Issuu on Google+

This article was downloaded by: [University of Otago] On: 27 October 2011, At: 17:31 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Ergonomics Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/terg20

Can accelerometry be used to discriminate levels of activity? a

b

c

P. Hendrick , M.L. Bell , P.J. Bagge & S. Milosavljevic

a

a

Centre for Physiotherapy Research, School of Physiotherapy, University of Otago, Dunedin, New Zealand b

Department of Preventive and Social Medicine, University of Otago, Dunedin, New Zealand

c

Department of Community Medicine and Rehabilitation, Umea University, Sweden

Available online: 03 Sep 2009

To cite this article: P. Hendrick, M.L. Bell, P.J. Bagge & S. Milosavljevic (2009): Can accelerometry be used to discriminate levels of activity?, Ergonomics, 52:8, 1019-1025 To link to this article: http://dx.doi.org/10.1080/00140130902846464

PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.


Ergonomics Vol. 52, No. 8, August 2009, 1019–1025

Can accelerometry be used to discriminate levels of activity? P. Hendricka*, M.L. Bellb, P.J. Baggec and S. Milosavljevica

Downloaded by [University of Otago] at 17:31 27 October 2011

a

Centre for Physiotherapy Research, School of Physiotherapy, University of Otago, Dunedin, New Zealand; bDepartment of Preventive and Social Medicine, University of Otago, Dunedin, New Zealand; cDepartment of Community Medicine and Rehabilitation, Umea University, Sweden

The aims of this study were to investigate the associations between an activity logbook and the RT3 accelerometer and to assess whether the RT3 can discriminate activity levels in healthy adults. Ten participants completed two trials wearing an RT3 accelerometer over a 4–6 h period and completed a detailed activity log. Results showed a poor correlation between the RT3 in moderate activities (r ¼ 0.22) in comparison to low (r ¼ 0.52) and hard (r ¼ 0.70) from the logbook. A significant difference was found in average RT3 vector magnitude (VM) counts/min in each activity level (p 5 0.0001). Discriminant analysis demonstrated that an RT3VM counts/min value of approximately 500 was found to have high sensitivity (88%), and specificity (88%) for discriminating between low and moderate activity levels from the logbook. This study found that accelerometry has the potential to discriminate activity levels in free living. This study is the first to investigate whether tri-axial accelerometry can discriminate different levels of free-living activity recorded in an activity logbook. The RT3 accelerometer can discriminate between low and moderate physical activities and offers a methodology that may be applicable to future research in occupational settings. Keywords: physical activity; activity measurement; tri-axial accelerometry; short-term free living

1.

Introduction

There is a need for properly designed investigations employing objective measures to gather accurate and reliable data in the measurement of physical activity (PA) (Bulley et al. 2007). Various PA measurement tools have been employed in both the workplace and in free living to investigate the relationship between PA and musculoskeletal disorders (Schneider and Becker 2005, Schofield et al. 2005, Morken et al. 2007), to promote PA within the workplace (Thomas and Williams 2006) and to assess the effect of workplace PA programmes (Naito et al. 2008, Opdenacker et al. 2008). Accelerometers and recall questionnaires are two instruments that have been used to measure both free living (Booth 2000, Schutz et al. 2001) and workplace (Ruiz-Tendero et al. 2006) PA levels. Their simultaneous use is argued to provide an improved estimation of the types and intensities of PA (Keim et al. 2004, Macfarlane et al. 2006). The ability of a PA measurement to discriminate the intensity and duration of activity levels in free living is important for the assessment of differences within populations and to evaluate change over time (Lamonte and Ainsworth 2001). Tri-axial accelerometers (TAs) have demonstrated the most effective objective method to distinguish differences in activity levels between individuals, when compared to either *Corresponding author. Email: paul.hendrick@otago.ac.nz ISSN 0014-0139 print/ISSN 1366-5847 online Ó 2009 Taylor & Francis DOI: 10.1080/00140130902846464 http://www.informaworld.com

