Monitoring Activity in Individuals with Parkinson Disease: A Validity Study : Journal of Neurologic Physical Therapy

Secondary Logo

Journal Logo

ARTICLE

Monitoring Activity in Individuals with Parkinson Disease

A Validity Study

White, Daniel K. MSPT, NCS1; Wagenaar, Robert C PhD2; Ellis, Terry PT, PhD, NCS3

Author Information
Journal of Neurologic Physical Therapy 30(1):p 12-21, March 2006. | DOI: 10.1097/01.NPT.0000282145.10822.20
  • Free

Abstract

INTRODUCTION

Parkinson disease (PD) is a progressive neurodegenerative disorder prevalent in 128 to 187 individuals per 100,000 people.1The incidence of PD is increasing as the population ages. Individuals with PD have deficits at the body functions and structures level, which include rigidity, bradykinesia, tremor, and loss of postural control leading to activity limitations such as difficulty getting out of bed, dressing, getting up from a chair, and walking in the home and community settings.2 These limitations result in diminished participation leading to a decline in quality of life.3, 4 Clinicians and scientists assess individuals with PD in order to gauge disease severity, optimize medication schedules, and determine the effectiveness of intervention approaches. Commonly used methods of assessment include self-report and performance based measures focused on measuring body functions and structures,5,6 activities,7–9 and quality of life.10,11 These methods of assessment are often administered in a clinical or home environment at a discrete point in time. Given that the examination typically takes place in a fixed environment in a short time period, the generalizability of these measures across multiple settings over a continuous time period comes into question.

The ability to measure variation of functional status across different settings and time periods is crucial when measuring disablement experienced by individuals with PD. While PD medication has been shown to lessen the symptoms of PD,12the effects of medication often wear off through the day leading to considerable fluctuation in body functions and structures13,14and activity limitations.15,16 One major challenge clinicians and researchers face is determining how to best capture time dependent changes in functional status when measuring movement behavior. Attempts have been made to minimize variation of results through administering tests at a consistent time of day and measuring individuals at the peak and trough of the medication cycle. The major drawback to these methods is the limited transfer of findings to other time periods.

To better generalize to the full spectrum of functional ability, a longer testing period is necessary in a variety of environmental contexts. An instrument that potentially could fulfill these needs is an activity monitor (AM). The purpose of an AM is to continuously capture functional movements in a variety of environments, such as the clinic, home, community, and work settings without a clinician or researcher being present. The device is comprised of sensors placed on one or more body parts and a unit that records the signals from the sensors. The monitors are self contained and can be worn for periods up to several days. In individuals with PD, these recording devices have been used to study body function and structure limitations such as tremor and dyskinesias17–20 and limitations of activity such as time spent in body positions (eg, supine, sitting, and standing), movements (eg, walking and cycling), and transitions between body positions.21,22

In the present study, we investigated the validity of assessing functional status through the use of an AM (Vitaport 3: Temec Instruments BV, the Netherlands)23,24 in individuals with PD. The AM measures the length of time spent in specific positions including supine, sidelying, prone, sitting, and standing. The AM also counts the number of transitions between these positions and measures length of time spent walking, bicycling, and climbing up and down stairs. In addition, the AM calculates the number of walking episodes greater than 5 and 10 seconds, stride frequency, and provides an estimation of the overall duration of all activities. To calculate these measures of activity, accelerometer data is downloaded from the AM into a computer to allow for signal processing. The time spent in body positions, the number of transitions between body positions, and measures of gait are calculated through a series of algorithms embedded in software developed and distributed by the manufacturer. The AM can be used for all populations, healthy and diseased, and has been studied in adults with congestive heart failure and PD, and adolescents with meningomyelocele.22,25–27

While use of the AM has the potential to yield valuable quantitative measures of functioning in home and community environments, only a few studies have addressed the validity of the AM in measuring functional activities in those settings and only one study has focused on individuals with PD. Dunnewold et al examined the sensitivity and specificity of the AM to detect lying, sitting, and standing upright in 50 individuals with PD.27 Both walking and standing were classified as ‘standing upright.’ The participants were instructed to perform a fixed set of activities for 10 minutes followed by 5 minutes of activity chosen by the subjects. The duration of time spent in body positions was compared between the AM measures and video recordings. The mean specificity and sensitivity values for the entire module were 99.6% and 99.8% respectively; however, the accuracy of the AM with recording functional activities, such as walking and stair climbing, still needs to be determined in individuals with PD.

