Exploring Actical Accelerometers as an Objective Measure of Physical Activity in People With Multiple Sclerosis

Article Outline

• Abstract
• Methods
  • Participants
  • Instruments
    • Actical accelerometers
  • Criterion Measures
    • Heart rate
    • Perceived exertion
    • Six-minute walk test
    • Thirty-second chair stand test
  • Procedures
  • Analyses
    • Feasibility and acceptability
    • Test-retest reliability and validity
• Results
  • Feasibility and Acceptability
  • Test-Retest Reliability
  • Validity
• Discussion
• Conclusions
• Acknowledgments
• Appendix 1: Activities Completed by Participants During the Accelerometer Testing Sessions
• References
• Copyright

Abstract 

Kayes NM, Schluter PJ, McPherson KM, Leete M, Mawston G, Taylor D. Exploring Actical accelerometers as an objective measure of physical activity in people with multiple sclerosis.

Objective

To assess the feasibility, acceptability, and psychometric properties of Actical accelerometers in people with multiple sclerosis (MS).

Design

Participants attended 2 testing sessions 7 days apart in which they completed 6 activities ranging in intensity while wearing an Actical accelerometer and Polar heart rate monitor. Perceived exertion was recorded after each activity.

Setting

University research center.

Participants

People (N=31) with a definite diagnosis of MS were purposefully selected, aiming for diversity in level of reported disability, age, sex, and type of MS.

Interventions

Not applicable.

Main Outcome Measures

Actical accelerometer, Polar S810i and RS800sd heart rate monitors, Borg rating of perceived exertion, six-minute walk test (6MWT), 30-second chair stand test.

Results

Accelerometers had good feasibility and acceptability in people with MS. Test-retest reliability was poor for sedentary and free-living activities, with low to moderate intraclass correlation coefficients (.00–.75), but was better for more vigorous or rhythmic activities (.85–.90). Bland-Altman 95% limits of agreement for average accelerometer counts were wide, ranging from ±16 (newspaper reading) to ±1330 (6MWT). Validity was not established with 95% prediction intervals showing high variability for all activities.

Conclusions

The psychometric problems highlighted here suggest Actical accelerometers should be used with caution in people with MS as a measure of physical activity, particularly when measuring comparatively sedentary or free-living activities.

Key Words: Activities of daily living, Exercise, Movement, Multiple sclerosis, Rehabilitation

List of Abbreviations: GEE, generalized estimating equation, %HRR, percentage heart rate reserve, ICC, intraclass correlation coefficient, IQR, interquartile range, MS, multiple sclerosis, PAR-Q, physical activity readiness questionnaire, RPE, rating of perceived exertion, 6MWD, six-minute walk distance, 6MWT, six-minute walk test

MULTIPLE SCLEROSIS is a demyelinating disease of the central nervous system and the most common chronic disabling neurologic condition in young adults. While evidence suggests that exercise and physical activity can result in a wide range of health benefits for people with MS,¹, ², ³, ⁴, ⁵, ⁶, ⁷, ⁸ participation remains low,⁹, ¹⁰, ¹¹, ¹², ¹³ leading to increased risk of secondary conditions¹⁴, ¹⁵, ¹⁶ and physical limitation, potentially contributing to the social isolation and reduced general well being reported in the MS population.¹⁷ Research seeking to understand exercise behavior better in people with MS has received increased attention,¹⁸, ¹⁹, ²⁰ but has been hampered by the lack of valid and reliable measures of physical activity available for use in this population.²¹, ²²

Recently, portable activity monitors such as pedometers and accelerometers have been growing in popularity as objective measures of physical activity, particularly in the general, healthy population.²³, ²⁴, ²⁵ However, the accuracy of these devices in disability populations requires consideration because of the complexity of measuring physical activity in such populations. For example, many people with MS engage in low levels of physical activity, have gait and ambulatory difficulties, rely on a range of assistive devices, have limitations in fine motor control and lack balance and coordination, all of which could affect the precision of activity monitors.

Pedometers have faced criticism²¹ because their accuracy deteriorates at slow walking speeds,²⁶, ²⁷ a characteristic of some people experiencing disability. Furthermore, pedometers measure only steps taken, thereby ignoring a range of free-living physical activities, such as gardening or washing, and other activities at the low end of the activity spectrum, which may be important to capture for people experiencing disability. Compared with pedometers, accelerometry-based devices such as the Actical accelerometers²⁸, ^a may offer important advantages because they have been shown to be sensitive in detecting varying levels of activity at the lower end of the physical activity spectrum¹²; detect movement in multiple planes²⁵; and are compact, waterproof, and can be worn in a variety of land-based and aquatic activities.²⁸ A range of accelerometers have been used previously in people with disability, including people with MS.¹², ²⁹, ³⁰, ³¹, ³² However, despite their wide use, the ability of accelerometers to detect the amount and intensity of activity reliably in the MS population has not been established. In addition, other important outcome measure criteria, such as feasibility and acceptability, are yet to be explored.