indirect calorimetry in the laboratory (Westerterp 1999) or doubly labelled water (DLW), accepted as the gold standard measure for energy expenditure (EE) in free living (Bouten et al. 1996). However, the use of DLW is costly, particularly in large populations, and correlations are often dependent upon subject characteristics (Plasqui and Westerterp 2007). Importantly, DLW cannot be used to examine patterns of PA or information about the frequency, intensity, type or duration of PA (Plasqui et al. 2005). Questionnaires are therefore utilised to provide a low cost, yet indirect, measure of the type, intensity and duration of such activities (Ken-Dror et al. 2005). However, correlations between measured tri-axial accelerometry and recall questionnaires have been found to vary considerably (r ¼ 0.30 to 0.90) within normal populations (Matthews and Freedson 1995, Leenders et al. 2000, Philippaerts et al. 2001, HaydenWade et al. 2003) and across different activity levels (Hendelman et al. 2000, Hayden-Wade et al. 2003). The RT3 TA has previously demonstrated high intra-monitor but variable inter-monitor reliability in the laboratory, dependent upon activity (Rowlands et al. 2004). The ability of the RT3 to discriminate activity levels has recently been evaluated in both children (Chu et al. 2007) and older adults (Sumukadas et al. 2008) under standardised conditions in the laboratory. Recent


Downloaded by [University of Otago] at 17:31 27 October 2011

1020

P. Hendrick et al.

research employing modelling of accelerometry data demonstrates the potential to differentiate individual activities in free living (Allen et al. 2006, Ermes et al. 2008). However, there appears to be no studies that have assessed and evaluated the ability of the RT3 or any other TA to discriminate activity levels from a recall instrument in a free-living population. As monitoring times for PA in free living are commonly between 3 to 7 d (Washburn and Ficker 1999), issues of validity and reliability have also been raised (Rikli 2000, Mader et al. 2006), particularly in relation to recall bias, which can affect reporting accuracy of the recorded and performed activities (Durante and Ainsworth 1996, Shephard 2003). Previous studies have utilised short-term, free-living activity monitoring (12 h or less) in both children (Tremblay et al. 2001) and an adult population (Hendelman et al. 2000) to assess the relationship between activity recall and accelerometry data. The current study has similarly employed short-term activity monitoring to: (1) assess whether there is a difference in RT3 vector magnitude (VM) counts/min compared to activities recorded and classified from the logbook into light, moderate and hard; (2) assess the levels of correlation between the RT3 data and the logbook for different activity levels; (3) assess the ability of the RT3 to discriminate freeliving activity levels recorded from the logbook.

2. 2.1.

Methods Participants

Following public advertising, five male and five female university students volunteered to participate. Inclusion criteria were: 1) in general good health with no history of a medical condition that could limit PA levels; 2) over the age of 18 years. Exclusion criteria were: 1) inability to remember to wear the RT3 daily and/or to record daily activities; 2) any history of current or past medical problems that prevented participants undertaking usual day-to-day activities; 3) inability to walk independently within the home and outside. All participants signed a consent form approved under Category B standards by the University of Otago, School of Physiotherapy Human Ethics Committee. 2.2.

three axes at 1-min intervals (mode 3). Information recorded in the activity logbook included current levels of free time PA over the 4–6 h period when the RT3 monitor was worn. Each participant also received instructions on the use of the RT3 and the activity log. Participants were asked to wear monitors on two occasions on weekdays, 1 week apart, for four to six consecutive hours between 09.00 and 22.00 hours. The rationale was to record the maximum amount and variety of free-living activities for each participant within the constraints of a 4–6 h monitoring period. The choice of two collection periods per participant was the methodology chosen to gather such data. They were instructed to record all activities of 5-min duration or greater in the activity log. The RT3 monitor was beltmounted over the right waist (belt-line) and used during all activities, excluding water-based activities (such as showering, surfing and swimming). Participants recorded any monitor removal in their logbooks and returned both monitors and logbooks at the end of the 4–6 h recording session. A careful perusal of the logbook data with the participant present reviewed the record of daily living activities undertaken during the recording period. Further clarification was sought where temporal activity records of 5 min or greater were missing. This step also ensured that the intensity and duration of described activities were also recorded. Participants were also asked for feedback on completing the logbook and if any problems had occurred while wearing the activity monitor. The RT3 data were downloaded and stored in a computer database. Activity monitoring was repeated the following week and all participants used the same monitor for both trials.

Protocol

Following provision of informed written consent, participants had their weight, height, age and sex recorded. Monitor input and data collection mode of the RT3 were set to simultaneously collect data in the

2.3.

