ORIGINAL ARTICLE
`Measuring Free-Living Physical Activity in Adults With and
`Without Neurologic Dysfunction With a Triaxial Accelerometer
`Leigh A. Hale, PhD, Jaya Pal, MPhty, Ines Becker, PGDipPhyt
`
`1765
`
`ABSTRACT. Hale LA, Pal J, Becker I. Measuring free-
`living physical activity in adults with and without neurologic
`dysfunction with a triaxial accelerometer. Arch Phys Med
`Rehabil 2008;89:1765-71.
`Objective: To investigate the reliability, validity, and utility
`of a triaxial accelerometer to measure physical activity in the
`free-living environment in adults with and without neurologic
`dysfunction.
`Design: Repeated-measures design.
`Setting: General community.
`Participants: Volunteer sample of 17 men and 30 women
`(age range, 28⫺91y) living in the community with stroke of
`greater than 6 months in duration (n⫽20), Parkinson disease
`(n⫽7), or multiple sclerosis (n⫽11), and healthy but sedentary
`controls (n⫽9).
`Interventions: Not applicable.
`Main Outcome Measures: Physical activity measured with
`the TriTrac RT3 accelerometer, 7-day recall questionnaire, and
`activity diary.
`Results: The accelerometer reliably measured free-living
`physical activity (intraclass correlation coefficient, .85; 95%
`confidence interval, .74⫺.91; P⫽.000). The standard error of
`measurement indicated that a second test would differ from a
`baseline test by ⫾23%. Mean daily RT3 data collected in the
`first 3 days differed significantly from that of the mean daily
`RT3 data collected over 7 days. The RT3 appeared to distin-
`guish level of mobility better than the 7-day recall question-
`naire, and participants found the RT3 to be a user-friendly and
`acceptable measure of physical activity.
`Conclusions: The triaxial accelerometer provided a stable
`measure of free-living physical activity, was found to distin-
`guish between people with varying levels of mobility, and was
`well tolerated by participants. The results indicate that collect-
`ing data for 3 days was not reflective of data collected over 7
`days.
`Key Words: Exercise; Neurologic manifestation; Question-
`naires; Rehabilitation.
`© 2008 by the American Congress of Rehabilitation Medi-
`cine and the American Academy of Physical Medicine and
`Rehabilitation
`
`INCREASING PHYSICAL ACTIVITY
`
`important
`is an
`health goal for both people with and without disability,1-3
`necessitating an accurate method of measuring daily physical
`activity. Physical activity questionnaires and diaries are com-
`monly used but rely on recall and honest reporting and require
`people to have no cognitive deficits and no potential for bias in
`reporting results.4-6 Motion sensors, such as pedometers and
`accelerometers, provide an objective method of measuring
`physical activity. Pedometers are simple to use and inexpensive
`but may be less accurate at slow speeds of walking.7 Uniaxial
`and triaxial accelerometers measure the acceleration of move-
`ment and can quantify movement intensity, frequency, and
`duration.8 Triaxial accelerometers capture movement in 3 or-
`thogonal planes, potentially providing a comprehensive mea-
`surement of the variety of movements performed by people in
`their day-to-day life. However, the increased sensitivity of
`3-dimensional measurement may reduce the reliability of data
`on repeated measurements; uniaxial accelerometers’ 1-direc-
`tional capture of movement may provide more stable data.8
`The TriTrac RT3 accelerometera is a triaxial accelerometer
`that may be suitable for sustained tracking of physical activity
`in the home environment. It is small (65g), capable of collect-
`ing and storing data in 1-minute epochs for 21 days, and has no
`external controls that can be manipulated during data collec-
`tion.9 To date the attributes of the RT3 have been investigated
`in the laboratory with mechanical devices,8,9 treadmill walk-
`ing,10-13 and discrete physical tasks.10,11,14 Populations tested
`have included healthy adults,10,11,13 children,11,12 and adults
`with MS.14 Most studies reported good intramonitor reliability;
`however intermonitor variance has been demonstrated, indicat-
`ing that the same monitor should be used for the same partic-
`ipant in a repeated-measures design.8-13 One study has reported
`on the use of the RT3 outside of a laboratory, and in this study,
`the reliability of the RT3 to measure activity during a physical
`education program in school children with visual impairment
`was reported to be good.15 To our knowledge, no study has
`investigated the attributes of the RT3 in measuring daily phys-
`ical activity in the free-living environment.