Feasibility is the impact that use of an instrument has on staff and researchers and requires consideration of issues such as time taken for staff training, ease of use, processing time, and data preparation,³³ whereas acceptability refers to how acceptable the instrument is from the perspective of the participant and is concerned with minimizing participant burden as much as possible.³³ Ideally, both criteria should be satisfied.

The specific objectives of this study were to (1) assess the feasibility and acceptability of the Actical accelerometer in people with MS, and (2) explore the test-retest reliability and validity of the Actical accelerometer in people with MS.

Back to Article Outline

Methods 

Ethical approval for this study was obtained from the Health and Disability Ethics Committee in New Zealand.

Participants 

Participants were recruited for this study through the local MS Societies and District Health Board in New Zealand as part of a larger ongoing study exploring the barriers and facilitators to physical activity for people with MS. All those who agreed to participate in the larger study were asked to indicate on their consent form whether they would also be willing to participate in this accelerometer testing study. Potential participants were eligible to take part if they had received a definite diagnosis of MS by a neurologist, were able to communicate with the researchers, and were independently mobile (with or without the use of an assistive device). Potential participants were excluded if they had any medical condition that precluded them from participation in any physical activity component of the study. Of those who gave consent, we aimed to recruit 30 participants, a sample size commonly used in this population and studies of this type,³⁰, ³², ³⁴ in which capturing a diverse sample is fundamental to making appropriate assessments of instrument reliability and acceptability.³⁵ Participants were purposefully selected, aiming for diversity in level of reported disability, age, sex, and type of MS. To do this, we initially stratified participants into low, medium, and high disability groups, based on a combination of self-reported disability items, such as the number of assistive devices and use of arms and legs (full, partial, or no use). From each of these strata, we then purposefully selected participants aiming for diversity in age, sex, and type of MS. Selected participants were then contacted via the telephone and screened for eligibility using a modified version of the PAR-Q.³⁶ If participants answered yes to any of the questions on the PAR-Q, then medical clearance was obtained from their general practitioner before taking part in the testing.

Instruments 

Actical accelerometers 

Actical accelerometers²⁸ detect low frequency (0.5–3.2Hz) gravitational forces (.05–2.0g) common to human movement and generate a signal, proportional to the magnitude and duration of the sensed acceleration, which is digitized and summed over a user-specified time-interval (epoch) length of 15, 30, or 60 seconds.³⁷ This digitized value is known as an activity count. Data can be displayed in activity counts, or Actical software can convert these counts to energy expenditure (kcal and metabolic equivalents [METS]). In the current study, 15-second epochs were selected and investigated because we were interested in testing the reliability and validity of the accelerometer in measuring short bouts of activity likely to be common in people with MS. Raw activity count data were used rather than energy expenditure data because there are currently no existing algorithms for predicting energy expenditure in people experiencing disability,³⁸ so the conversion of raw activity counts to energy expenditure may be problematic in people with MS. Furthermore, energy expenditure equations used by Actical have been found to underestimate time spent in vigorous activities and overestimate walking and sedentary activities.³⁹

Criterion Measures 

In the absence of a criterion standard measure of physical activity in people with MS, a range of proxy criterion measures were used to assess validity, including both physiologic (heart rate) and self-reported (RPE) activity intensity and actual observed physical activity on 2 standardized measures: the 6MWT⁴⁰ and the 30-second chair stand test.⁴¹

Heart rate 

R-R interval heart rate data were recorded simultaneously with accelerometer data via Polar S810i and RS800sd heart rate monitors.^b R-R interval heart rate data were converted to average heart rate (in beats a minute). Raw R-R interval data were filtered using the Polar Precision Performance Software, version 4.10.029.^b Filtered data were then exported to Microsoft Excel^c and converted to %HRR using the Karvonen formula⁴²:

where HR_activity is equal to the average beats a minute during activity completion and HR_max is equal to 220 minus the person's age. The Karvonen method has the advantage over predicted percentage heart rate maximum because it takes into account resting heart rate, which has been reported to be significantly lower in MS groups compared with age-matched healthy people.⁴³

Perceived exertion 

The Borg RPE⁴⁴ was used to measure self-reported activity intensity. The Borg RPE is a measure of subjective feeling of exertion during physical activity in which the person undertaking the activity rates activity intensity on a scale ranging from 6 (no exertion at all) to 20 (maximal exertion).

