any form of external financial support.
References 1. Bailey RC, Olson J, Pepper SL, Porszasz J, Barstow TJ, Cooper DM. The level and tempo of children’s physical activities: an observational study. Med Sci Sports Exerc 1995;27(7):1033–41. 2. Baquet G, Stratton G, Van Praagh E, Berthoin S. Improving physical activity assessment in prepubertal children with highfrequency accelerometry monitoring: a methodological issue. Prev Med 2007;44(2):143–7. 3. Bassett Jr DR, Strath SJ. Use of pedometers to assess physical activity. In: Welk GJ, editor. Physical activity assessment for health-related research. Champaign, IL: Human Kinetics; 2002. p. 163–78. 4. Behrens TK, Dinger MK. Motion sensor reactivity in physically active young adults. Res Q Exerc Sport 2007;78(2):1–8. 5. Chen KY, Bassett Jr DR. The technology of accelerometry-based activity monitors: current and future. Med Sci Sports Exerc 2005;37(11 Suppl):S490–500. 6. Chen KY, Sun M. Improving energy expenditure estimation by using a triaxial accelerometer. J Appl Physiol 1997;83(6):2112–22.
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533 7. Crouter SE, Schneider PL, Bassett Jr DR. Spring-levered versus piezoelectric pedometer accuracy in overweight and obese adults. Med Sci Sports Exerc 2005;37(10):1673–9. 8. Duncan JS, Schofield G, Duncan EK, Hinckson EA. Effects of age, walking speed, and body composition on pedometer accuracy in children. Res Q Exerc Sport 2007;78(5):420–8. 9. Freedson P, Pober D, Janz KF. Calibration of accelerometer output for children. Med Sci Sports Exerc 2005;37(11 Suppl):S523–30. 10. Le Masurier GC, Lee SM, Tudor-Locke C. Motion sensor accuracy under controlled and free-living conditions. Med Sci Sports Exerc 2004;36(5):905–10. 11. Masse LC, Fuemmeler BF, Anderson CB, Matthews CE, Trost SG, Catellier DJ, et al. Accelerometer data reduction: a comparison of four reduction algorithms on select outcome variables. Med Sci Sports Exerc 2005;37(11 Suppl):S544–54. 12. Matthews CE. Calibration of accelerometer output for adults. Med Sci Sports Exerc 2005;37(11 Suppl):S512–22. 13. McClain JJ, Sisson SB, Washington TL, Craig CL, Tudor-Locke C. Comparison of Kenz Lifecorder EX and ActiGraph accelerometers in 10-yr-old children. Med Sci Sports Exerc 2007;39(4):630–8. 14. Melanson EL, Knoll JR, Bell ML, Donahoo WT, Hill JO, Nysse LJ, et al. Commercially available pedometers: considerations for accurate step counting. Prev Med 2004;39(2):361–8. 15. National Association of Physical Education Sports. Guidelines for appropriate physical activity for elementary school children: update 2004. Reston, VA: National Association of Physical Education and Sports; 2004. 16. 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(1):87–96. 17. Pate RR, Pratt M, Blair SN, Haskell WL, Macera CA, Bouchard C, et al. Physical activity and public health. A recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine. JAMA 1995;273(5):402–7. 18. 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):1419–27. 19. Schneider PL, Crouter SE, Bassett DR. Pedometer measures of freeliving physical activity: comparison of 13 models. Med Sci Sports Exerc 2004;36(2):331–5. 20. Schneider PL, Crouter SE, Lukajic O, Bassett Jr DR. Accuracy and reliability of 10 pedometers for measuring steps over a 400-m walk. Med Sci Sports Exerc 2003;35(10):1779–84. 21. Scruggs PW, Beveridge SK, Eisenman PA, Watson DL, Shultz BB, Ransdell LB. Quantifying physical activity via pedometry in elementary physical education. Med Sci Sports Exerc 2003;35(6):1065–71. 22. Scruggs PW, Beveridge SK, Watson DL, Clocksin BD. Quantifying physical activity in first-through fourth-grade physical education via pedometry. Res Q Exerc Sport 2005;76(2):166–75. 23. Treuth MS, Schmitz K, Catellier DJ, McMurray RG, Murray DM, Almeida MJ, et al. Defining accelerometer thresholds for activity intensities in adolescent girls. Med Sci Sports Exerc 2004;36(7): 1259–66. 24. Troiano RP. A timely meeting: objective measurement of physical activity. Med Sci Sports Exerc 2005;37(11 Suppl):S487–9. 25. Trost SG, McIver KL, Pate RR. Conducting accelerometer-based activity assessments in field-based research. Med Sci Sports Exerc 2005;37(11 Suppl):S531–43.