`The purpose of this study was to investigate the reliability,
`validity, and utility of the RT3 to measure physical activity in
`the free-living environment in adults with and without neuro-
`logic dysfunction. More specifically, we wished to explore the
`test-retest reliability and sensitivity of the RT3 in free-living
`compared with that of the 7-day recall questionnaire, and
`whether it was necessary to measure activity for 7 days as has
`
`From the REAL Neurology Research Group, Centre for Physiotherapy Research
`and School of Physiotherapy, University of Otago, Dunedin, New Zealand.
`Preliminary results presented to the Neurosymposium, Neurological Special Inter-
`est Group of the New Zealand Physiotherapy Society, May 12, 2007, Nelson, New
`Zealand.
`Supported by the University of Otago (research grant).
`No commercial party having a direct financial interest in the results of the research
`supporting this article has or will confer a benefit on the authors or on any organi-
`zation with which the authors are associated.
`Reprint requests to Leigh A. Hale, PhD, Centre for Physiotherapy Research, University
`of Otago, PO Box 56, Dunedin, New Zealand, e-mail: leigh.hale@otago.ac.nz.
`0003-9993/08/8909-00997$34.00/0
`doi:10.1016/j.apmr.2008.02.027
`
`List of Abbreviations
`
`confidence interval
`intraclass correlation coefficient
`multiple sclerosis
`mean vector magnitude
`Parkinson disease
`Rivermead Mobility Index
`receiver operating characteristic
`
`CI
`ICC
`MS
`MVM
`PD
`RMI
`ROC
`
`Arch Phys Med Rehabil Vol 89, September 2008
`
`Exhibit 1042 page 1 of 7
`DENTAL IMAGING
`
`

`

`1766
`
`MEASURING FREE-LIVING PHYSICAL ACTIVITY, Hale
`
`been the case in many previous studies measuring activity in
`free-living.16
`
`METHODS
`
`Sampling
`Volunteers with PD, MS, or stroke, and sedentary, healthy
`participants were recruited via local service organizations and
`public advertising. Sample size was calculated by the method
`described by Bonett,17 using data obtained from a pilot study
`investigating the reliability of the RT3 to measure walking in
`people with MS.14 To obtain an ICC with a 95% CI of width
`0.2 using 2 repeated measurements, a sample size of 53 was
`calculated.
`Inclusion criteria. Participants had to be of good health,
`living in the community, and able to walk independently within
`the home with or without appliances. Participants with neuro-
`logic dysfunction had to have a definite diagnosis of PD,18
`MS,19 or stroke20 of more than 6 months. Adult control par-
`ticipants were recruited if they self-reported to be sedentary.
`Exclusion criteria included the inability to understand the re-
`quirements of the study (eg, because of dementia or receptive
`aphasia) and the presence of short-term memory loss. Written
`informed consent was gained from all participants. The study
`was approved by the local regional ethical committee (no.
`LRS/05/09/029).
`
`Equipment
`The RT3 is battery operated and uses an integrated computer
`chip to measure movement across 3 orthogonal planes: vertical
`(x), anteroposterior (y), and mediolateral (z). The RT3 mea-
`sures the mean acceleration (in m/s2) for each of the 3 planes
`across set 1-second or 1-minute intervals and presents these
`data in a digital format called activity counts. The exact rela-
`tionship between the acceleration data and the displayed activ-
`ity count has not been described by the manufacturers. Activity
`counts for each plane can be summarized by calculating the
`MVM (⫽ [x2 ⫹ y2 ⫹ z2]0.5), which is also expressed in activity
`units (http://www.stayhealthy.com). Because it is not possible
`to calibrate an RT3 unit, the reliability of each RT3 unit to
`measure motion per se was established in the laboratory prior
`to the start of the study with the use of repeated measurements
`both on a mechanical device and with discrete, standardized
`motor tasks. Six new monitors were tested and found to be
`reliable; no monitor had to be excluded.
`Questionnaires used in this study included the RMI21 and the
`7-day recall questionnaire,16,22 a validated, interviewer-admin-
`istered questionnaire that asks respondents to recall activities
`they have performed over the past 7 days. It has been used in
`previous studies to investigate activity in adults with MS.23,24
`Participants were asked to keep a daily activity log in which
`they recorded, for each hour of the day, the activities they had
`been involved in, such as shopping or walking. This informa-
`tion was used to verify the collected RT3 data. A utility
`questionnaire was specifically developed for this study to eval-
`uate participants’ opinions of using the RT3.