Measurement procedures

2.3.1. Accelerometry The RT3 monitor is a small TA used to measure PA and is approximately the size of a personal pager (71 6 56 6 28 mm, weight 65 g, with one AAA sized battery). When firmly attached to clothing or a belt at the waist, it measures the acceleration of bodily movement in three directions, vertical (X), anterior– posterior (Y) and medio–lateral (Z), converting this movement into raw counts. VM is the square root of the sum of the squared counts in each direction and is utilised as a measure of accelerometry output representing PA. A full description of this technology is described in detail elsewhere (Chen and Bassett 2005). The RT3 monitor can store data for up to 7 d while simultaneously recording from all three axes at 1-min intervals. Proprietary interface software allows the download of data to an appropriate computer database. The software converts activity counts into


1021

Ergonomics

Downloaded by [University of Otago] at 17:31 27 October 2011

kcal/min from a physiological regression equation developed by the manufacturer, helping to calculate both activity related EE and total EE. 2.3.1.1. Accelerometry testing. An inter- and intramonitor reliability test conducted prior to the study involved each subject performing three standardised timed activities (running on a treadmill at 7.5 miles per hour (mph), walking at 3 mph and stair climbing) with each activity repeated 10 times while wearing six RT3 monitors. Results showed that the intra-monitor coefficient of variation (CV) (measurement error) ranged from 3% to 12%, whereas the inter-monitor CV ranged from 11% to 49%. The results showed that there was greater variability in the three older monitors (intra CV range 2.1–12%, inter CV range 2.1–21.6%) than the three newer monitors (intra CV range 3.1– 8.5%, inter CV range 1.6–10.7%). Therefore, the three new monitors were used in order to optimise reliability. 2.3.2. Activity logbook For this study, an activity logbook recorded PA during monitoring periods. The lack of a validated short-term activity recall questionnaire meant that a detailed activity log was developed and utilised as the primary measure of activity recall. The participant was required to record start and stop time, intensity (e.g. slow, brisk, fast) and details of any activities lasting greater than 5 min (e.g. walking uphill or carrying a bag) every 15 min in their activity logbook (Appendix 1). This was based upon the participant’s own perception of that activity, where moderate activity was always equated to the intensity of their ‘normal’ walking pace and hard to ‘running or cycling’. 2.3.2.1. Testing the activity logbook. The logbook was pilot tested on two participants over 4–6 h of free living and modified based upon their feedback. Based on this feedback this period of monitoring was accepted as allowing a more detailed assessment of each participant’s activities helping to minimise recall failure, optimise compliance and potentially capture a wide range of free-living daily activities. 2.4.

Data analysis

Logbook data were manually transferred into an ExcelTM database (Microsoft Corporation, Redmond, WA, USA) together with the EE of each activity calculated using metabolic equivalent (MET) values published in the Compendium of physical activities (Ainsworth et al. 2000). Activities not described in the compendium were recoded to a similar activity. EE (MET/kg per hour or kcal) for all activities were

computed by multiplying each individual’s weight (kg), duration of the activity (hours) and the MET value for that activity. Activities were deleted from the logbook if a corresponding RT3 record was not available. Also, any cycling activity performed would have been removed from comparative analysis with the RT3 as it is recognised that waist-worn accelerometry is not that accurate at estimating EE from cycling. No cycling was recorded by any of the participants. 2.4.1. Logbook reliability A MET coding crosscheck investigated the reliability of converting activities from the logbook into MET values. A blinded second assessor assigned MET values to three randomly selected logbooks with MET values demonstrating acceptable agreement if they were within 0.5 MET. A 93% level of agreement allowed activities to be MET coded and classified into low (1–2.9 MET), moderate (3–5.9 MET) and hard (46 MET) activity levels (Hendelman et al. 2000). 2.5.