In 4 studies evaluating the validity of the AM, healthy individuals21 and participants with heart failure,22 transtibial amputations,28 and failed back surgeries29 were videotaped while performing functional activities, such as sitting, standing, and walking in their own manner and pace. Despite all studies using the same type of AM, agreement in these studies between the AM and video analysis (VA) varied from 13% to 100% due to (1) the variation in inclusion criteria related to the duration of activity and (2) inclusion of different activities. For example, activities (eg, sitting and standing) lasting less than 20 seconds were excluded from analysis in 2 studies22,29 while activities lasting less than 60 seconds were excluded from another study.21 The fourth study did not specify if activities below a given duration of time were excluded.28 In addition, 3 studies did not distinguish between different activities such as walking, running, bicycling, or stair climbing and combined those into one ‘dynamic’ activity measure.21,28,29 Finally, in one study that examined walking, the ability of the AM to detect individuals walking at slow velocities relevant to home and community settings appeared limited. Van den Berg-Emons et al reported a 13% accuracy of detecting slow walking with the AM when compared to VA.22 Further investigation of the AM's ability to detect slow walking is needed given this is a common symptom for individual's with PD.

Software used in the previous studies included a smoothing procedure that classified activities lasting less than 5 seconds as the same movement or posture preceding the short activity. For instance, if an individual in a chair quickly stood up and sat down in less than 5 seconds, the AM would classify the entire activity as sitting. The manner in which this procedure affects the validity of AM to detect functional tasks and transitions is not known since previous studies excluded activities lasting less than 20 to 60 seconds as part of their analyses. In addition, the influence of the type of protocol on validity of the AM needs further investigation. Bussmann et al found agreement to be higher during a protocol where subjects were asked to perform specific activities (96%) compared to a protocol where subjects were allowed to self-select activities over a 4-hour period (88%).21 These authors provided no explanation for the observed difference.21

Limited evidence exists regarding the AM's ability to (1) accurately distinguish between functional activities and (2) determine the length of time spent performing each activity. The purpose of this research is to determine the validity of the AM in measuring functional activities in individuals with PD. On the basis of findings reported in the literature and recent software updates, we hypothesize that the AM will accurately identify body positions and movements (ie, supine, sidelying, prone, sit, stand, walking, and bicycling) as well as accurately count the number of transitions between body positions and the number of 5 and 10 second walking periods in individuals with PD. Van den Berg-Emons et al reported that the AM misclassified slow walking as standing most likely due to the low frequency of movements. With recent software developments by the manufacturer, we hypothesize that there will be no difference between VA and AM in detecting walking at slow, comfortable, and fast velocities as well as recording stride frequency accurately during comfortable walking velocity. To test the AM's ability to detect various activities for discrete durations, subjects performed a series of activities in a systematic predetermined sequence and duration. A similar set of activities were performed in a random sequence at a self-selected pace. The purpose of the random sequence was to test the AM's ability to accurately detect movement reflecting the arbitrary order of activities which takes place in the home and community settings. We hypothesize the AM will detect activity with similar accuracy during the fixed and random modules in individuals with PD.

METHODS

Subjects

Individuals included in the study (1) had a diagnosis of idiopathic PD, (2) had a Hoehn and Yahr (H & Y) score of 2 or 3, (3) were between the ages of 40 to 79 years, (4) were on a stable dose of antiparkinsonian medications for 2 weeks prior to and during the course of the study, (5) were able to walk independently, (6) were able to understand and communicate with personnel, (7) lived independently in the Boston area, and (8) had no other neurological, cardiopulmonary, or orthopedic medical disorders that could interfere with their mobility. Informed consent was obtained from all subjects prior to participating in the study. The study was approved by the Boston University Institutional Review Board. The present study was coupled with a study on test-retest reliability of the AM.