`
`Procedure
`The following measurements were taken or recorded:
`weight, height, age, sex, and level of mobility (using the
`RMI21). The RT3 was programmed via computer interface with
`the participant’s personal data (sex, age, height, weight) prior
`to testing and set to sample data for all 3 axes every minute.
`The participant was instructed when to wear the RT3 unit and
`how to complete the daily activity log. For participants unable
`
`Arch Phys Med Rehabil Vol 89, September 2008
`
`to complete the daily diary, we made arrangements for another
`person to complete it under the participant’s instruction. The
`RT3 was attached to the participant’s waist belt in a central
`back position and switched on to start measuring and recording
`activity data; the time of activation was recorded. The central
`back position was chosen to locate the RT3 to be as close to the
`body’s center of gravity as possible25 and to allow for potential
`asymmetrical movement as a result of the neurologic condi-
`tion.14 Participants were asked to wear the RT3 during waking
`hours (except when bathing, swimming, or lying in bed) for 7
`consecutive days while maintaining their typical weekly sched-
`ules. Seven days later, the RT3 unit was collected from the
`participant, the time this occurred was recorded, and the data
`were downloaded via the computer software. At this point, the
`7-day recall questionnaire was administered. Eight weeks later
`the procedure was repeated, starting on the same day and time
`of the week (a Monday, Tuesday, or Wednesday) and using the
`same RT3 unit. Participants then completed the utility ques-
`tionnaire. Participants were telephoned during the test week to
`ensure there were no problems and reminded to wear the RT3
`and to complete the daily activity log.
`
`Data Analysis
`For each participant, the MVM from activity data (in activity
`units [AU]) for each 24-hour period, beginning at the time the
`RT3 was activated, was summed to provide a daily activity
`count. This daily activity data and the recorded data in the
`activity log were compared for obvious inconsistencies (eg,
`failure to wear the RT3, equipment failure). Data considered
`erroneous were not included in the statistical analysis. The
`mean daily data for the first 3 days and for 7 days of measuring
`were calculated. The 7-day recall questionnaire and RMI data
`were scored. The 7-day recall questionnaire scores were con-
`verted to kilocalories as described by Sallis et al.22
`All statistical calculations were performed using the SPSS
`softwareb for Windows with the level of significance set at P less
`than .05. All demographic and measured data were analyzed
`descriptively.
`RT3 data were analyzed for each of the 2 test periods, for
`both the 3-day period data (mean daily MVM over 3 days) and
`the 7-day period data (mean daily MVM over 7 days), with
`intraclass coefficients (ICC2,1) using a 2-way random effects
`model with absolute agreement, and with the SE of measure-
`ment.26 The ICC and SE of measurement were calculated as
`follows:
`
`ICC ⫽
`
`between-subject variance
`between-subject variance ⫹ within-subject variance
`
`SE of measurement
`⫽ square root of the within-subject variance
`The strength of all correlations computed were determined
`as follows: 0.00 to 0.25, little or no correlation; 0.26 to 0.49,
`low correlation; 0.50 to 0.69, moderate correlation; 0.70 to
`0.89, high correlation; and 0.90 to 1.00, very high correlation.27
`To investigate whether daily activity data over 3 week days
`differed significantly from daily activity data measured over 7
`days, the mean daily MVMs for each period were compared
`using a paired t test and its 95% CI for the difference. If the
`95% CI lay below the smallest meaningful difference between
`the 2 measurements, they were determined to be equivalent.
`The level of agreement between the 2 scores was established
`with the Bland-Altman method.28
`
`Exhibit 1042 page 2 of 7
`DENTAL IMAGING
`
`

`

`MEASURING FREE-LIVING PHYSICAL ACTIVITY, Hale
`
`1767
`
`Characteristics
`
`Total (N⫽47)
`
`MS (n⫽11)
`
`PD (n⫽7)
`
`Stroke (n⫽20)
`
`Controls (n⫽9)
`
`Table 1: Demographic Characteristics of Participants
`
`Sex
`Men
`Women
`Age (y)
`
`RMI (total/15)
`
`Height (cm)
`
`Weight (kg)
`
`17
`30
`63.7⫾15.5
`(28–91)
`12.5⫾2.6
`(4–15)
`165.4⫾8.0
`(147–182)
`76.7⫾15.9
`(46–124)
`
`3
`8
`50.7⫾11.8
`(35–70)
`12.4⫾2.2
`(8–15)
`164.9⫾9.5
`(155–182)
`69.2⫾14.2
`(49–85)
`
`3
`4
`75.3⫾7.7
`(68–91)
`11.7⫾2.6
`(7–15)
`160.6⫾8.5
`(147–180)
`79.4⫾6.4
`(73–88)
`
`10
`10
`72⫾7.1
`(57–86)
`11.9⫾2.9
`(4–15)
`166.2⫾6.8
`(150–178)
`83.2⫾18.5
`(46–124)
`
`1
`8
`51⫾18.1
`(28–76)
`14.9⫾0.3
`(15–15)
`167.8⫾7.4
`(152–179)
`70.1⫾11.1
`(58–96)
`
`NOTE. Values are n or mean ⫾ SD (range).