Six-minute walk test 

The 6MWT⁴⁰ is a measure of functional exercise capacity in which people are asked to walk as far as they can for 6 minutes. In the current study, the 6MWT was performed indoors in a 30-meter stretch of hallway on a hardened and flat surface. The 6MWD was used as the criterion measure. This is the total distance (m) covered by the participant during the 6-minute period. The protocol for administering this test and instructing participants followed that outlined by the American Thoracic Society.⁴⁰

Thirty-second chair stand test 

The 30-second chair stand test⁴¹ involves counting the number of times a person can stand fully upright from a seated position over a 30-second period without pushing off with their arms. The number of chair stands the participant completed over the 30-second period was used as the criterion measure. The protocol for administering this test followed that outlined by Rikli and Jones.⁴¹

Procedures 

All eligible participants were scheduled to attend 2 testing sessions, 7 days apart, at the Health and Rehabilitation Research Centre at Auckland University of Technology. On arrival at each testing session, participants were fitted with an Actical accelerometer and a Polar heart rate monitor. Accelerometers were mounted onto waistbands and fitted around the participants' waists over the iliac crest of the left hip, in accordance with the manufacturer's recommendations. The Polar heart rate transmitter was moistened and strapped to the chest. Participants were then instructed on the use of the Borg RPE, and resting heart rate was recorded.

During testing sessions, participants completed a series of 6 activities ranging in intensity from sedentary to vigorous (appendix 1) while wearing the Actical accelerometer and Polar heart rate monitor. Activities were conducted in random order for each participant at each testing session, with the exception of the 6MWT, which was always conducted last in order to eliminate the possibility that it would affect the participant's ability to complete subsequent activities. A set of standardized instructions was read to each participant before completing each activity.

At the commencement and completion of each activity, the marker button on the accelerometer and the start/finish button on the heart rate monitor were pressed simultaneously to enable comparison of data. On completion of each activity, participants were also asked to rate the activity intensity using the Borg RPE. Between activities, participants rested until their heart rate had returned to resting rate before commencing the next activity. On completion of the testing session, participants were asked to rate how comfortable they felt the accelerometer was to wear on a 5-point Likert scale (with 1 being “not comfortable at all” and 5 being “very comfortable”) and were to report any feedback they had about the accelerometer.

Raw data from the accelerometer and heart rate monitor were downloaded using the Actical manufacturer's software²⁸ and Polar Pro Trainer 5 software,^b respectively. Raw 15-second epoch activity data were then imported into Microsoft Excel and total activity count for each individual activity obtained. For most analyses, the total activity count was then converted into average activity counts a minute using the equation below in order to make it comparable to our primary criterion measure, %HRR, which was calculated using average heart rate (beats a minute) as described previously.

It was necessary to use this method rather than simply collect activity data using 60-second epochs for 3 reasons. First, some of the activities were less than 1 minute in length (eg, 30-second chair stand). Second, there was no way of ensuring that an activity started at the beginning of an epoch, resulting in the likelihood that many of the activities would start and finish midway through a 60-second epoch period. While this is also possible with 15-second epochs, there is less risk of the accelerometer picking up movement unrelated to the activity within that same epoch period with a shorter epoch. Last, because some activities resulted in partial epochs (eg, if an activity was 1 minute 7 seconds in length, it resulted in 4 full epochs and 1 partial epoch), then averaging total activity count over the number of epochs would have been misleading. In all of these instances, averaging total activity count across the total length of the activity and then converting this into average activity counts a minute allowed for higher accuracy.

Analyses 

Feasibility and acceptability 

Feasibility was assessed informally drawing on the experiences of the research team throughout the accelerometer testing period. Consideration was given to ease of use, staff training, processing time pretesting and posttesting, fitting of the accelerometer, and data preparation. Acceptability was assessed by calculating percentages of participant responses on the Likert scale and collating participant feedback.

Test-retest reliability and validity 

Using the methods described, raw accelerometer data were converted to average activity counts a minute for most analyses, with the exception of the validity analyses using actual observed activity as a criterion measure (6MWT and 30-second chair stand), where total activity count was used. Recognizing the likelihood for highly skewed empirical distributions, medians, IQRs, and ranges were employed to report estimates of location and spread. Test-retest reliability was assessed using the 1-way analysis of variance ICC and Bland-Altman 95% limits of agreement method.⁴⁵ Graphical checks advocated by Bland and Altman³⁵, ⁴⁵ were undertaken to detect whether important distributional violations existed. Because of the repeated nature of the data, validity was assessed using GEE models. Initially, scatter plots and lowess curves (a nonparametric regression estimator) were drawn to depict relationships graphically between accelerometer counts and chosen criterion measures (principally, %HRR, and the Borg RPE). GEE models were then fitted to estimate parameters and relationships between variables for each of the 6 activities. Unstructured within-person correlation matrices were specified and robust (Huber-White) sandwich estimators of variance employed to calculate SEs and confidence intervals. Diagnostic and residual checks of these analyses followed those recommended by Dupont.⁴⁶ All analyses were performed using SAS version 8.2^d and Stata version 8.0.^e