Instruments

Two uni-axial (M92962) and 1 bi-axial (M92961) piezo-resistive accelerometers were used as sensors (size: 1.5 X 1.5 X.75 cm) (Temec Instrucments, BV, the Netherlands; see Figure 1). Sensors were attached to both thighs (lateral mid-shank) and to the middle of the sternum using TENS/NMES electrode pads (series 500; Empi, St. Paul, Minn) and double-sided tape. The pads holding the accelerometers were positioned on the skin parallel to the horizontal axis. Accelerometers were connected to a Vitaport3 data recorder (size 9cm X 4.5cm X 15cm) inserted in a padded bag that was carried around the waist (see Figure 2). The data recorder stores signals from the accelerometers and the hand held switch in a 128MB Compact Flash card at a sampling frequency of 32 Hz. The data recorder and bag weigh 1.36 kg. The data were downloaded into a computer (Pentium 4 at 1.2 GHz with 512MB RAM; Dell Inc, Round Rock, Tex) for detecting activities using a proprietary signal processing and inferencing language (S.P.I.L.) software package (Automatic Kinematic Analysis Version 9.0 June 2002). This software processes accelerometer data in 3 steps. The first step (feature extraction) processes signals with a low-pass filter to detect the relative angle of the accelerometer data to gravity or horizontal surfaces, and with a high-pass filter to detect repetitive frequencies that occur during activities such as walking. The next step (posture/motion detection) classifies processed data in 1 second intervals into categories of activities. These activities include the time spent in body positions (ie, supine, sidelying, prone, sitting, and standing), in addition to walking, cycling, running, and climbing up and down stairs. Movements that do not fit a specific category are classified as general movement. The classification of activities occurs by comparing processed signals to pre-established minimum and maximum values in an activity detection knowledge base. The activity detection knowledge base is a database of predetermined minimum and maximum signal values set by the manufacturer. The third step (post-processing) involves classifying activities lasting less than 5 seconds as the same movement or posture which preceded the short activity. This is done since most false activity detections are present for short durations of time.30 A more detailed description of AM signal processing has been published elsewhere.22,25 Functional activities during the present study were also recorded with a digital video camera (Canon Mini DV, Lake Success, NY).

F1-3
Figure 1:
The Activity Monitor includes the data recorder, marker, and 3 sensors.
F2-3
Figure 2:
Activity monitor shown as attached to an individual.

Protocol

A research assistant took no longer than 15 minutes to place the AM on the subject during the subject's self reported medication peak, or ‘on time.’ After receiving instruction, the subjects followed an experimental protocol consisting of 2 modules followed by a stair climbing task. The first module consisted of a series of tasks performed in a fixed order for predetermined durations. The second module consisted of a series of tasks performed in a random order for both predetermined and self selected durations. The beginning and end of each module was simultaneously marked by the AM and videotape by pushing the AM's marker switch and providing a hand signal (see Figure 1). The entire protocol (both movement modules and stair climbing) took no longer than 25 minutes. To avoid bias, the purpose of using the AM was explained to participants after completing the experimental protocol.

The fixed module consisted of a sequence of predetermined activities with fixed durations given to subjects in a stepwise fashion (see Table 1). The module began with subjects in a supine body position. Subjects were asked to transition to several body positions followed by walking at various speeds and durations around a small oval track (3.2m × 2.1m). Lastly, subjects were instructed to run for 15 seconds around the oval track. The fixed module ended with subjects sitting.

T1-3
Table 1:
Listed is the Sequential Order of Activities and Activity Duration in Seconds (sec) for the Fixed Module

The random module began after a 30- to 60-second break with the subjects seated in a chair. Subjects were asked to randomly choose one activity card from a deck of 11, perform the activity described on the card, and return to the chair to select another card. Thus each subject performed activities in a unique order. Activities on the cards were similar to those carried out in the fixed module with the exclusion of the prone body position and the addition of bicycling on a stationary bicycle (see Table 2).

T2-3
Table 2:
Listed are the Descriptions of the Tasks for Each Activity Card Used in the Random Module

Immediately following the conclusion of both modules, participants were brought to a stairwell of 10 stairs and were instructed to walk up the stairs, turn around, and walk back down the stairs in a comfortable manner.