`
`The relationships of MVM and 7-day recall questionnaire
`data to level of mobility (measured by the RMI) were investi-
`gated using scatterplots, linear regression (R2), and ROC anal-
`yses curves. The area under the ROC curve was calculated
`under the nonparametric assumption. The closer this area was
`to 1.0, the more accurately the activity data could be deemed to
`distinguish between levels of mobility. The hypotheses were
`that
`the greater the loss of mobility,
`the less activity the
`participant would perform, and that the accelerometer would be
`able to detect this better than the 7-day recall questionnaire.
`
`RESULTS
`Fifty-two participants were recruited into the study, but data
`were complete for only 47 (age, 64⫾15y; PD, n⫽7; MS,
`n⫽11; stroke, n⫽20; controls, n⫽9). Table 1 reflects the de-
`mographic characteristics of these participants; data were not
`normally distributed. Five participants did not complete testing
`because of monitor fault (n⫽3), development of an acute
`medical condition (n⫽1), and death (n⫽1).
`The average number of hours of collected daily activity was
`11 hours. No participants included in the analysis had less than
`10 hours of collected RT3 data a day. Table 2 shows the MVM
`and 7-day recall questionnaire data collected for each of the test
`periods; these data were not normally distributed. Data consid-
`ered to be outliers can be seen in the box and whisker graphs
`
`in figure 1, representing the distribution of the activity data
`(MVM) by diagnosis for each of the test periods. Inspection of
`the daily activity logs revealed that these outlying data were a
`result of the type of activity participants had engaged in, which
`varied considerably compared with other participants within
`the same diagnostic group.
`Table 3 displays the ICCs and SE of measurement calculated
`for the RT3 MVM data collected over the 2 test periods as well
`as those collected in the first 3 days of each test period. The
`ICCs for the total group and for the diagnostic subgroups
`demonstrated high to very high correlation with the exception
`of the 3 day MS and stroke scores, which were only moderately
`correlated. Inspection of the box and whisker plots for the
`stroke subgroup indicated disparate data for 2 participants, and
`this was explained and verified by the varying degrees of
`weekly physical activity recorded in their daily activity logs. In
`table 3, the SE of measurement is presented as a percentage of
`the mean data for each parameter collected in the first week of
`data collection. The absolute reliability (SE of measurement,
`expressed as percentage) for the 7-day total data was 23%. The
`95% CI for the total data had a width of .17, meeting the a
`priori power calculation of the study. The percentage SE of
`measurement for the 3-day data was larger at 27% and the 95%
`CI width was slightly greater (.21) than the a priori power
`calculation.
`
`Table 2: The Group Means Physical Activity Data Collected Over 7 Days or 3 Days for Each Participant Group
`
`Test Period
`
`Total (N⫽47)
`
`MS (n⫽11)
`
`PD (n⫽7)
`
`Stroke (n⫽20)
`
`Controls (n⫽9)
`
`7 days
`Test period 1
`MVM (AU)
`Test period 2
`MVM (AU)
`7 days
`Test period 1
`7-d RQ (kcal)
`Test period 2
`7-d RQ (kcal)
`3 days
`Test period 1
`MVM (AU)
`Test period 2
`MVM (AU)
`
`894,236⫾534,844
`
`1,085,849⫾373,047
`
`854,660⫾433,264
`
`673,920⫾379,495
`
`1,385,760⫾719,868
`
`852,528⫾510,239
`
`965,707⫾398,308
`
`754,492⫾470,446
`
`573,403⫾266,993
`
`1,490,363⫾510,675
`
`2468⫾496
`
`2442⫾514
`
`2180⫾454
`
`2166⫾486
`
`2484⫾273
`
`2430⫾228
`
`3645⫾572
`
`2633⫾607
`
`2413⫾354
`
`2363⫾327
`
`412,990⫾283,841
`
`520,822⫾210,673
`
`470,821⫾383,874
`
`320,463⫾190,549
`
`625,401⫾323,440
`
`379,142⫾271,720
`
`430,842⫾174,992
`
`408,617⫾460,806
`
`250,269⫾116,847
`
`611,578⫾291,424
`
`NOTE. Values are mean ⫾ SD.