Back to Article Outline

Results 

From the 282 people involved in the larger study, 128 (45%) consented to take part in the accelerometer testing. Initially, 30 of these were purposefully selected, and all were deemed eligible to take part. Two potential participants failed to attend the testing sessions, so a further 3 participants were selected to account for those and the potential for future dropouts. However, there were no further dropouts, leaving a total of 31 participants. The sample had a median age of 50 years (range, 34–80y); 21 (68%) were women; and 29 (94%) were white, 1 (3%) was New Zealand Maori, and 1 (3%) was a Pacific Islander. The sample was diverse in terms of illness-related variables, with a median time since diagnosis of 7 years (range, 1–40y); 12 (39%) had chronic progressive MS, 11 (35%) had relapsing-remitting, 5 (16%) secondary progressive, and 3 (10%) a benign course of illness. Fourteen (45%) participants used an assistive device of some kind either inside or outside the home, predominantly using a cane, a walker, or both interchangeably.

Summary statistics of the average activity counts a minute, %HRR, and Borg RPE at each time point are shown in table 1. Available data for analysis fluctuate between activities because 2 participants did not attend their second testing session for reasons unrelated to testing (ie, schedule clashes and feeling unwell); some participants opted not to complete all activities, either at 1 or both time points, because either they or the researcher conducting the testing had concerns about their ability to complete these activities safely; and on a small number of occasions, the heart rate monitor failed to detect the participant's heart rate during testing. Average accelerometer counts were observed to have relatively large ranges compared with %HRR and the Borg RPE, particularly for the more vigorous activities such as the 6MWT (range, 1–6275), the 30-second chair stand (range, 34–6418), and the stair climb (range, 0–3472).

Table 1. Descriptive Statistics of Average Accelerometer Counts, Percentage Heart Rate Reserve and the Borg Rating of Perceived Exertion by Activity and Time Point for Participants Completing the Activity at Both Time Points

Feasibility and Acceptability 

The research team found the Actical accelerometers and the Actical software package very straightforward and easy to use. Staff training required was minimal with one 60-minute face-to-face session, after which staff worked independently, familiarizing themselves with the monitor and software, and ran through a mock testing session. The time spent setting up the accelerometers for use with participants and downloading the data was minimal, taking no more than 5 minutes a participant pretesting and posttesting. Fitting the accelerometer onto the participant was simple, taking no more than 1 minute to fit pretesting. Furthermore, the tamper-free design of the monitors made them particularly useful and appropriate for research, eliminating the potential for participant interference during data collection. However, there were some minor technical difficulties. First, the marker button did not always work on the first occasion, so the researcher had to press it twice each time to ensure it inserted a marker into the data set to indicate the start and finish time of each activity. Second, the custom interval function of the software that enables computation of summary data for a user-selected time period can only be used when working with energy expenditure data and so could not be applied to raw activity count data. Rather, the raw data had to be imported into Microsoft Excel to manage it more effectively.

Acceptability of the accelerometers was very high with participants, with most participants (90%) rating the accelerometer “very comfortable” (5/5) to wear. The remaining participants rated it moderately comfortable to wear, with 2 participants rating it 4 out of 5 and 1 rating it 3 out of 5. Participants commented that they “didn't notice it,” were “not even aware that it was there,” and “almost forgot it was there.” Even the participant who rated it as 3 out of 5 commented that they “didn't know it was there,” suggesting that the small, compact design of the accelerometers generated very little participant burden.

Test-Retest Reliability 

Results of the average accelerometer counts test-retest reliability analyses with both the ICCs and Bland-Altman 95% limits of agreement displayed for each activity are presented in table 2. ICCs ranged from .00 (newspaper reading) to .90 (6MWT), and the 95% limits of agreement were wide overall, ranging from ±16 (newspaper reading) to ±1330 (6MWT).

Table 2. Test-Retest Reliability of Average Accelerometer Counts by Activity for Participants Completing the Activity at Both Time Points

Abbreviation: CI, confidence interval.

⁎P value from the Student t test comparing the estimated bias against the null value of 0.

†While estimated, the distribution of ICC is highly skewed, and hence the reported estimates and CIs may be relatively unstable.