Statistical methods

SAS version 9.1 (SAS Institute and Inc., Cary, NC, USA) was used for statistical analyses. Activity log data were graphically compared with RT3 VM scores to check for transcription errors and outliers. If errors were found, data were either recalculated or, in a very small number of cases, removed. In total, 19 activities from the logbooks were removed: seven activities due to removal of the RT3; 12 activities as they were shorter than 5 min in duration. Total and average RT3 VM counts/min and the sum of kcal derived from the RT3 monitor (RT3 kcal) was calculated for each recorded activity of 5-min duration or greater using Microsoft Excel. The logbook kcal value (logkcal) was calculated from the ascribed MET value of each recorded activity in the activity logbook (Ainsworth et al. 2000). This study dealt with the repeated measures (nonindependence) aspect of the data in two ways. For most of the analyses, the data were aggregated by finding the total kcal or VM count for each of the 10 participants, separately by trial (resulting in 20 data records) or by level (resulting in 24 records).These totals were then divided by the total number of minutes for each of the participants (by trial or level). Additionally, mixed models and generalised estimating equations (12) were used on the non-aggregated data. Correlation between the outcomes by trial and by level was investigated using Spearman’s correlation coefficient. The non-parametric correlation was chosen because of the small sample size. Mixed models (Fitzmaurice et al. 2004.) were used to estimate the mean RT3 VM per min, logkcal per


Downloaded by [University of Otago] at 17:31 27 October 2011

1022

P. Hendrick et al.

min and RT3 kcal per min as a function of activity level (low, medium, hard). The mixed model accounts for the repeated measures nature of the data as well as the imbalance (i.e. that participants had differing numbers of bouts of activity and at different levels). Level was used as a fixed effect and subject and trial were used as random effects. The researchers weighted by the number of minutes in each activity. Discrimination analyses were performed to estimate the area under the receiver operating characteristic (ROC) curve for RT3 VM per min and the logkcal per min in order to: 1) determine the ability of the RT3 to discriminate between low and moderate levels of activity (as defined by MET values); 2) determine the optimum cut point for the RT3 VM per min; 3) compare discrimination between the RT3 and the logbook. Hard activity was not used, as there were very little data in this category and the results would not have been reliable. Models for sensitivity and specificity were fit in the SAS procedure GENMOD, which accounts for non-independent observations. ROC curves were calculated from the output of these models. 3. 3.1.

Results Participant’s activity description

Participants (n ¼ 10) had a mean age of 24.5 (SD 4.9) years, height of 170.5 (SD 6.9) cm, weight of 70.5 (SD 5.6) kg and BMI of 24.6 (SD 2.7). In total, 185 activities of 5-min duration or greater were recorded over the two trial periods while the mean recording time for all 20 trials was 5.4 h. Duration of recorded activities ranged from 5 to 185 min, with a mean of 33 (SD 33.7) min. Of the 185 activities, 118 were classified as low activity, 59 were moderate and eight were hard activity. All participants engaged in low activity, nine engaged in moderate activity and five participants engaged in hard and very hard activities. Descriptive statistics for nonaggregated data are presented in Table 1. 3.2. RT3 scores within low, moderate and high activity levels The results of a mixed model ANOVA estimating the mean RT3 VM scores/min for each of the Table 1. Descriptive statistics for the non-aggregated data (n ¼ 185).

logkcal RT3 kcal RT3 VM counts/min

Mean

SD

Minimum

Maximum

3.03 2.99 644

2.57 2.43 800

1.08 0.50 3

22.94 20.52 5146

VM ¼ vector magnitude.

activity levels are shown in Table 2. Comparisons of these scores between each of the recorded activity levels show a statistically significant difference in scores between each of the recorded activity levels. 3.3. Correlations of RT3 scores with the logbook within low, moderate and high activity levels Correlations between the logbook and the RT3 (kcal and VM) within each of the activity levels are presented in Table 3. Low and high activity levels demonstrate moderate to good correlations (r ¼ 0.52, 0.90). Correlations within moderate activity were substantially lower for the logkcal to RT3 kcal (r ¼ 0.22) and RT3 VM counts/min score (r ¼ 0.48). 3.4. Discriminatory analysis of RT3 scores for activity level The discrimination analyses found the area under the ROC curve for RT3 VM counts/min to be 0.93 (95% CI 0.88, 0.97) and 0.96 (95% CI 0.93, 0.98) for logkcal. These areas were not found to be statistically different, p ¼ 0.2. The cut-off points, which maximised Table 2. Results of three mixed model ANOVA comparing average outcomes per min by activity level. Activity Level

n*

Low

118

Moderate Hard

59 8

RT3 VM counts/min (95% CI)

RT3 kcal/min (95% CI)

Logbook kcal/min (95% CI)

152 (85, 219) 1009 (865, 1154) 2442 (2175, 2709)

1.75 (1.41, 2.10) 3.80 (3.18, 4.43) 10.01 (8.79, 11.2)

1.83 (0.99, 2.67) 3.90 (2.94, 4.87) 10.85 (10.14, 13.77)

Note: Each of the pairwise tests of differences was statistically significant with p 5 0.0001. VM ¼ vector magnitude. n* ¼ number of recorded observations.