`Abbreviation: 7-d RQ, 7-day recall questionnaire.
`
`Arch Phys Med Rehabil Vol 89, September 2008
`
`Exhibit 1042 page 3 of 7
`DENTAL IMAGING
`
`

`

`MEASURING FREE-LIVING PHYSICAL ACTIVITY, Hale
`
`1768
`
`A 250000
`
`s
`~
`·.:: .. Q.,
`'C
`0
`-"' ~
`-"'
`::;
`~
`
`200000
`
`,l-9
`
`1500000
`
`100000
`
`50000
`
`0
`
`T
`
`I
`-
`1
`
`l
`
`035
`
`~~ .l.
`
`■ -,T
`T T
`
`Mean +2S0
`
`Mean
`0
`
`Mean -2S0
`
`-200000
`
`>,
`
`:;
`> :;
`" Q ,:,
`'O ;
`~
`£
`.5
`" .c
`'o
`" " i
`
`0
`
`oo
`
`0
`
`0
`
`0
`
`0
`
`0
`
`0
`
`0
`
`0
`
`0 O
`
`0
`
`0
`
`0
`
`0
`
`0
`
`0
`
`200000
`
`400000
`
`
`
`VM - Mea n 3-Day MVM Mean Daily 7-Day M
`
`Fig 2. Bland-Altman analysis comparing MVM (in AU) collected
`over 7 days with that collected over 3 days.
`
`line (mean of the difference) and only 6 (6%) of 96 data points
`fell outside of the mean of the difference ⫾2 SD range
`(9938⫾43,150 AU). However, this analysis indicated that the
`3-day mean daily MVM data could differ from the 7-day mean
`daily MVM data by 86,300 AU. Given that the mean daily
`MVM ⫾ SD for the 7-day and 3-day data were 124,831⫾
`74,373 AU and 132,252⫾92,394 AU, respectively, this differ-
`ence may be considered large.
`Scatterplot and regressional analysis established that the
`RMI data accounted for only a small percentage of the varia-
`tion in activity data; RT3 data (MVM) had a slighter higher
`linear correlation with the RMI data (R2⫽.12, 16%) than the
`7-day recall questionnaire data (R2⫽.01, 1%). An ROC anal-
`ysis of these parameters revealed that the area under the curve
`for the MVM data was .72 (b⫽.02) and for the 7-day recall
`questionnaire was .61 (b⫽.13), indicating that the RT3 accel-
`erometer was more sensitive in distinguishing between people
`with varying levels of mobility than the 7-day recall question-
`naire.
`The results to the closed questions of the utility question-
`naire are provided in table 4. Question 1 asked whether wear-
`ing the accelerometer every day for 7 days was an acceptable
`method to measure daily activity. Questions 2, 3, and 4 en-
`quired how easy it was to remember to wear the accelerometer
`every day, whether it interfered with the daily routine, and
`whether it was annoying to wear. Question 5 checked whether
`
`2
`
`4
`
`la
`
`2a
`
`3a
`
`4a
`
`Diagnosis
`
`•
`111T
`J..,.
`.l
`,-T
`--- T T
`
`,.JS
`
`-r-
`
`047
`
`B 2500000
`
`Q.,
`
`s
`~
`0 ·.:: ..
`'C
`~
`'C
`=
`.....
`::;
`~
`
`2000000
`
`1500000
`
`1000000
`
`500000
`
`0
`
`2
`
`4
`
`la
`
`2a
`
`3a
`
`4a
`
`Diagnosis
`
`Fig 1. MVM (in AU) collected over 7 days and over 3 days for each
`test period versus diagnosis: (A) first test period and (B) second test
`period. Diagnosis legend: 1, PD; 2, stroke; 3, MS; 4, control group
`data collected over 7 days; 1a, PD; 2a, stroke; 3a, MS; 4a, control
`group data collected over 3 days. Legend: *participants with out-
`lying results.