Validity 

Scatter plots of average accelerometer counts against %HRR and the Borg RPE for each activity are depicted in Fig 1, Fig 2. Examination of the scatter plots suggests that only weak relationships exist between average accelerometer counts and both %HRR and Borg RPE. In particular, sedentary (newspaper reading) and free living (washing) activities were seen to have relatively high variability in %HRR and Borg RPE measurements compared with the average accelerometer counts.

• 
  • View Large Image
  • Download to PowerPoint

• Fig 1. 

  Scatterplot of average accelerometer counts against %HRR by each activity.

• 
  • View Large Image
  • Download to PowerPoint

• Fig 2. 

  Scatterplot of average accelerometer counts against the Borg RPE by each activity.

Results from the GEE analysis together with the Bland-Altman 95% prediction intervals by each activity for all the criterion measures are presented in table 3. The parameter estimates indicate that a significant linear relationship exists between %HRR and the average accelerometer counts for the 6MWT, 30-second chair stand, and vacuuming activities. However, for the Borg RPE, only the average accelerometer counts obtained on the 6MWT and the stair climb activities had a significant relationship. For both criterion measures, the prediction intervals show high variability ranging from ±5.7 (newspaper reading) to ±19.7 (6MWT) for %HRR and ±3.6 (newspaper reading) to ±5.3 (washing) for Borg RPE.

Table 3. Parameter Estimates With Associated 95% Confidence Intervals and P Values of Generalized Estimating Equation Analysis Together With the Bland-Altman 95% Prediction Intervals for all Criterion Measures

Abbreviation: CI, confidence interval.

Significant linear relationships also exist between total accelerometer counts and the 6MWD and the number of chair stands on the 30-second chair stand test (see table 3). However, once again, the prediction intervals show high variability at ±195 (6MWD) and ±6.2 (number of chair stands).

Back to Article Outline

Discussion 

Accelerometry-based devices are increasingly used as an objective measure of physical activity in both research and practice. They are generally believed to be both accurate and reliable and in some cases have been used as a criterion measure for validating other measures of physical activity.⁴⁷, ⁴⁸ However, few studies have evaluated their use in people with chronic disabling conditions.²⁹, ³⁰, ³¹ This study has explored the use of the Actical accelerometer as an objective measure of physical activity in people with MS under standardized conditions. A range of activities were investigated, varying in both activity intensity (from sedentary to vigorous) and type of activity (from free-living to structured standardized activities).

Feasibility and acceptability of the Actical accelerometers was good, with the research team reporting them to be easy to use and implement, experiencing only minor technical difficulties, and participants finding them to be very comfortable to wear. However, test-retest reliability analyses indicated mixed findings. Low ICCs were obtained for sedentary (newspaper reading) and free-living activities (washing), while moderate to high ICCs were obtained on the more vigorous activities (6MWT) and rhythmic activities that required an alternate change in direction or change in acceleration (30-second chair stand). Bland-Altman 95% limits of agreement were more modest for all activities, showing high variability between time points.

Similarly, while validity analyses indicated a significant relationship between activity intensity and accelerometer counts on the more vigorous or rhythmic activities, such as the 6MWT, no such relationship existed on sedentary and free-living activities. The only exception was the vacuuming activity, considered a free-living activity, where accelerometer counts were significantly related to %HRR. It is possible that this is because vacuuming sometimes requires a relatively rapid change in hip position, and the accelerometer may be more sensitive to this type of movement. Despite this, the wide 95% prediction intervals obtained for all activities suggest that accelerometer counts do not predict either physiologic (%HRR) or self-reported (Borg RPE) activity intensity very well.

These findings suggest the validity of the Actical accelerometer for use in people with MS should be questioned; however, when interpreting these results, it is important to consider alternative reasons for the poor validity observed. Monitor placement may be 1 such reason. As one would expect, activity monitors are likely to measure motion most accurately when they are placed on the part of the body where the motion occurs. It is possible, then, that for free-living activities such as washing and vacuuming, hip placement might record fewer activity counts than wrist placement, for example. However, wrist placement is also subject to potential error because it is more likely to pick up motion unrelated to the task, such as fidgeting and muscle spasms. Furthermore, researchers exploring the ability of Actical accelerometers placed at different locations (wrist, hip, ankle) to predict energy expenditure in healthy persons over a range of free-living activities found hip location to be the only location that accurately predicted energy expenditure at all intensity levels (light, moderate, vigorous).³⁷ With this in mind, we would suggest that hip placement had little impact on the outcome of the validity analyses.