533
26. 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):1732–8. 27. Tudor-Locke C, McClain JJ, Abraham TL, Sisson SB, Washington TL. Pedometry methods for assessing free-living youth. Res Q Exerc Sport in press. 28. Tudor-Locke CE, Myers AM. Methodological considerations for researchers and practitioners using pedometers to measure physical (ambulatory) activity. Res Q Exerc Sport 2001;72(1):1–12. 29. Welk GJ. Principles of design and analyses for the calibration of accelerometry-based activity monitors. Med Sci Sports Exerc 2005;37(11 Suppl):S501–11. 30. Welk GJ. Use of accelerometery-based activity monitors to assess physical activity. In: Welk GJ, editor. Physical activity assessment for health-related research. Champaign, IL: Human Kinetics; 2002. p. 125–41. 31. Welk GJ, Corbin CB, Dale D. Measurement issues in the assessment of physical activity in children. Res Q Exerc Sport 2000;71(Suppl 2):59–73. 32. Wickel EE, Welk GJ, Eisenmann JC. Concurrent validation of the Bouchard Diary with an accelerometry-based monitor. Med Sci Sports Exerc 2006;38(2):373–9. 33. Beets MW, Patton MM, Edwards S. The accuracy of pedometer steps and time during walking in children. Med Sci Sports Exerc 2005;37(3):513–20. 34. Oliver M, Schofield GM, Kolt GS, Schluter PJ. Pedometer accuracy in physical activity assessment of preschool children. J Sci Med Sport 2007;10(5):303–10. 35. Crouter SE, Schneider PL, Karabulut M, Bassett Jr DR. Validity of 10 electronic pedometers for measuring steps, distance, and energy cost. Med Sci Sports Exerc 2003;35(8):1455–60. 36. Kumahara H, Schutz Y, Ayabe M, Yoshioka M, Yoshitake Y, Shindo M, et al. The use of uniaxial accelerometry for the assessment of physicalactivity-related energy expenditure: a validation study against wholebody indirect calorimetry. Br J Nutr 2004;91(2):235–43. 37. Brage S, Brage N, Wedderkopp N, Froberg K. Reliablity and Validity of the Computer Science and Applications Accelerometer in a Mechanical Setting. Meas Phys Educ Exerc Sci 2003;7(2):101–19. 38. Mattocks C, Leary S, Ness A, Deere K, Saunders J, Tilling K, et al. Calibration of an accelerometer during free-living activities in children. Int J Pediatr Obes 2007;2(4):218–26. 39. Pate RR, Almeida MJ, McIver KL, Pfeiffer KA, Dowda M. Validation and calibration of an accelerometer in preschool children Obesity (Silver Spring) 2006;14(11):2000–6. 40. Trost SG, Way R, Okely AD. Predictive validity of three ActiGraph energy expenditure equations for children. Med Sci Sports Exerc 2006;38(2):380–7. 41. Welk GJ, Schaben JA, Morrow Jr JR. Reliability of accelerometrybased activity monitors: a generalizability study. Med Sci Sports Exerc 2004;36(9):1637–45. 42. Heil DP. Predicting activity energy expenditure using the Actical activity monitor. Res Q Exerc Sport 2006;77(1):64–80. 43. Pfeiffer KA, McIver KL, Dowda M, Almeida MJ, Pate RR. Validation and calibration of the Actical accelerometer in preschool children. Med Sci Sports Exerc 2006;38(1):152–7. 44. Puyau MR, Adolph AL, Vohra FA, Zakeri I, Butte NF. Prediction of activity energy expenditure using accelerometers in children. Med Sci Sports Exerc 2004;36(9):1625–31.
Journal of Science and Medicine in Sport 12 (2009) 526–533
Opinion piece
Objective monitoring of physical activity in children: considerations for instrument selection James J. McClain a,∗ , Catrine Tudor-Locke b a
Cancer Prevention Fellowship Program, Office of Preventive Oncology, National Cancer Institute, National Institutes of Health, United States b Walking Behavior Laboratory, Pennington Biomedical Research Center, United States Received 14 May 2008; received in revised form 26 August 2008; accepted 11 September 2008
Abstract There has been a rapid recent increase in both the number and type of objective physical activity (PA) assessment instruments which are commercially available to researchers, practitioners, and consumers. Although this has provided improved capacity for PA assessment, it also presents a somewhat bewildering range of options related to instrument selection for users of these technologies. The purpose of this review is to provide a primer to guide selection of instruments for the objective monitoring of children’s PA. In an effort to inform without overwhelming, it is not intended to be exhaustive in terms of all available instruments. A general overview is provided of two primary categories of objective monitors: pedometers and accelerometers. Within each category we focus on distinctly relevant options and features important to consider during instrument selection. In general, the desired outcome measure will determine the specific instrument category, options, and features from which the ultimate instrument choice is made. Other considerations include evidence of validity and reliability, cost, computer interface and download options, memory capacity, data aggregation and storage methods, and general participant and researcher burden associated with instrument use. There is no single objective PA assessment instrument that is appropriate for all situations, populations, and research questions. Further, we can anticipate that the commercial nature of these instruments will drive an even greater range of features and options in the future, increasing both the opportunity and the challenge for objectively assessing PA in children. © 2008 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved. Keywords: Measurement; Physical activity; Children; Adolescent; Pedometer; Accelerometer
1. Introduction Over the past 20 years, improved awareness of the health benefits of physical activity (PA)17 has pressured development, validation, and application of new tools to objectively monitor this behaviour for the purpose of surveillance, intervention, or program evaluation. As a result, there has been a rapid increase in both the number and type of objective PA assessment instruments, including pedometers and accelerometers, which are commercially available to researchers, practitioners, and consumers.5,19 The end users of these technologies can be overwhelmed and confused when attempting to select from the array of instruments, options, and features offered. Instrument selection is further ∗
Corresponding author. E-mail address: [email protected] (J.J. McClain).