Table 3. Spearman’s correlation coefficients between logbook and RT3 per min accelerometer at low, medium and high activity levels where data is aggregated by participant and by activity level. Low (n ¼ 10) r value (95% CI)

Moderate (n ¼ 9) r value (95% CI)

Hard (n ¼ 5) r value (95% CI)

0.70 0.22 0.52 Logkcal – p ¼ 0.2 p ¼ 0.6 p ¼ 0.1 Total VM counts/ (70.16, 0.87) (70.53, 0.77) (70.48, 0.98) min 0.90 0.80 0.67 Logkcal – p ¼ 0.04 p ¼ 0.01 p ¼ 0.03 per min (0.09, 0.99) (0.29, 0.97) (0.07, 0.9) RT3kcals VM ¼ vector magnitude.


Ergonomics sensitivity and specificity, were found to be approximately 500 VM counts/min and 2.8 kcal/min. This value (500 VM counts/min) demonstrated a high sensitivity (0.88, 95% CI 0.77, 0.95) and specificity (0.88, 95% CI 0.82, 0.94) for discriminating between low and moderate activities.

Downloaded by [University of Otago] at 17:31 27 October 2011

4.

Discussion

This study found significant differences when comparing average RT3 VM counts/min scores with low, moderate and hard activities recorded from the activity logbook. Pearson correlations between the logbook and the RT3 demonstrate stronger associations at the low and high activity levels when compared to the moderate activity levels. Despite the variable correlation findings within each of the activity levels, RT3 activity counts demonstrated a high sensitivity, specificity and diagnostic accuracy for discriminating between low and moderate activity levels. 4.1.

Activity level analysis

Table 3 shows that the average RT3 VM counts/min score is able to distinguish between activities classified as light, moderate and hard, based upon their MET score derived from an activity logbook record. The relatively narrow width of the 95% CI for the VM counts/min scores within each activity level indicates an accurate estimation of RT3 scores within each of the activity levels in this free-living cohort. Previous research employing the TracmorTM tri-axial monitor established cut-off values in free living for accelerometry counts into light, moderate and hard activities (Westerterp 2001). To the authors’ knowledge, no previous cut-off values have been published for RT3 VM activity counts/min scores in a free-living population. The correlations between the logbook and the RT3 activity counts were moderate for low and hard activities and poor within the moderate activity level. The confidence intervals within all activity levels were wide due to the small number of participants, particularly seen in those undertaking hard activities (n ¼ 5). Previous studies have generally reported low to moderate correlations between tri-axial accelerometry and a recall instrument within the different activity levels (Hayden-Wade et al. 2003, Dubbert et al. 2004, Macfarlane et al. 2006). The current results may partly reflect the relatively small numbers of participants in the current study. Lower recording compliance or alternatively inaccuracies in recording activities and/or a reduced association with movement could explain the poorer correlation seen in moderate level activities. However, the results also demonstrate that RT3 activity counts/min have a high sensitivity and

1023

specificity for discriminating between low and moderate activity levels. This is a significant finding, particularly with reference to monitoring free-living activity and health recommendations (Blair 2003). The cut-off point of 500 counts/min is slightly higher than the findings from a recent laboratory-based study of 20 elderly and physically impaired participants (Sumukadas et al. 2008), which reported a cut-off point of 250 counts/min discriminated between sedentary and moderate activities. Chu et al. (2007), however, found that an RT3 counts/s threshold of 31 (equivalent of 1860 counts/min) had both a high sensitivity (84%) and specificity (72%) for distinguishing low from moderate activities in a group of young children. Both of these studies employed standardised tasks within a laboratory setting and the difference between these values is a reflection of not only the different populations but also the fact that the current study was performed in free living and employed a recall questionnaire as the comparative measure. It has also been shown that the counts generated by an accelerometer are dependent not only on the intensity but also the types of activities performed (Swartz et al. 2000). Cut-off values generated from standardised laboratory-based experiments may not fully reflect the full range and types of activities performed within the light and moderate activity categories in a free-living environment. Factors such as body type, range and types of activities performed and inter-instrument variability will result in significant between-subject variation and potential misclassification of activity. 4.2.