`
`Paired Student t test analysis demonstrated a significant
`difference between the 7-day and 3-day data (P⫽.03). Bland-
`Altman analysis (fig 2) of these data showed good levels of
`agreement because most data points clustered around the zero
`
`Table 3: Test-Retest Reliability Data for Each Participant Group
`
`Test Period
`
`Total (N⫽47)
`
`MS (n⫽11)
`
`PD (n⫽7)
`
`Stroke (n⫽20)
`
`Controls (n⫽9)
`
`7-Day MVM test period 1 vs test period 2
`ICC
`95% CI
`P
`SEM (AU) (%)
`3-Day MVM test period 1 vs test period 2
`ICC
`95% CI
`P
`SEM (AU) (%)
`
`Abbreviation: SEM, SE of measurement.
`
`Arch Phys Med Rehabil Vol 89, September 2008
`
`.85
`.74–.91
`.00
`204,435 (23)
`
`.84
`.70–.91
`.00
`111,588 (27)
`
`.83
`.49–.95
`.00
`182,637 (17)
`
`.62
`.12–.88
`.01
`130,465 (25)
`
`.81
`.29–.96
`.01
`198,273 (23)
`
`.90
`.57–.98
`.001
`133,923 (28)
`
`.68
`.36–.86
`.00
`187,556 (28)
`
`.54
`.16–.79
`.00
`110,860 (35)
`
`.82
`.42–.96
`.002
`261,362 (19)
`
`.97
`.87–.99
`.00
`54,423 (9)
`
`Exhibit 1042 page 4 of 7
`DENTAL IMAGING
`
`

`

`MEASURING FREE-LIVING PHYSICAL ACTIVITY, Hale
`
`1769
`
`Table 4: Utility Questionnaire Data (Nⴝ47)
`
`Test Week 1
`
`Test Week 2
`
`Question
`
`Mean (Median)
`
`SD
`
`Range
`
`Mean (Median)
`
`SD
`
`Range
`
`Question 1 (1 ⫽ not acceptable; 9 ⫽ very acceptable)
`Question 2 (1 ⫽ difficult to remember; 9 ⫽ no problem)
`Question 3 (1 ⫽ interfered greatly; 9 ⫽ did not interfere at all)
`Question 4 (1 ⫽ most annoying; 9 ⫽ not annoying at all)
`Question 5
`
`Question 6
`
`NOTE. Values are in centimeters.
`
`2–9
`1.5
`7.5 (8)
`1–9
`2.2
`7.2 (8)
`2.5–9.0
`1.2
`7.8 (8)
`2.5–9.0
`1.5
`7.4 (8)
`Yes ⫽ 18 (38%), No ⫽ 26 (55%)
`Maybe ⫽ 3 (7%)
`Yes ⫽ 42 (89%), No ⫽ 1 (2%)
`Maybe ⫽ 4 (9%)
`
`1.5–9.0
`1.8
`7.2 (8)
`1.5–9.0
`2.0
`7.0 (8)
`2–9
`1.6
`7.5 (8)
`1–9
`2.2
`7.2 (8)
`Yes ⫽ 11 (23%), No ⫽ 33 (70%)
`Maybe ⫽ 3 (7%)
`Yes ⫽ 40 (85%), No ⫽ 2 (4%)
`Maybe ⫽ 5 (11%)
`
`the participant would mind wearing the accelerometer again as
`part of a research project. Question 6 sought
`to establish
`whether the accelerometer was a user-friendly method of mea-
`suring daily activity. The 1 open-ended question, which asked
`participants to comment on using the RT3, resulted in 67
`statements from 35 participants. These statements were
`grouped into common themes as follows: positioning the RT3
`in the middle of the back was uncomfortable (especially when
`sitting or driving), 24 (36%) of 67; participants were worried
`that the RT3 would fall off, especially in the toilet or bathroom,
`17 (25%) of 67; keeping the diary was burdensome, 4 (6%) of
`67; the RT3 was too big, 2 (3%) of 67; and some participants
`found the RT3 easy to wear and had no problems, 10 (15%)
`of 67.
`
`DISCUSSION
`The 8-week test-retest reliability of the RT3 accelerometer
`was good for data collected over 7 days (ICC⫽.85; 95% CI,
`.74⫺.91; P⫽.000) and 3 days (ICC⫽.84; 95% CI, .70⫺.91;
`P⫽.000). The absolute reliability for the total data, as calcu-
`lated with the SE of measurement, indicated that a second test
`would differ from a baseline test by ⫾23% (⫾204,435 AU),
`signifying that if the RT3 was to be used as a measure of
`change in physical activity levels, the minimal detectable dif-
`ference would have to be greater than 23% of the baseline
`measurement to allow for normal variance in weekly physical
`activity. Matthews et al29 reported that intraindividual variance
`accounted for 30% to 45% of the overall variance in acceler-
`ometer counts in healthy adults (n⫽92) measured over 21
`consecutive days (using the Computer Science Applications
`accelerometer). A 23% variation could therefore be considered
`a reasonable fluctuation in weekly activity patterns. It is not
`clear, however, what a 23% change in RT3 activity data would
`mean clinically. What increase in level and type of activity
`would this represent? It would probably depend on the level of
`activity the person was engaged in at baseline. An increase of
`23% on a very sedentary lifestyle would be far more meaning-
`ful than the same increase in activity in a person who was
`extremely active.