complicated for those who study children’s PA due to: (1) the challenge associated with detecting the typically short and sporadic nature of children’s PA patterns1 ; (2) the diversity of developmental maturity/age among potential participants (i.e., from infants and toddlers to adolescents); and, (3) children’s inherent curiosity regarding wearable technologies and the associated potential for reactivity to monitoring. As a result, researchers and practitioners scramble to make, at times, under-informed choices with regard to instrument selection. Therefore, the purpose of this review is to provide a primer to guide selection of instruments for the objective monitoring of children’s PA. In an effort to inform without overwhelming, it is not intended to be exhaustive in terms of all available instruments. We begin with an overview of the two primary categories of objective monitors: pedometers and accelerometers. Within each category we focus on
1440-2440/$ – see front matter © 2008 Sports Medicine Australia. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.jsams.2008.09.012
Table 1 Summary of instrument categories, instrument specific data outputs, function, cost, and select validity and reliability references. Data outputs
Data aggregation and epoch lengths
Memory capacity
Instrument and accessories price ($US, as of 5/1/2008)
Select Refs validity/reliability
Yamax SW-200 (spring lever pedometer) www.new-lifestyles.com www.healthmg.com.au www.optimalhealthproducts.com Yamax PW-611 (piezo-electric pedometer) www.healthmg.com.au www.optimalhealthproducts.com
(1) Steps
Data aggregated since last manual reset
NA
Instrument: $17–30
8,27,33,34
(1) Steps
Data aggregated daily
7 days (on-instrument)
Instrument: $48–70
NA
(1) Steps
Data aggregated daily
7 days (on-instrument)
Instrument: $20–45
NA
(2) Aerobic steps (i.e., accumulated at >60 steps/min for >10 consecutive minutes) (1) Steps
Data aggregated daily
7 days (on-instrument)
HJ-150: $10 HJ-151: $13
NA
(2) Moderate steps (HJ-151) (3) MVPA (minutes) (HJ-151) (1) Steps
Data aggregated daily
7 days (on-instrument)
Instrument: $60
7,8,19,20,35
(1) Steps
Data aggregated daily
7 days (on-instrument)
Instrument: $43
NA
(1) Steps
Data aggregated daily (epoch length: 4 s)
7 days (on-instrument)
Instrument: $48
NA
Data aggregated daily (epoch length: 4 s)
7 days (on-instrument)
Instrument: $57–80
NA
(2) Movement time (3) Distance (4) Walking speed (5) PAEE Omron HJ-112/113 (piezo-electric pedometer) www.healthmg.com.au www.onlinefitness.com
Omron HJ-150 & 151 (piezo-electric pedometer) www.onlinefitness.com
NL-2000 (piezo-electric pedometer) www.new-lifestyles.com
(2) PAEE (3) TEE NL-800 (piezo-electric pedometer) www.new-lifestyles.com NL-1000 (accelerometer) www.new-lifestyles.com
(2) MVPA (min:s) (3) Distance Kenz Lifecorder e-Step (accelerometer) www.healthmg.com.au www.pedometerswithattitude.com
(1) Steps
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533
Instrument/selected distributor websites
(2) MVPA (min:s) (3) PAEE
527
9,23,37–41
42–44
Instrument: $335 USB cable and software: $349
Instrument: $450 infrared reader port and software: $500
2. Pedometers
Note: PAEE, physical activity energy expenditure; TDEE, total daily energy expenditure.
(2) Counts (3) PAEE
Data aggregated by epoch (epoch range: 15 s to 1 min) (1) Steps Actical (accelerometer) www.minimitter.com
(2) Counts (3) PAEE
Data aggregated by epoch (epoch range: 1 s to 5 min) (1) Steps ActiGraph (accelerometer) www.theActiGraph.com
(2) Frequency counts for epochs classified as sedentary, microactivity, and 9 PA intensity levels (3) PAEE (4) TDEE
distinctly relevant options and features important to consider during instrument selection. Table 1 summarises the primary instrument categories, options, and features covered herein. Finally, we discuss both participant and researcher burdens associated with objective PA assessment, specifically as it pertains to children.
5–22 days for 15 s to 1 min epochs (downloadable)
13,35,36
Instrument: $245–300 USB cable and software: $350 (1) Steps Kenz Lifecorder EX (accelerometer) www.healthmg.com.au www.new-lifestyles.com
Data aggregated daily (epoch length: 4 s)
7 days (on-instrument) 200 days (downloadable)
5–345 days for 1 s to 1 min epochs (downloadable)
Select Refs validity/reliability Memory capacity Data aggregation and epoch lengths Data outputs Instrument/selected distributor websites
Table 1 (Continued)
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533
Instrument and accessories price ($US, as of 5/1/2008)
528
As a category of objective PA assessment instruments, pedometers offer a typically simple and low cost estimate of total volume of PA which is outputted as the number of steps taken. The majority of pedometer instruments currently available detect steps by using a horizontal, spring-suspended lever arm which moves up and down with vertical accelerations of the hip.3 An event (i.e., step) is recorded when a sufficiently forceful (i.e., above the sensitivity threshold of the specific pedometer) vertical hip acceleration deflects the lever arm to complete an electronic circuit.3 Sensitivity thresholds for detecting steps can vary between available instruments.20 In general, the electronic circuitry within a pedometer accumulates steps and displays this information on a digital screen.3 A number of brand to brand comparisons have been conducted in adults10,14,19 but not in paediatric populations. Thirty of the 34 articles identified in a recent pedometry methods review article reported using the Yamax pedometer (Yamax Corporation, Tokyo, Japan) to determine free-living PA in youth.27 In adults the Yamax pedometer is considered a criterion pedometer, against which others may be compared.19 Although different models of the Yamax have been used, they do not differ on their internal step counting mechanism, just on their offerings of additional outputs (e.g., distance travel, energy expended). These additional outputs are produced through the imputation of an internal microprocessor that considers additional information (e.g., stride length) inputted beforehand by the user. From a measurement perspective, the process of manipulating the raw step data to derive estimates of distance or energy expenditure introduces additional layers of error.28 Pedometers are generally not designed to detect time in specific intensity categories (e.g., time in moderate-tovigorous PA; MVPA) and therefore are not an appropriate choice for end users whose specific research questions are focused on these parameters. That being said, in a more delimited activity time, for example, recess or physical education (PE) class time, the pedometer output of steps taken can be manipulated to derive steps/min.27 To this end, Scruggs et al.21,22 have worked to provide preliminary steps/min cut points potentially useful for interpreting children’s time in MVPA during PE class specifically. Pedometer manufacturers are also now beginning to offer additional features claiming estimates of activity time (e.g., accumulated time of stepping) and also time in MVPA (e.g., time accumulated above a specified stepping cadence). However, these new offerings