Another issue to consider is the possibility that activity intensity (measured by %HRR and Borg RPE) may be limited in usefulness as a criterion measure of physical activity. Heart rate monitors and ratings of perceived exertion have faced some criticism previously.²¹, ⁴⁹, ⁵⁰, ⁵¹ Heart rate can be influenced by a number of factors other than participation in physical activity, such as emotional stress and physical fitness,⁵¹ and has been found to overestimate light activity and underestimate moderate activity.⁴⁹ In addition, there are some factors to take into consideration specific to people with disability when using heart rate monitors, such as the impact that the disabling condition has on the participants' autonomic nervous system and the impact of certain medications on heart rate.²¹ However, as described, this study attempted to reduce the impact of these influences by using %HRR. Finally, Borg RPE has been reported to have high interindividual variability,⁵⁰ which could affect its validity as a criterion measure.

Despite the concerns noted about using heart rate or Borg RPE as criterion measures, we suggest a number of factors indicate the validity of the Actical accelerometer for use in people with MS must still be questioned. First, on the free-living activities (such as washing and vacuuming), the accelerometer counts obtained seemed markedly low and disproportionate to the nature of the activity and, indeed, the self-rating of the activity. Second, the 95% prediction intervals were wide for all activities, even for those activities for which significant relationships between the criterion measure and activity counts were observed. Furthermore, wide 95% prediction intervals were evident even when actual observed activity (6MWD and number of chair stands completed in the 30-second chair stand test) was used as a criterion measure rather than measures of activity intensity.

Back to Article Outline

Conclusions 

Despite the intuitive appeal of omnidirectional accelerometers such as the Actical, concerns about the quality of the data they provide have been identified. We suggest they should be used with caution in people with MS and possibly other people with chronic disabling conditions, particularly when intending to measure sedentary or free-living activities. This study has highlighted that the accuracy and reliability of accelerometry-based devices cannot be assumed across populations. Researchers and practitioners should ensure their device of choice has been well validated and is stable in their population of interest before using it in future research and practice.

Suppliers

Back to Article Outline

Acknowledgments 

We thank the local MS Societies and District Health Board for their support and help with recruitment. We acknowledge members of the Person Centred Rehabilitation Team at AUT University for their contribution to data collection.

Back to Article Outline

Appendix 1: Activities Completed by Participants During the Accelerometer Testing Sessions 