`The good test-retest reliability found in our study was similar
`to that found for other types of accelerometers during free-
`living activity monitoring trials. The ActiGraph accelerometer
`yielded an ICC of .93 while monitoring activity counts a day
`for 7 days in a sample of people with MS30 and the StepWatch
`step activity monitor an ICC of .86 and .89 over 7 days in
`adults with and without neurologic disorders, respectively.31
`Wearing an activity monitor for 1 week could be considered
`by some people to be onerous; however, the result of this study
`found the mean daily data collected in the first 3 days, despite
`its stability, to be significantly different from those collected
`
`over 1 week. The study by Matthews et al29 demonstrated the
`variable nature of daily physical activity, and these researchers
`proposed that a 7-day monitoring period provides the most
`reliable measurement of physical activity. In our study, the first
`3 days of each test period were week days, but because most
`participants in the study were unemployed, this was not con-
`sidered a problem. However, employment may not be an issue,
`because Motl et al30 demonstrated high reliability for both the
`pedometer and the ActiGraph accelerometer for all combina-
`tions of and types of days over a 1-week period in 193 adults
`with MS, 56% of whom were employed. To standardize for
`potential initial motivation in wearing the RT3, we chose to use
`the first 3 days of data collection for our analysis; however,
`further analysis could include comparing 7-day data with other
`groupings of 3-day data as undertaken by Motl.30
`The RT3 was found in this study to be able to distinguish
`between levels of mobility; this was not apparent when the
`7-day recall questionnaire data were plotted against the RMI
`data. Previous laboratory-based studies have demonstrated the
`RT3’s ability to distinguish between different velocities of
`treadmill walking and between low-intensity activities.11,13
`However, 1 study showed that as the intensity of activity
`increases, the degree of differentiation decreases, and sug-
`gested that the RT3 may be best suited to measuring activity in
`sedentary groups,11 such as used in our study.
`Participants did not find the RT3 a problem to wear and
`considered it to be a user-friendly, acceptable method for
`measuring physical activity. However, the location of the RT3
`in the middle of the back was considered by 36% of partici-
`pants to be uncomfortable, especially when sitting and driving,
`and it is suggested that the RT3 be worn on the side of the waist
`in the future. Some participants (25%) were worried the RT3
`would fall out of the provided holster. A more secure holster
`such as used for mobile cellular phones that are firmly attached
`to a belt would possibly be more secure than the present
`commercially supplied clip-on holster.
`Although the findings of this study are supportive of the use
`of the RT3 as a measure of free-living physical activity, the
`daily physical activity log kept by participants allowed us to
`verify RT3 data that appeared incorrect. This would imply that
`the RT3 should be used in conjunction with a simple daily
`activity log, which detracts from the utility of the instrument,
`but together provides a fairly comprehensive description and
`measure of daily physical activity.
`
`Study Limitations
`Our test-retest duration in this study could be considered
`long at 8 weeks and may have been a limitation. This duration
`was chosen because 8 weeks appears to be the minimum
`reported time for exercise to optimize muscle strength and
`
`Arch Phys Med Rehabil Vol 89, September 2008
`
`Exhibit 1042 page 5 of 7
`DENTAL IMAGING
`
`

`

`1770
`
`MEASURING FREE-LIVING PHYSICAL ACTIVITY, Hale
`
`cardiovascular function in order to potentially enhance physical
`activity in people with chronic neurologic dysfunction.32,33
`However, over an 8-week period, a change in seasons or in
`weather conditions can result in a change in physical activities
`(eg, from being able to walk in the community on a sunny
`autumn day to being unable to walk outside because of icy
`conditions). These variations were noted in the activity diaries.