Advantages of short term activity monitoring

The advantages of the short-term recording period utilised in the current study are the enhanced compliance to wearing the monitors where no participant drop-out occurred in either of the two trials and no monitor data loss occurred, as previously reported in more prolonged monitoring (Philippaerts et al. 2001). It is also reasonable to argue that the detailed nature of the logbook optimised and improved accuracy and validity of the recorded activities. However, short-term activity monitoring has also been shown to alter activity behaviour (Clemes et al. 2008), suggesting that caution should be applied in accepting short-term activity measurement as an indicator of habitual free-living activity. 4.3.

Study limitations and suggestions for future work

It is recognised that use of a logbook is essentially a nonobjective measure of activity levels. The major deciding factors in the use of the logbook were low cost and ease of use for the participant, allowing a detailed and


Downloaded by [University of Otago] at 17:31 27 October 2011

1024

P. Hendrick et al.

descriptive record of the types of free-living activities to compare to the accelerometry data. Recording activity every 5 min over a relatively short time frame (4–6 h) was expected to enhance the relative accuracy of logbook data and reduce some of the inherent bias seen previously in recall instruments (Neilson et al. 2008). Future research should also compare accelerometry and recall instruments to other objective measures such as global positioning system tracking and/or physiological measures such as heart rate monitors (Bouchard and Trudeau 2008), DLW (Plasqui and Westerterp 2007) or indirect calorimetry (Kavouras et al. 2008). Such comparisons would allow a clearer understanding of the effectiveness of accelerometry to differentiate freeliving activity. It may be a combination of these measures that will ultimately offer the best options for supporting accelerometry as a discriminator of daily living activities (Rodriguez et al. 2002, 2005). The strength of the associations reported here requires validation in larger prospective trials. Participants in the present study were young students and therefore results may not be generalisable to all normal working populations. However, the authors consider that this study design could readily be applied to workplace investigations in order to help determine whether accelerometry can accurately discriminate given activity levels in different age groups and occupational activities. Establishment of cut-off values for different occupational groups would allow measurement of specific PA dimensions such as type, frequency, intensity and duration with an objective PA measure. This information would also enable a clearer interpretation of comparative accelerometry data between studies. In summary, this study showed that associations between an RT3 activity monitor and the logbook can be variable and dependent upon the activity levels of the participants. The use of provisional cut-off values and a sensitivity analysis for the RT3 VM activity counts, however, demonstrate an ability to discriminate low and moderate activity levels. Acknowledgements The authors acknowledge and thank Professor David Baxter for his support in this research and assistance in manuscript design. Funding was provided by Centre for Physiotherapy Research.

References Ainsworth, B.E., et al., 2000. Compendium of physical activities: An update of activity codes and MET intensities. Medicine Science in Sports and Exercise, 32, S498–S504. Allen, F.R., et al., 2006. Classification of a known sequence of motions and postures from accelerometry data using adapted Gaussian mixture models. Physiological Measurement, 27, 935–951.