Data Analysis

Three observers independently rated videotapes of the experimental protocol from each of the 11 subjects to establish inter-rater reliability. The raters were research assistants and achieved an Intraclass Correlation Coefficient [ICC (3,3)] of 0.997 for the fixed module and an ICC (3,3) of 0.991 for the random module. Duration of time in supine, sidelying, prone, sitting, standing, and walking was collected from the AM and VA for both modules at 1-second intervals. If a movement could not be classified into any of these positions, the activity was classified in the unknown or general movement category. The number of transitions between sit and stand were used for analysis since the majority of transitions occurred between these body positions. In addition, the total number of transitions was used for analysis. Walking periods lasting longer than 5- and 10-seconds were counted from the VA for both modules. Stride frequency was collected from the 60-second walking episode in the fixed module. Raters calculated stride frequency by counting the number of steps during the 60-second walking period. The AM detected ascending and descending stairs in 0 out of 9 individuals. The AM appeared to incorrectly classify other movements as stair climbing; hence we discontinued stair climbing from further analysis. The AM detected running in only 4 out of 8 participants who were able to run; therefore, no further statistical analyses were applied.

To investigate whether the AM adequately records the duration of activities, such as walking at slow, comfortable, and fast speeds, stride frequency, and bicycling, both the ICC (2,2) and the Spearman Rank Correlation Coefficient (rs) between VA and AM outcomes were calculated. The ICC was used to provide a parametric measure of correlation and agreement between the VA and AM. Since a small number of subjects were recruited for the study, the Spearman Rank Correlation Coefficient was used to provide a nonparametric analysis of the strength of association between the AM and VA. The ICC and rs were not calculated for walking durations in the fixed module due to the limited variability of time spent walking and the lack of transitions that occurred between walking bouts. Instead, the Wilcoxon Signed Rank Test (pw) was applied to determine if systematic differences existed between the VA and AM. The ICC and rs were calculated for the walking durations in the random module. To investigate if the AM adequately records the number of transitions between sit and stand, the total number of transitions, and the number of 5- and 10-second walking periods, the Kappa Statistic (K) was used to provide a chance-corrected measure of agreement between the AM and VA,31 and the pw was calculated to determine if systematic differences existed between the VA and AM. The mean differences and 95% confidence intervals between the AM and VA are presented for these measures. The K was also calculated for transitions between other body positions (eg, supine to sit, sit to supine, etc) to provide a measure of agreement between the AM and VA. However, mean differences and 95% confidence intervals between the AM and VA were not calculated due to the low number of transitions that occurred among these body positions. For the number of 5 and 10-second walking periods, p was calculated to examine if systematic differences existed between the AM and VA.

To investigate if the AM significantly over- or under-reported activity measures in one module compared to the other, the difference between AM and VA outcomes for each module were compared calculating the pw. All statistics were calculated using SAS (SAS Institute Inc, Cary, NC) with an alpha level of significance at 0.05.

RESULTS

Eleven individuals met the inclusion criteria and participated in the study. Subject characteristics are presented in Table 3. One participant (#6) frequently showed freezing and stopping during walking, thus his duration of time spent walking, walking periods, and stride frequency in both modules were excluded from analysis.

T3-3
Table 3:
Listed are the participant's gender (M=male, F=female), age in years (yrs), Hoehn and Yahr (H & Y) score, and time since diagnosis (DX) in years (yrs). Means and standard deviations (std) are also presented.

Fixed Module

The rs between the AM and VA ranged from 0.83 to 0.98 (p = 0.001) with the ICCs ranging from 0.54 to 0.95 for duration of time in body positions (supine, prone, sidelying, sitting, and standing; see Table 4). However, the AM data showed significantly increased duration of time compared to VA, ranging from 4.9 sec for the prone body position (95% CI [0.51 9.24]) to 19.6 sec for standing (95% CI [-6.5 45.8]).

T4-3
Table 4:
Listed are the times in seconds subjects spent in each body position for the fixed and random module as recorded by the VA and AM. Mean difference between the AM and VA (ΔM), the 95% Confidence interval (CI) for the mean difference, Spearman Rank Correlation Coefficient (rs), Intraclass correlation coefficient (2,2) (ICC), and the Wilcoxon Signed Rank Test (pw) are presented. If subjects were not able to attain a particular body position this was marked as a 0 for VA.