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533
have not been well validated in absolute terms and the degree to which such estimates agree with each other across different instruments is as yet unknown, especially in paediatric populations. Determination of MVPA from children’s stepping cadence (e.g., steps/min) during free-living is particularly challenging, due to the nature of their typical PA behaviours which consist of shorter duration and higher frequency movement bouts.1,2 Researchers and practitioners must remain in a “buyer beware” stance as there is no appointed authority to establish or regulate measurement standards. Despite these cautionary statements, pedometry as a whole continues to advance, especially with regards to the measurement of children’s PA. As noted above, young peoples’ activity patterns have been described as intermittent or sporadic, characterised by brief bursts of intense movement intermingled with bouts of light and sedentary activity.31 Given these aspects of young people’s movement behaviours, and the youth-appropriate public health focus on accumulation of daily PA,15 it follows that a cumulative record of steps taken over the course of the day is a suitable marker to track in youth. Further, pedometers are relatively inexpensive (∼ =$10– 50 per instrument, with a few exceptions) and are therefore feasible for widespread use by a range of researchers and practitioners working directly with young people. Recently, piezoelectric pedometers have emerged within the commercial market. In fact, many well-known pedometer manufacturers now offer at least one piezoelectric model, including the Yamax PW-611; New-Lifestyles NL-2000 and NL-800; and Omron models HJ-112, HJ-150, and HJ-151. These pedometers use a basic accelerometer-type mechanism to detect steps. Briefly, the mechanism consists of a horizontal suspended beam and a piezoelectric crystal which directly measures vertical accelerations, recording a step if detected an acceleration is above manufacturer-defined sensitivity thresholds.20 From a mechanical perspective, piezoelectric pedometers are similar to most other accelerometers (i.e., described in detail below). However, they are classified as pedometers since instrument outputs include only steps and variables derived from detected steps (e.g., estimates of PAEE, distance, or walking speed calculated from total steps and user inputs such as stride length; and intensity based on steps/min thresholds). Far less research is available for piezoelectric pedometers, due primarily to their recent market emergence. Initial research in adults7,14 and at least one study in children,8 have demonstrated improved precision in step counting with these new piezoelectric pedometers (i.e., versus spring-suspended lever arm models) at slower walking speeds and when tilted at increased angles from vertical (i.e., such as on participants with increased abdominal adiposity).
3. Accelerometers As noted above, many PA research questions and behaviour change goals may only require a simple indicator of total volume of daily PA participation, an output avail-
529
able from a standard pedometer. However, other questions or goals may be related to more specific components of PA such as the frequency, intensity and duration of daily PA. Intensity and duration of PA are of particular interest within surveillance research24 due to their relationship to current PA public health guidelines.17 A wide range of accelerometers are available which allow objective assessment of PA intensity and duration among children. A detailed discussion of the instrumentation and mechanical properties of accelerometer sensors is provided by Chen and Bassett.5 A brief overview is also provided below. Body worn accelerometers detect accelerations (measured in gravitational acceleration units (g; 1 g = 9.8 ms−2 ) at specific attachment points (such as the midaxillary line of the hip) on the body.30 Typically, a single sensor is positioned in line with the vertical axis of the body.5,30 However, some basic accelerometers use multiple sensors (e.g., contained within the same casing and positioned on the vertical, longitudinal, and/or transverse axis) to detect accelerations in multiple axes.5,6 The sensor used by most instruments consists of a piezoelectric element and a seismic mass within an enclosed casing.5 When the sensor is exposed to an acceleration, the seismic mass (which places force on the piezoelectric element) causes the piezoelectric element to deform (i.e., bend or compress depending on the structure of the particular sensor).5 This deformation produces a displaced and detectable electrical charge (positive or negative) to build up on one side of the sensor, generating a variable output voltage signal that is proportional to the applied acceleration.5 Voltage outputs are then converted into unitless numerical values typically called counts.5 Counts are a linear reflection of the sum of the voltage amplitude (i.e., a scalar measure of a wave signal’s magnitude of oscillation) detected. Counts are summed and stored (i.e., for most instruments) over a relatively brief length of time (typically ranging from 1 s up to 1 min) called an epoch or sampling interval.5 Additionally, the absolute number of combined positive and negative accelerations detected by raw voltage signal outputs is interpreted as number of steps taken and this distinct output can also be stored on an epoch-by-epoch basis by many accelerometers.5 The resulting epoch-by-epoch outputs of counts can be utilised in their raw form as a measure of activity volume (i.e., total counts) or activity rate (i.e., counts/min). They can also be transformed and/or re-coded to derive frequency, intensity and duration of PA, or PA energy expenditure (PAEE) estimates based on validated prediction models or count cutpoints.9,29 Details about this process are given below. However, interpreting raw data outputs (i.e., counts) as a primary indicator of PA participation can minimise error associated with data manipulations aimed at producing derived estimates of PAEE or intensity.30 Most accelerometer researchers have used count cutpoints based on validation research to derive time in intensity variables from raw count data.9,12 Using these cutpoints, epochs can then be categorised and the number of epochs in a specific intensity category (i.e., sedentary, light, moderate, or