Blair, S.N., 2003. Physical activity, epidemiology, public health, and the American College of Sports Medicine. Medicine Science in Sports and Exercise, 35, 1463. Booth, M., 2000. Assessment of physical activity: An international perspective. Research Quarterly in Exercise and Sport, 71, 114–120. Bouchard, D.R. and Trudeau, F., 2008. Estimation of energy expenditure in a work environment: Comparison of accelerometry and oxygen consumption/heart rate regression. Ergonomics, 51, 663–670. Bouten, C.V.C., et al., 1996. Daily physical activity assessment: Comparison between movement registration and doubly labeled water. Journal of Applied Physiology, 81, 1019–1026. Bulley, C., et al., 2007. A critical review of the validity of measuring stages of change in relation to exercise and moderate physical activity. Critical Public Health, 17, 17–30. Chen, K.Y. and Bassett, D.R. Jr., 2005. The technology of accelerometry-based activity monitors: Current and future. Medicine Science in Sports and Exercise, 37, S490–S500. Chu, E.Y.W., McManus, A.M., and Yu, C.C.W., 2007. Calibration of the RT3 accelerometer for ambulation and nonambulation in children. Medicine Science in Sports and Exercise, 39, 2085–2091. Clemes, S.A., Matchett, N., and Wane, S.L., 2008. Reactivity: An issue for short-term pedometer studies? British Journal of Sports Medicine, 42, 68–70. Dubbert, P.M., et al., 2004. Evaluation of the 7-day Physical Activity Recall in urban and rural men. Medicine Science in Sports and Exercise, 36, 1646–1654. Durante, R. and Ainsworth, B.E., 1996. The recall of physical activity: Using a cognitive model of the question-answering process. Medicine Science in Sports and Exercise, 28, 1282–1291. Ermes, M., et al., 2008. Detection of daily activities and sports with wearable sensors in controlled and uncontrolled conditions. IEEE Transactions on Information Technology in Biomedicine, 12, 20–26. Fitzmaurice, G.M., Laird, N.M., and Ware, J.H., 2004. Applied longitudinal analysis. Hoboken, NJ, USA: John Wiley and Sons, Inc. Hayden-Wade, H.A., et al., 2003. Validation of the telephone and in-person interview versions of the 7-day PAR. Medicine Science in Sports and Exercise, 35, 801–809. Hendelman, D., et al., 2000. Validity of accelerometry for the assessment of moderate intensity physical activity in the field. Medicine Science in Sports and Exercise, 32, S442– S449. Keim, N.L., Blanton, C.A., and Kretsch, M.J., 2004. America’s obesity epidemic: measuring physical activity to promote an active lifestyle. Journal of the American Dietetic Association, 104, 1398–1409. Ken-Dror, G., et al., 2005. Measurement and assessment of habitual physical activity in epidemiological studies. Harefuah, 1443, 200–205. Kavouras, S., et al., 2008. Assessment of energy expenditure in children using the RT3 accelerometer. Journal of Sports Sciences, 26, 959–966. Lamonte, M.J. and Ainsworth, B.E., 2001. Quantifying energy expenditure and physical activity in the context of dose response. Medicine & Science in Sports & Exercise, 33, S370–S378. Leenders, N.Y.J., Sherman, W.M., and Nagaraja, H.N., 2000. Comparisons of four methods of estimating physical activity in adult women. Medicine Science in Sports and Exercise, 32, 1320–1326.