The K between the AM and VA for the number of transitions ranged from K=0.54 (p = 0.07) for transitions from sit to stand to K=1.0 (p < 0.001) for transitions from stand to sit (see Table 5) and no significant differences between AM and VA were found for the number of transitions between these body positions. For transitions among other body positions, the K between the AM and VA ranged from K=0.42 (p = 0.07) for supine to sit transitions to K=1.0 (p < 0.0001) for prone to sit transitions, with the exception of transitions from sit to sidelying, where K=0, (p = 0.99; see Table 6).

T5-3
Table 5:
Listed are the summary statistics for number of transitions from sit to stand, stand to sit and the total number of transitions, and 5- and 10-second walking periods for both modules. Statistics include the mean difference between the AM and VA (ΔM), the 95% Confidence intervals (CI) of the mean difference, the Wilcoxon Signed Rank Test (Pw), the Kappa Statistic (K), and p-values for the respective Kappa Statistic.
T6-3
Table 6:
Listed are the Kappa statistics (K) and respective p-values for agreement between the AM and VA for transitions between body positions and walking periods for the fixed and random modules. N/A indicates the transition did not occur in the respective module.

There was no significant difference between AM and VA for total time spent walking, which included slow, comfortable, and fast velocities. High correlation coefficients between the AM and VA were found for stride frequency (rs = 1.0, p < 0.001; ICC = 0.86) with no significant difference between the AM and VA. The K values found between the AM and VA were low for the number of 5- and 10-second walking periods (K = 0.028, p = 0.54; K = 0.21, p = 0.43, respectively; see Table 5). There were no significant differences between the AM and VA for the number of 5- and 10-second walking periods.

Random Module

The r2 between the AM and VA for the durations for body positions (ie, supine, prone, sidelying, sitting, and standing) ranged from 0.88 to 0.98 (p < 0.0001 to p = 0.003) and the ICC ranged from 0.89 to 0.98 (see Table 4). Significantly increased time durations for supine and sidelying positions were found for the AM when compared to VA though no significant differences were observed for sitting and standing positions.

The K between the AM and VA for the number of transitions from sit to stand, stand to sit, and total transitions were low and not significant (see Table 5). The AM data showed a significantly lower number of transitions from sit to stand, stand to sit, and the total number of transitions compared to VA. For transitions among other body positions, the K between AM and VA ranged from K = 0.46 (p = 0.25) for transitions from sidelying to sit to K= 0.0 for transitions from supine to stand (p = 0.98), sit to stand (p = 0.03), and stand to sit (p = 0.026; see Table 6).

The correlations coefficients for the duration of time walking at slow, comfortable, and fast velocities were high and significant (rs = 0.75, p = 0.01; ICC = 0.73; see Table 7). However, the AM data showed significantly increased walking durations compared to VA (pw = 0.05). The K between the AM and VA was low for the number of 5- and 10 second walking periods (K = 0.41, p = 0.72; K = −0.17, p = 0.72, respectively; see Table 5). The AM counted significantly more 5- and 10-second walking periods compared to VA (see Table 5).

T7-3
Table 7:
The table lists summary statistics for duration of time walking in both modules and bicycling in the random module. The mean difference between VA. and AM (ΔM), 95% Confidence intervals (CI) for mean differences, and the Wilcoxon Signed Rank Test (pw) are listed. The Spearman Rank Correlation Coefficient (rs), p-values for respective Spearman Rank Correlation Coefficients (p), and the Intraclass Correlation Coefficient (2,2) (ICC) are listed comparing the AM to VA for the duration of walking and bicycling in the random module.

The AM detected bicycling in 8 of 9 participants. The correlation coefficients between the AM and VA was moderately high (rs = 0.64, p = 0.033; ICC = 0.77), and there was no significant difference in reported time (see Table 7).

Differences between the fixed and random modules

The AM's over-reporting of duration of time in body positions was not significantly different between modules. While the AM under-reported the number of transitions in both modules compared to VA, the difference was larger between the AM and VA in the random module compared to the fixed module. The AM reported significantly lower transitions compared to VA for sitting-to-standing, standing-to-sitting, and total-transitions (pw = 0.004) in the random module compared to the fixed module. There were no significant differences for duration of time walking between modules. The AM reported significantly increased 5-and 10-second walking periods compared to VA (pw = 0.02, pw = 0.04, respectively) for the random module compared to the fixed module.