530
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533
vigorous) can be multiplied by the epoch length to estimate accumulated time in specific intensities of PA. For example, if data were collected over a single day for a child using 30 s epochs, and counts recorded in 60 distinct epoch periods exceeded an established moderate count cutpoint, then we could estimate that child’s accumulated time in MVPA would be 1800 s (i.e., 30 s × 60 = 1800 s) or 30 min. Validation research has established children’s count cutpoints for moderate and vigorous PA for most accelerometers.9 The extent, however, that these studies provide appropriate cutpoints across the spectrum of paediatric populations from preschool to elementary to adolescent age groups varies by instrument.9 With the above caveat, sedentary and light intensity count cutpoints have also been established for some paediatric populations among specific instruments.9,23 When selecting an accelerometer to assess PA intensity and duration, it is important to determine if available count cutpoints have been validated in a similar population. Published reviews of child-specific accelerometer activity count cutpoints can also be used as a resource to inform instrument selection.9 The most widely used and extensively validated9 accelerometer for assessment of PA among children is the ActiGraph (ActiGraph LLC, Pensacola, FL, USA). The widespread use of the ActiGraph, in comparison to competing instruments such as the Actical (Mini Mitter, Bend, OR, USA), has propagated largely due to the above-mentioned disproportionate volume of available validation research. As a result, selection and usage of the ActiGraph has perhaps provided researchers the ability to collect data on a wider range of populations, due to the availability of validated count cutpoints necessary for producing derived PA intensity outcomes. However, there is relatively little functional difference between the ActiGraph and the Actical, in terms of their internal piezoelectric sensors’ ability to quantify accelerations associated with children’s PA. Typically, accelerometers range in price from $50 to over $400 per unit. The large discrepancy in costs between brands and/or models of accelerometers are primarily associated with differences in features such as personal computer (PC) interface and download options, memory capacity, and data aggregation and storage methods. A principle difference between many accelerometers is their ability to download data and their mode of interface with a PC. Most accelerometers models released in the 1990s, including the Actical and a previous version of the ActiGraph (model #7164), download to a PC using infrared reader ports or docking stations, with unit prices for these ancillary hardware requirements as high as $500. More recently released (or updated) downloadable accelerometers, such as the ActiGraph (model GT1 M) and the Lifecorder EX, have on-instrument Universal Service Bus (USB) ports for PC connectivity and generally have minimal additional costs for connection hardware (e.g., USB cable). In addition, instrument software prices for downloadable accelerometers typically range from $200 to $400.
A new class of low cost accelerometers (i.e., under $100) has recently emerged onto the commercial market including the New Lifestyles NL-1000 and the Kenz Lifecorder e-Step. These low cost accelerometers utilise the raw voltage output signal from their internal sensor to specifically classify PA intensity during each epoch using a modifiable intensity threshold, similar to higher cost instruments. This function differentiates the NL-1000 and e-Step from above mentioned piezoelectric pedometers (i.e., one of which, the NL-800, shares the same internal sensor), which simply record an event or step each time the raw voltage output signal exceeds a manufacturer-defined sensitivity thresholds. Both of these instruments output steps and time in MVPA as well as other outputs such as estimated distance (NL-1000) or PAEE (eStep). However, these low cost instruments do not have the ability to interface with or download to a PC. Instead, both have a LCD display screen similar to a pedometer, by which they display available outputs for the current day, or data stored in memory from the previous 7 days. For example, accumulated time in MVPA for each of the previous 7 days could be accessed on the display screen by cycling through the memory mode using the on-screen buttons. As mentioned above, the NL-1000 and e-Step both also allow the user/researcher to modify the manufacturer’s default settings for the MVPA intensity range, although this specific function is “hidden” and must be accessed using a specific button sequence on the keypad. This process reduces risk of accidental resetting by wearers. On the one hand, the ability to modify thresholds for MVPA intensity is of potential utility, allowing for population specific (e.g., preschool, elementary, adolescent, and even adult age groups) validation of optimal ranges. On the other hand, lack of a standardised usage of a single MVPA range can possibly impair direct comparisons between populations and between studies. Perhaps of greater concern is the lack of a verifiable record of the MVPA range used during data collection; such a record would be typical of data which is downloaded and processed in a post hoc manner. Little validation research is available on these emerging instruments in children, although both the NL-1000 and e-Step use the same internal sensor and intensity classification scheme as the Lifecorder EX, which has displayed convergent validity with MVPA outputs from several commonly used children’s ActiGraph count cutpoints.13 As mentioned above, nearly all pedometers are equipped with a digital screen display of instrument outputs (e.g., steps). In contrast, accelerometers have primarily been utilised as a blinded research instrument and few were capable of providing feedback to the user prior to downloading and post processing of data. Lifecorder EX (released in 2004) has a multi-function display which can be programmed during instrument initialisation to display a blank screen, or one, or all of its available outputs (i.e., steps, activity intensity, PAEE, TEE, and time). The optional screen displays on the Lifecorder EX, which allows the researcher to control feedback to the wearer, makes it well suited for