Back to Article Outline

References 

1. Taylor NF, Dodd KJ, Prasad D, Denisenko S. Progressive resistance exercise for people with multiple sclerosis. Disabil Rehabil. 2006;28:1119–1126
   • View In Article
   • MEDLINE
   • CrossRef
2. Stuifbergen AK, Blozis SA, Harrison TC, Becker HA. Exercise, functional limitations, and quality of life: a longitudinal study of persons with multiple sclerosis. Arch Phys Med Rehabil. 2006;87:935–943
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (126 KB)
   • CrossRef
3. Romberg A, Virtanen A, Ruutiainen J, et al. Effects of a 6-month exercise program on patients with multiple sclerosis—a randomized study. Neurology. 2004;63:2034–2038
   • View In Article
4. Roehrs TG, Karst GM. Effects of an aquatics exercise program on quality of life measures for individuals with progressive multiple sclerosis. J Neurol Phys Ther. 2004;28:63–71
   • View In Article
5. Rietberg MB, Brooks D, Uitdehaag BM, Kwakkel G. Exercise therapy for multiple sclerosis. Cochrane Database Syst Rev. 2005;(1):CD003980
   • View In Article
6. Petajan JH, Gappmaier E, White AT, Spencer MK, Mino L, Hicks RW. Impact of aerobic training on fitness and quality of life in multiple sclerosis. Ann Neurol. 1996;39:432–441
   • View In Article
   • MEDLINE
   • CrossRef
7. Oken BS, Kishiyama S, Zajdel D, et al. Randomized controlled trial of yoga and exercise in multiple sclerosis. Neurology. 2004;62:2058–2064
   • View In Article
8. Mostert S, Kesselring J. Effects of a short-term exercise training program on aerobic fitness, fatigue, health perception and activity level of subjects with multiple sclerosis. Mult Scler. 2002;8:161–168
   • View In Article
   • MEDLINE
   • CrossRef
9. Stuifbergen AK, Roberts GJ. Health promotion practices of women with multiple sclerosis. Arch Phys Med Rehabil. 1997;78(Suppl 5):S3–S9
   • View In Article
   • Abstract
   • Full-Text PDF (851 KB)
   • CrossRef
10. Stuifbergen AK. Physical activity and perceived health status in persons with multiple sclerosis. J Neurosci Nurs. 1997;29:238–243
    • View In Article
    • MEDLINE
    • CrossRef
11. Santiago MC, Coyle CP. Leisure-time physical activity among women with mobility impairments: implications for health promotion and leisure education. Ther Recreation J. 2004;38:188–205
    • View In Article
12. Ng AV, Kent-Braun JA. Quantitation of lower physical activity in persons with multiple sclerosis. Med Sci Sport Exerc. 1997;29:517–523
    • View In Article
    • MEDLINE
    • CrossRef
13. Motl RW, McAuley E, Snook EM. Physical activity and multiple sclerosis: a meta-analysis. Mult Scler. 2005;11:459–463
    • View In Article
    • MEDLINE
    • CrossRef
14. Slawta JN, Wilcox AR, McCubbin JA, Nalle DJ, Fox SD, Anderson G. Health behaviors, body composition, and coronary heart disease risk in women with multiple sclerosis. Arch Phys Med Rehabil. 2003;84:1823–1830
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (84 KB)
    • CrossRef
15. Slawta JN, McCubbin JA, Wilcox AR, Fox SD, Nalle DJ, Anderson G. Coronary heart disease risk between active and inactive women with multiple sclerosis. Med Sci Sport Exerc. 2002;34:905–912
    • View In Article
    • MEDLINE
    • CrossRef
16. Santiago MC, Coyle CP. Leisure-time physical activity and secondary conditions in women with physical disabilities. Disabil Rehabil. 2004;26:485–494
    • View In Article
    • MEDLINE
    • CrossRef
17. Sutherland G, Andersen MB. Exercise and multiple sclerosis: physiological, psychological, and quality of life issues. J Sports Med Phys Fit. 2001;41:421–432
    • View In Article
18. Stuifbergen AK, Rogers S. Health promotion: an essential component of rehabilitation for persons with chronic disabling conditions. Adv Nurs Sci. 1997;19:1–20
    • View In Article
19. Motl RW, Snook EM, McAuley E, Scott JA, Douglass ML. Correlates of physical activity among individuals with multiple sclerosis. Ann Behav Med. 2006;32:154–161
    • View In Article
    • MEDLINE
    • CrossRef
20. Motl RW, Snook EM, McAuley E, et al. Symptoms, self-efficacy, and physical activity among individuals with multiple sclerosis. Res Nurs Health. 2006;29:597–606
    • View In Article
    • MEDLINE
    • CrossRef
21. Warms C. Physical activity measurement in persons with chronic and disabling conditions: methods, strategies and issues. Fam Community Health. 2006;29(1 Suppl):78S–88S
    • View In Article
    • MEDLINE
22. Tudor-Locke CE, Myers AM. Challenges and opportunities for measuring physical activity in sedentary adults. Sports Med. 2001;31:91–100
    • View In Article
    • MEDLINE
    • CrossRef
23. Ward DS, Evenson KR, Vaughn A, Rodgers AB, Troiano RP. Accelerometer use in physical activity: best practices and research recommendations. Med Sci Sport Exerc. 2005;37(11 Suppl):S582–S588
    • View In Article
    • MEDLINE
    • CrossRef
24. Trost SG, McIver KL, Pate RR. Conducting accelerometer-based activity assessments in field-based research. Med Sci Sport Exerc. 2005;37(11 Suppl):S531–S543
    • View In Article
    • MEDLINE
    • CrossRef
25. Chen KY, Bassett DR. The technology of accelerometry-based activity monitors: current and future. Med Sci Sport Exerc. 2005;37(11 Suppl):S490–S500
    • View In Article
    • MEDLINE
    • CrossRef