`Seasonal variation may explain why, with the exception of the
`control group, the MVM data were lower for the second test
`period than for the first. A second possible explanation for this
`reduction in MVM data may be that the RT3 had an incentive
`role in the first week of testing. Participants knew that their
`activity levels were being monitored, and they were therefore
`motivated to do more activity. By the second test period, the
`novelty of wearing the RT3 had worn off, and it thus had less
`of an inspiring role. Although the RT3 measures the amount of
`activity undertaken, unlike a pedometer, this accumulating
`measurement cannot be read on the monitor screen, thus pos-
`sibly lessening the RT3’s ability to motivate. It was interesting
`to note that the differences in the 7-day recall questionnaire
`data between test periods were minimal in comparison with the
`MVM data.
`A further limitation of this study was the small sample size.
`We had aimed for 53 participants but collected complete data
`for only 47 participants; however, the width of the 95% CI for
`the total group ICC was within our a priori decision of 0.2. The
`internal validity of the subgroup analysis in this study was
`particularly limited by the small sample size, but the external
`validity of the study results was strengthened by the diverse
`nature of the sample.
`One of the problems encountered in this study was that of
`monitor dysfunction in 3 monitors, which resulted in lost data.
`In 1 case, the reason for monitor fault was unknown. The
`reasons for the other 2 monitors’ dysfunctions were that the
`monitors were dropped a couple of times, 1 monitor into water.
`Monitor failure is one of the limitations of equipment moni-
`toring of daily physical activity.34
`
`CONCLUSIONS
`The results of this study, based on 47 people both with
`and without neurologic dysfunction, indicate that the triaxial
`RT3 accelerometer provides a stable measure of free-living
`physical activity. On repeated testing,
`the value of the
`second measurement may differ by 23%. The RT3 can
`distinguish between people with varying levels of mobility
`and is well tolerated by participants. The results indicate that
`collecting data for 3 days is not reflective of data collected
`over 7 days.
`
`References
`1. Ministry of Health. Healthy eating— healthy action/Oranga Kai—
`Oranga Pumau: a background, 2003. Wellington: Ministry of Health;
`2003. Available at: http://www.moh.govt.nz/healthyeatinghealthyaction.
`Accessed June 3, 2008.
`2. World Health Organization. Fifty-Seventh World Health Assem-
`bly. Global strategy on diet, physical activity and health. Geneva:
`WHO; 2004. No. A57/9.
`3. Cooper RA, Quatrano LA, Axelson PW, et al. Research on phys-
`ical activity and health among people with disabilities: a consen-
`sus statement. J Rehabil Res Dev 1999;36:142-54.
`4. Bassett DR Jr, Cureton AL, Ainsworth BE. Measurement of daily
`walking distance— questionnaire versus pedometer. Med Sci
`Sports Exerc 2000;32:1018-23.
`
`Arch Phys Med Rehabil Vol 89, September 2008
`
`5. Mulcare JA, Mathews T. Final report: physical activity in persons
`with multiple sclerosis. Washington (DC): Department of Veter-
`ans Affairs; 2004. Merit review no. B747-3RS.
`6. Mathie MJ, Coster AC, Lovell NH, Celler BG. Accelerometry:
`providing an integrated, practical method for long-term, ambulatory
`monitoring of human movement. Physiol Meas 2004;25:R1-20.
`7. Melanson EL, Knoll JR, Bell ML, et al. Commercially available
`pedometers: considerations for accurate step counting. Prev Med
`2004;39:361-8.
`8. Esliger DW, Tremblay M. Technical reliability assessment of
`three accelerometer models in a mechanical setup. Med Sci Sports
`Exerc 2006;38:2173-81.
`9. Powell SM, Jones DI, Rowlands AV. Technical variability of the
`RT3 accelerometer. Med Sci Sports Exerc 2003;35:1773-8.
`10. Powell SM, Rowlands AV. Intermonitor variability of the RT3
`accelerometer during typical physical activities. Med Sci Sports
`Exerc 2004;36:324-30.
`11. Rowlands AV, Thomas PM, Eston RG, Topping R. Validation of
`the RT3 triaxial accelerometer for the assessment of physical
`activity. Med Sci Sports Exerc 2004;36:518-24.
`12. Chu EY, McManus AM, Yu CC. Calibration of the RT3 acceler-
`ometer for ambulation and nonambulation in children. Med Sci
`Sports Exerc 2007;39:2085-91.
`13. King AG, Torres N, Potter C, Brooks TJ, Coleman KJ. Compar-
`ison of activity monitors to estimate energy costs of treadmill
`exercise. Med Sci Sports Exerc 2004;36:1244-51.
`14. Hale L, William K, Ashton C, Connole T, McDowell H, Taylor C.
`Investigating the reliability and validity of the TriTrac RT3 Ac-
`celerometer in measuring mobility