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533
use in both surveillance (by providing no direct feedback and thereby reducing potential for reactivity) and behaviour change interventions (by providing immediate feedback to the user). There is conflicting evidence of reactivity (i.e., increasing PA as a result of being monitored) in children’s PA assessment literature.4 Ultimately, however, the specific research question will dictate whether or not such feedback is preferable or not, and this in turn will inform instrument selection. Another important consideration for instrument selection is specific data aggregation, storage and memory features of an accelerometer. Data aggregation refers to a method of continually summing sampled data before it is stored in a permanent memory. Three principal methods and a range of data storage rates or epochs are utilised by objective assessment instruments. The three methods include aggregation of data: (1) since last manual reset, (2) daily, and (3) by epoch. Pedometers with no memory capacity are the only instruments which aggregate data exclusively since the last manual reset. Pedometers with memory (e.g., Yamax CW-600 or NL-800 for example) and at least three accelerometers (e.g., Lifecorder EX, NL-1000, and e-Step) continuously aggregate data and store it to permanent memory once daily. The principal limitation of this process of data aggregation and storage is that the ability to examine data from specific time segments of the day (e.g., during PE or recess for children) is lost unless data are self-recorded (if possible) at the time of transitions between activities.26 Data aggregation by epoch is the most common method used by accelerometers. This method allows data to be interpreted (i.e., using validated cutpoints or prediction equations) for short duration intervals and allows daily data to be segmented as desired (e.g., segmenting before school, in class, PE, recess, lunch and after school periods) to determine the contribution of specific periods of the day to total daily PA. Instrument epoch lengths are either manufacturerdetermined (i.e., unmodifiable) or researcher-selected (i.e., from a range of available epoch lengths). The ActiGraph has a minimum epoch length of 1 s, while the Actical has a 15 s minimum sampling interval. With adults, use of a 1 min epoch is typical for accelerometer assessment of PA.25 However, research suggests that due to children’s specific movement patterns (which tend to consist of shorter and more frequent bouts of PA)1 , use of epochs shorter than 1 min may be required16,25 . Therefore, the ability to use an epoch (user/researcher-selected or manufacturerdetermined) shorter than 1 min is a critical consideration in instrument selection for PA assessment in children.25 In addition, exploratory data processing methods under development are currently investigating the validity of high frequency sampling (i.e., using 1 s or 2 s epochs),2 and raw acceleration signal processing (i.e., extracting additional features such as median and peak acceleration from the raw signal collected at a rate of 32 Hz or samples/second, and coefficient of variation over a brief period such as 10 s)18 as methods for improving validity of PA outcomes.
531
A range of memory features and capacities are available for accelerometers. The only accelerometers which offer an on-instrument memory are the NL-1000 and e-Step (i.e., past 7 days and summed past week). All other accelerometers offer a downloadable memory of varying lengths. Accelerometer downloadable memory capacities are largely dependent on the data aggregation methods (e.g., daily or by epoch) and epoch length (i.e., shorter epoch lengths result in few days of memory capacity due to finite memory capacity). For the ActiGraph and Actical, data can be collected over 5 days at the minimum epoch lengths of 1 s and 15 s, respectively. Using 1 min epochs, the ActiGraph and Actical can collect data for over 1 year and 22 days, respectively. In comparison, the Lifecorder EX, which stores data daily, can collect data for over 200 days using its manufacturer-determined 4 s epochs. For instruments that have adopted (e.g., ActiGraph) or that choose to convert to using a rechargeable lithium ion battery as a power source (i.e., versus the typical coin cell battery used in most instruments), data collection is more likely to be limited by battery life than by memory capacity. For example, battery life for an ActiGraph with a full charge is approximately 14 days during data collection. However, these types of batteries can be recharged during non-wear periods (e.g., overnight) using a USB connection to a computer. However, this feature then demands that the wearer have ready access to a computer and be compliant with such a regimen, concerns that may limit this manner of use in the assessment of children’s PA.
4. Participant and researcher burden Participant burden during objective PA assessment is comprised of two primary factors: (1) ease and comfort of instrument wear, and (2) data recording requirements. With regard to the wearing issues, the majority of instruments (including most pedometers and many simple accelerometers) are mounted at the waist, are mostly unobtrusive, and cause relatively minimal participant burden or discomfort. Most waist-mounted pedometers and some accelerometers have pre-molded or manufactured plastic or metal belt clips which allow them to be easily attached to the waist line of clothing or to a belt. However, some instruments require the use of an additional elastic belt to properly fit the instrument to the wearer. Use of a separate elastic belt (in contrast to a manufactured on-instrument clip) may be considered as an additional burden to some children (i.e., due to comfort, inconvenience, or fashion). However, both an instrument and elastic belt could easily be hidden from view of peers by simply wearing it under an un-tucked shirt, or other outer layers of clothing, in most cases. Affixing instruments to elastic belts may actually allow children to more quickly and independently prepare, attach, and wear the instrument (e.g., in settings such a relatively brief PE class); and may reduce the chance of a child inadvertently losing the instrument.