1025

Downloaded by [University of Otago] at 17:31 27 October 2011

Ergonomics Macfarlane, D.J., et al., 2006. Convergent validity of six methods to assess physical activity in daily life. Journal of Applied Physiology, 101, 1328–1334. Mader, U., et al., 2006. Validity of four short physical activity questionnaires in middle-aged persons. Medicine Science in Sports and Exercise, 38, 1255–1266. Matthews, C.E. and Freedson, P.S., 1995. Field trial of a three-dimensional activity monitor: comparison with self report. Medicine Science in Sports and Exercise, 27, 1071– 1078. Morken, T., Magerøy, N., and Moen, B.E., 2007. Physical activity is associated with a low prevalence of musculoskeletal disorders in the Royal Norwegian Navy: A cross sectional study. BMC Musculoskeletal Disorders, 8, 56. Naito, M., et al., 2008. Effect of a 4-year workplacebased physical activity intervention program on the blood lipid profiles of participating employees: The highrisk and population strategy for occupational health promotion (HIPOP-OHP) study. Atherosclerosis, 197, 784–790. Neilson, H.K., et al., 2008. Estimating activity energy expenditure: How valid are physical activity questionnaires? American Journal of Clinical Nutrition, 87, 279–291. Opdenacker, J., et al., 2008. Effectiveness of a lifestyle physical activity intervention in a women’s organization. Journal of Women’s Health, 17, 413–421. Philippaerts, R.M., Westerterp, K.R., and Lefevre, J., 2001. Comparison of two questionnaires with a tri-axial accelerometer to assess physical activity patterns. International Journal of Sports Medicine, 22, 34–39. Plasqui, G. and Westerterp, K.R., 2007. Physical activity assessment with accelerometers: An evaluation against doubly labeled water. Obesity, 15, 2371–2379. Plasqui, G., et al., 2005. Measuring free-living energy expenditure and physical activity with triaxial accelerometry. Obesity Research, 13, 1363–1369. Rikli, R.E., 2000. Reliability, validity, and methodological issues in assessing physical activity in older adults. Research Quarterly in Exercise and Sport, 71, 89–96. Rodriguez, D.A., Brown, A.L., and Troped, P.J., 2005. Portable global positioning units to complement accelerometry-based physical activity monitors. Medicine Science in Sports and Exercise, 37, S572–S581. Rodriguez, G., et al., 2002. Comparison of the TriTrac-R3D accelerometer and a self-report activity diary with heartrate monitoring for the assessment of energy expenditure in children. British Journal of Nutrition, 87, 623–631. Rowlands, A.V., et al., 2004. Validation of the RT3 triaxial accelerometer for the assessment of physical activity. Medicine Science in Sports and Exercise, 36, 518–524. Ruiz-Tendero, G., et al., 2006. Measurement of physical activity levels of workers on a Spanish University campus using accelerometry technology. Journal of Human Movement Studies, 51, 321–335. Schneider, S. and Becker, S., 2005. Prevalence of physical activity among the working population and correlation with work-related factors: Results from the first German National Health Survey. Journal of Occupational Health, 47, 414–423. Schutz, Y., Weinsier, R.L., and Hunter, G.R., 2001. Assessment of free-living physical activity in humans: an overview of currently available and proposed new measures. Obesity Research, 9, 368–379. Shephard, R.J., 2003. Limits to the measurement of habitual physical activity by questionnaires. British Journal of Sports Medicine, 37, 197–206.

Sumukadas, D., Laidlaw, S., and Witham, M.D., 2008. Using the RT3 accelerometer to measure everyday activity in functionally impaired older people. Aging Clinical and Experimental Research, 20, 15–18. Swartz, A.M., et al., 2000. Estimation of energy expenditure using CSA accelerometers at hip and wrist sites. Medicine Science in Sports and Exercise, 32, S450–S456. Thomas, L. and Williams, M., 2006. Promoting physical activity in the workplace: Using pedometers to increase daily activity levels. Health Promotion Journal of Australia, 17, 97–102. Tremblay, M.S., Inman, J.W., and Williams, J.D., 2001. Peliminary evaluation of a video questionnaire to assess activity levels of children. Medicine Science in Sports and Exercise, 33, 2139–2144. Washburn, R.A. and Ficker, J.L., 1999. Physical Activity Scale for the Elderly (PASE): the relationship with activity measured by a portable accelerometer. Journal of Sports Medicine and Physical Fitness, 39, 336–340. Westerterp, K.R., 1999. Physical activity assessment with accelerometers. International Journal of Obesity, 23, S45– S49. Westerterp, K.R., 2001. Pattern and intensity of physical activity. Nature, 410, 539.

Appendix 1. Single Day Activity log-book This is an example page to show you how to fill out the Activity Log. (The RT3 started recording at 8.30am in this demonstration case) (1) Please remember to wear the RT3 during all activities. (2) Detail your activities in each quarter hour that you have worn the RT3. (3) Remember to remove the RT3 when engaging in water activities. Time Date

8.00 8.15 8.30 8.45 9.00

Detail your main activities of 5 minute duration or greater for each of the quarter hour periods when wearing the RT3. Please document if the RT3 is removed

am am am am am

9.15 am 9.30 am 9.45 am 10.00 am 10.30 am 10.45 am 11.00 11.15 11.30 11.45 12.00

am am am am pm

Walked to paper shop 12 min Bought paper and walked home 15 min Sat and read 10 min On telephone 5 min Light housework including dishes and cleaning 15 min Light housework including dishes and cleaning 15 min Gardening: mowing lawn on flat 15 min Gardening: mowing lawn on steep section 15 min Gardening: Pruning hedge 10 min Standing 5 min Gardening: Tidying up and putting rubbish in bags 10 min Rest 5 min Shower – removed RT3 20 min Sat at computer 15 min Sat at computer 15 min Prepared and ate lunch 30 min


can accelerometry predict activity levels