26. Motl RW, McAuley E, Snook EM, Scott JA. Accuracy of two electronic pedometers for measuring steps taken under controlled conditions among ambulatory individuals with multiple sclerosis. Mult Scler. 2005;11:343–345
    • View In Article
    • MEDLINE
    • CrossRef
27. Crouter SE, Schneider PL, Karabulut M, Bassett DR. Validity of 10 electronic pedometers for measuring steps, distance, and energy cost. Med Sci Sport Exerc. 2003;35:1455–1460
    • View In Article
    • MEDLINE
    • CrossRef
28. Mini Mitter. Actical: physical activity monitoring system—instruction manual software version 2.0. Bend: Mini Mitter Co Inc; 2003;
    • View In Article
29. Motl RW, Zhu W, Park Y, McAuley E, Scott JA, Snook EM. Reliability of scores from physical activity monitors in adults with multiple sclerosis. Adapt Phys Act Q. 2007;24:245–253
    • View In Article
30. Motl RW, McAuley E, Snook EM, et al. Validity of physical activity measures in ambulatory individuals with multiple sclerosis. Disabil Rehabil. 2006;28:1151–1156
    • View In Article
    • MEDLINE
    • CrossRef
31. Gosney JL, Scott JA, Snook EM, Motl RW. Physical activity and multiple sclerosis: validity of self-report and objective measures. Fam Community Health. 2007;30:144–150
    • View In Article
32. Kayes NM, McPherson KM, Taylor D, Schluter PJ, Wilson BK, Kolt G. The Physical Activity and Disability Survey (PADS): reliability, validity and acceptability in people with multiple sclerosis. Clin Rehabil. 2007;21:628–639
    • View In Article
    • CrossRef
33. Fitzpatrick R, Davey C, Buxton MJ, Jones DR. Evaluating patient-based outcome measures for use in clinical trials. Health Technol Assess. 1998;2:1–71
    • View In Article
    • MEDLINE
34. Verheyden G, Nuyens G, Nieuwboer A, Van Asch P, Ketelaer P, De Weerdt W. Reliability and validity of trunk assessment for people with multiple sclerosis. Phys Ther. 2006;86:66–76
    • View In Article
    • MEDLINE
35. Bland JM, Altman DG. Applying the right statistics: analyses of measurement studies. Ultrasound Obst Gyn. 2003;22:85–93
    • View In Article
36. Thomas S, Reading J, Shephard RJ. Revision of the Physical Activity Readiness Questionnaire (PAR-Q). Can J Sports Sci. 1992;17:338–345
    • View In Article
    • MEDLINE
37. Heil DP. Predicting activity energy expenditure using the Actical activity monitor. Res Q Exercise Sport. 2006;77:64–80
    • View In Article
38. Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sport Exerc. 2000;32(9 Suppl):S498–S516
    • View In Article
    • MEDLINE
    • CrossRef
39. Crouter SE, Churilla JR, Bassett DR. Estimating energy expenditure using accelerometers. Eur J Appl Physiol. 2006;98:601–612
    • View In Article
    • MEDLINE
    • CrossRef
40. American Thoracic Society. ATS statement: guidelines for the six-minute walk test. Am J Respir Crit Care. 2002;166:111–117
    • View In Article
41. Rikli RE, Jones CJ. Development and validation of a functional fitness test for community-residing older adults. J Aging Phys Activ. 1999;7:121–161
    • View In Article
42. Karvonen J, Kertala K, Mustala O. The effects of training heart rate: a longitudinal study. Ann Med Exp Biol Fenn. 1957;35:307–315
    • View In Article
    • MEDLINE
43. Savci S, Inal-Ince D, Arikan H, et al. Six-minute walk distance as a measure of functional exercise capacity in multiple sclerosis. Disabil Rehabil. 2005;27:1365–1371
    • View In Article
    • MEDLINE
    • CrossRef
44. Borg GA. Perceived exertion: a note on “history” and methods. Med Sci Sport Exerc. 1973;5:90–93
    • View In Article
45. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307–310
    • View In Article
    • CrossRef
46. Dupont W. Statistical modelling for biomedical researchers: a simple introduction to the analysis of complex data. Cambridge: Cambridge Univ Pr; 2002;
    • View In Article
47. Orrell A, Doherty P, Coulton S, Miles J, Stamatakis E, Lewin R. Failure to validate the Health Survey for England physical activity module in a cardiac population. Health Policy. 2007;84:262–268
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (119 KB)
    • CrossRef
48. Yore MM, Ham SA, Ainsworth BE, et al. Reliability and validity of the instrument used in BRFSS to assess physical activity. Med Sci Sport Exerc. 2007;39:1267–1274
    • View In Article
    • CrossRef
49. Macfarlane DJ, Lee CC, Ho EY, Chan KL, Chan D. Convergent validity of six methods to assess physical activity in daily life. J Appl Physiol. 2006;101:1328–1334
    • View In Article
    • MEDLINE
    • CrossRef
50. Whaley MH, Brubaker PH, Kaminsky LA, Miller CR. Validity of rating of perceived exertion during graded exercise testing in apparently healthy adults and cardiac patients. J Cardiopulm Rehabil. 1997;17:261–267
    • View In Article
    • MEDLINE
    • CrossRef
51. Freedson PS, Miller K. Objective monitoring of physical activity using motion sensors and heart rate. Res Q Exercise Sport. 2000;71(2 Suppl):S21–S29
    • View In Article

• a Actical accelerometers; Mini Mitter, 20300 Empire Blvd # B3, Bend, OR 97701.
• b Polar Precision Performance Software, version 4.10.029; Polar Electro Oy, HQ Professorintie 5, FIN-90440 Kempele, Finland.
• c Microsoft Excel; Microsoft, One Microsoft Way, Redmond, WA 98052.
• d SAS version 8.2; SAS Institute, 100 SAS Campus Dr, Cary, NC 27513.
• e Stata version 8.0; StataCorp 4905 Lakeway Dr, College Station, TX 77845.