532
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533
In general, objective assessment instruments reduce participant recording burden by aggregating PA data electronically during the day. In contrast, paper based PA recording logs (i.e., such as the Bouchard 3-day PA record) requires the participant to record PA behaviours on a continuous basis throughout the monitored day.32 However, as mentioned above, some objective instruments have either no memory or limited on-instrument memory. In these cases, participants may still be required to self-record instrument outputs (e.g., steps, time in MVPA, etc.) on a PA log.25 This is typically done at the end of the day, however, specific research questions may require more frequent recording (e.g., before and after recess, etc.). Instruments which require such participant attention may decrease compliance (and therefore require a prompting schedule) or compromise the integrity of collected data (due to the potential for factiously recorded data).25 Such efforts may be particularly challenging for younger children. Previous research in paediatric populations overcame the risk of factiously recorded data by sealing pedometers when distributed to children in schools, and having study personnel return to the school daily to unseal children’s pedometers, record data, and reseal them prior to returning to each child. However, the act of recording behaviour is a theoretically based element of a successful behaviour change intervention, if such a feature is required. While participant burden is decreased by using downloadable instruments, researcher burden is often greatly increased by the use of such instruments. This is caused by the immense volume of data outputted by many instruments (e.g., up to thousands of data points per day) and the need to use sophisticated, researcher-designed post-processing programs (e.g., using database or SAS macro-coding programs) to screen data for factors such as minimum wear time, off-body periods, and data collection errors.11 Therefore, post-data collection processing (i.e., in order to obtain desired outcome measures) is perhaps the largest source of researcher burden with regards to use of downloadable objective PA assessment instruments. Additional time costs to the researcher associated with instrument initialisation, downloading, and data processing are important practical factors to consider when selecting an instrument. However, improvements in manufacturers’ software and/or sharing of processing macros and codes among researchers have the potential to greatly reduce this burden. A researcher’s available time, technical expertise, and ability to conduct the above mentioned data post-processing must be considered when selecting an instrument.
5. Summary The rapid increase in both the number and type of objective PA assessment instruments has provided improved capacity for PA assessment and increased options for instrument selection for users of these technologies. However, it also presents a sometimes confusing array of options with regard to instrument selection for research and applied uses of PA
assessment instruments. Perhaps the most important factor with regards to instrument selection is a researcher or practitioner’s required or desired PA outcome measure. In general, this desired outcome measure will determine the specific instrument category, options, and features from which the ultimate instrument choice is made. Other considerations include evidence of validity and reliability, cost, CP interface and download options, memory capacity, data aggregation and storage methods, and general participant and researcher burden associated with instrument use. There is no single objective PA assessment instrument that is appropriate for all situations, populations, and research questions. Further, we can anticipate that the commercial nature of these instruments will drive an even greater range of features and options in the future, increasing both the opportunity and the challenge of objectively assessing PA in children.
Practical implications • This review provides an overview of two categories of objective physical activity assessment instruments commonly used in children: pedometers and accelerometers. • Options and features important to consider prior to instrument selection for children’s physical activity assessment are discussed, providing a primer to guide researchers and practitioners in selection of instruments.
Disclosure statement The information contained within this review does not constitute an endorsement or recommendation on the behalf of the authors, of any product, manufacturer, or distributor discussed herein. Authors have no financial or other interest in the products or distributors of the products reviewed. This work was not supported by any form of external financial support.
References 1. Bailey RC, Olson J, Pepper SL, Porszasz J, Barstow TJ, Cooper DM. The level and tempo of children’s physical activities: an observational study. Med Sci Sports Exerc 1995;27(7):1033–41. 2. Baquet G, Stratton G, Van Praagh E, Berthoin S. Improving physical activity assessment in prepubertal children with highfrequency accelerometry monitoring: a methodological issue. Prev Med 2007;44(2):143–7. 3. Bassett Jr DR, Strath SJ. Use of pedometers to assess physical activity. In: Welk GJ, editor. Physical activity assessment for health-related research. Champaign, IL: Human Kinetics; 2002. p. 163–78. 4. Behrens TK, Dinger MK. Motion sensor reactivity in physically active young adults. Res Q Exerc Sport 2007;78(2):1–8. 5. Chen KY, Bassett Jr DR. The technology of accelerometry-based activity monitors: current and future. Med Sci Sports Exerc 2005;37(11 Suppl):S490–500. 6. Chen KY, Sun M. Improving energy expenditure estimation by using a triaxial accelerometer. J Appl Physiol 1997;83(6):2112–22.
J.J. McClain, C. Tudor-Locke / Journal of Science and Medicine in Sport 12 (2009) 526–533 7. Crouter SE, Schneider PL, Bassett Jr DR. Spring-levered versus piezoelectric pedometer accuracy in overweight and obese adults. Med Sci Sports Exerc 2005;37(10):1673–9. 8. Duncan JS, Schofield G, Duncan EK, Hinckson EA. Effects of age, walking speed, and body composition on pedometer accuracy in children. Res Q Exerc Sport 2007;78(5):420–8. 9. Freedson P, Pober D, Janz KF. Calibration of accelerometer output for children. Med Sci Sports Exerc 2005;37(11 Suppl):S523–30. 10. Le Masurier GC, Lee SM, Tudor-Locke C. Motion sensor accuracy under controlled and free-living conditions. Med Sci Sports Exerc 2004;36(5):905–10. 11. Masse LC, Fuemmeler BF, Anderson CB, Matthews CE, Trost SG, Catellier DJ, et al. Accelerometer data reduction: a comparison of four reduction algorithms on select outcome variables. Med Sci Sports Exerc 2005;37(11 Suppl):S544–54. 12. Matthews CE. Calibration of accelerometer output for adults. Med Sci Sports Exerc 2005;37(11 Suppl):S512–22. 13. McClain JJ, Sisson SB, Washington TL, Craig CL, Tudor-Locke C. Comparison of Kenz Lifecorder EX and ActiGraph accelerometers in 10-yr-old children. Med Sci Sports Exerc 2007;39(4):630–8. 14. Melanson EL, Knoll JR, Bell ML, Donahoo WT, Hill JO, Nysse LJ, et al. Commercially available pedometers: considerations for accurate step counting. Prev Med 2004;39(2):361–8. 15. National Association of Physical Education Sports. Guidelines for appropriate physical activity for elementary school children: update 2004. Reston, VA: National Association of Physical Education and Sports; 2004. 16. 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(1):87–96. 17. Pate RR, Pratt M, Blair SN, Haskell WL, Macera CA, Bouchard C, et al. Physical activity and public health. A recommendation from the Centers for Disease Control and Prevention and the American College of Sports Medicine. JAMA 1995;273(5):402–7. 18. 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):1419–27. 19. Schneider PL, Crouter SE, Bassett DR. Pedometer measures of freeliving physical activity: comparison of 13 models. Med Sci Sports Exerc 2004;36(2):331–5. 20. Schneider PL, Crouter SE, Lukajic O, Bassett Jr DR. Accuracy and reliability of 10 pedometers for measuring steps over a 400-m walk. Med Sci Sports Exerc 2003;35(10):1779–84. 21. Scruggs PW, Beveridge SK, Eisenman PA, Watson DL, Shultz BB, Ransdell LB. Quantifying physical activity via pedometry in elementary physical education. Med Sci Sports Exerc 2003;35(6):1065–71. 22. Scruggs PW, Beveridge SK, Watson DL, Clocksin BD. Quantifying physical activity in first-through fourth-grade physical education via pedometry. Res Q Exerc Sport 2005;76(2):166–75. 23. Treuth MS, Schmitz K, Catellier DJ, McMurray RG, Murray DM, Almeida MJ, et al. Defining accelerometer thresholds for activity intensities in adolescent girls. Med Sci Sports Exerc 2004;36(7): 1259–66. 24. Troiano RP. A timely meeting: objective measurement of physical activity. Med Sci Sports Exerc 2005;37(11 Suppl):S487–9. 25. Trost SG, McIver KL, Pate RR. Conducting accelerometer-based activity assessments in field-based research. Med Sci Sports Exerc 2005;37(11 Suppl):S531–43.
533
26. 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):1732–8. 27. Tudor-Locke C, McClain JJ, Abraham TL, Sisson SB, Washington TL. Pedometry methods for assessing free-living youth. Res Q Exerc Sport in press. 28. Tudor-Locke CE, Myers AM. Methodological considerations for researchers and practitioners using pedometers to measure physical (ambulatory) activity. Res Q Exerc Sport 2001;72(1):1–12. 29. Welk GJ. Principles of design and analyses for the calibration of accelerometry-based activity monitors. Med Sci Sports Exerc 2005;37(11 Suppl):S501–11. 30. Welk GJ. Use of accelerometery-based activity monitors to assess physical activity. In: Welk GJ, editor. Physical activity assessment for health-related research. Champaign, IL: Human Kinetics; 2002. p. 125–41. 31. Welk GJ, Corbin CB, Dale D. Measurement issues in the assessment of physical activity in children. Res Q Exerc Sport 2000;71(Suppl 2):59–73. 32. Wickel EE, Welk GJ, Eisenmann JC. Concurrent validation of the Bouchard Diary with an accelerometry-based monitor. Med Sci Sports Exerc 2006;38(2):373–9. 33. Beets MW, Patton MM, Edwards S. The accuracy of pedometer steps and time during walking in children. Med Sci Sports Exerc 2005;37(3):513–20. 34. Oliver M, Schofield GM, Kolt GS, Schluter PJ. Pedometer accuracy in physical activity assessment of preschool children. J Sci Med Sport 2007;10(5):303–10. 35. Crouter SE, Schneider PL, Karabulut M, Bassett Jr DR. Validity of 10 electronic pedometers for measuring steps, distance, and energy cost. Med Sci Sports Exerc 2003;35(8):1455–60. 36. Kumahara H, Schutz Y, Ayabe M, Yoshioka M, Yoshitake Y, Shindo M, et al. The use of uniaxial accelerometry for the assessment of physicalactivity-related energy expenditure: a validation study against wholebody indirect calorimetry. Br J Nutr 2004;91(2):235–43. 37. Brage S, Brage N, Wedderkopp N, Froberg K. Reliablity and Validity of the Computer Science and Applications Accelerometer in a Mechanical Setting. Meas Phys Educ Exerc Sci 2003;7(2):101–19. 38. Mattocks C, Leary S, Ness A, Deere K, Saunders J, Tilling K, et al. Calibration of an accelerometer during free-living activities in children. Int J Pediatr Obes 2007;2(4):218–26. 39. Pate RR, Almeida MJ, McIver KL, Pfeiffer KA, Dowda M. Validation and calibration of an accelerometer in preschool children Obesity (Silver Spring) 2006;14(11):2000–6. 40. Trost SG, Way R, Okely AD. Predictive validity of three ActiGraph energy expenditure equations for children. Med Sci Sports Exerc 2006;38(2):380–7. 41. Welk GJ, Schaben JA, Morrow Jr JR. Reliability of accelerometrybased activity monitors: a generalizability study. Med Sci Sports Exerc 2004;36(9):1637–45. 42. Heil DP. Predicting activity energy expenditure using the Actical activity monitor. Res Q Exerc Sport 2006;77(1):64–80. 43. Pfeiffer KA, McIver KL, Dowda M, Almeida MJ, Pate RR. Validation and calibration of the Actical accelerometer in preschool children. Med Sci Sports Exerc 2006;38(1):152–7. 44. Puyau MR, Adolph AL, Vohra FA, Zakeri I, Butte NF. Prediction of activity energy expenditure using accelerometers in children. Med Sci Sports Exerc 2004;36(9):1625–31.