DISCUSSION

The purpose of this study was to evaluate the validity of the AM to detect functional activities in individuals with PD by comparing AM outcomes to those obtained by VA. The degree of association for the duration of time in body positions, walking, stride frequency, and bicycling ranged from moderate to high. However, the AM data showed significantly increased time spent in most postures and movements compared to VA in both modules. The agreement between the AM and VA was also high for the number of transitions between sitting and standing, and the total number of transitions in the fixed module; however, this was not found for the random module. The AM accurately recorded walking at slow, comfortable, and fast velocities as there were no significant differences between the AM and VA for both modules and the correlation coefficients between the AM and VA were high. The correlation coefficients were low for the number of 5- and 10-second walking periods in both modules, however, the actual differences were small ranging from 1 to 3 over-reported walking periods recorded by the AM compared to VA. Comparing the number of transitions reported in both modules by the AM and VA, the AM reported transitions with significantly less accuracy in the random module than in the fixed module.

A limitation of the AM is the 5-second smoothing procedure, which results in the AM omitting transitions which occur before and after an activity lasting less than 5 seconds. This was more often the case in the random module compared to the fixed module. During the random module, subjects sat in a chair between each of the 11 tasks to select an activity card. Given that the duration of time it took subjects to sit and select the card and start the next activity was often less than 5 seconds, the AM under-reported transitions from sit to stand, and stand to sit during the random module. Additionally, the total number of transitions was over-reported by the AM, since most of the transitions occurred between sitting and standing. Despite this finding, the AM accurately reported in both modules (1) transitions between other body positions, such as supine-to-sit and supine-to-stand, (2) the duration of time in body positions, and (3) walking at slow, comfortable, and fast velocities. These findings support the hypothesis that the AM detects activities with similar accuracy between the fixed and random modules, which is an improvement from Bussmann et al's observation that the AM recording functional activities with less accuracy in a random module.21

Conducting this study in a research environment may have limited generalizability to the home and community settings. A high number of activities were conducted in a limited amount of time during each module. As part of an additional study on test-retest reliability of the AM, we recorded the transitions for participants with PD who wore the AM at home and in the community for a 24-hour period (White D, Wagenaar RC, Del Olmo M, Ellis T, unpublished data, February 2006). These participants averaged about 6 transitions per hour versus 120 transitions per hour in the present study. We predict that the accuracy of the AM's reporting of transitions and time durations of body positions will improve with a lower frequency of transitions. While the home setting would be a more ideal environment to compare the AM to VA, there are significant barriers to this, such as videotaping the subjects' activities at home and the subjects' privacy. A possible solution to better reflect activities in the home and community settings would be to match the frequency of activities in the clinical modules to these settings.

A majority of the subjects recruited in our study had a similar Hoehn and Yahr disease severity classification. Most individuals had a Hoehn and Yahr score of 2 (mild to moderate symptoms of PD) and only 3 had a Hoehn and Yahr score of 3 (more severe symptoms of PD). Future research should indicate whether the findings generalize to individuals with Hoehn and Yahr 3 and 4.

Despite these limitations, results reported previously in the literature are consistent with the results of the present study.21,22,28,29 Bussman et al21,28,29 reports agreement between the AM and VA ranging from 87 to 90%. In the present study, there was high agreement between the AM and VA for the duration of body positions and walking activities in the fixed and random modules and the number of transitions in the fixed module. The present study also confirms the findings by Dunnewold et al indicating that supine, sitting, and upright body positions in individuals with PD27 can be adequately detected with the AM. Additionally, the present study demonstrates adequate detection of prone and sidelying body positions by the AM. Previous studies were not able to differentiate between walking and standing; both were combined into an ‘upright’ category in Dunnewold et al's validity study of the AM in individuals with PD.27 The present study provides evidence that the AM can differentiate between standing and walking. Lastly, contrary to the finding reported by van den Berg-Emons et al, the present study demonstrates the AM can accurately detect walking at slow velocities.

The influence of the AM's 5-second smoothing procedure on reported transitions and durations needs to be further investigated. Specifically, future research should examine the impact of the 5-second smoothing procedure on the accuracy of AM at a frequency of activity similar to that occurring in the home and community settings. Further investigation should examine the validity and reliability of the AMs ability to assess stair climbing and expanded activities which incorporate the upper extremities (eg, reading, preparing a meal, and grooming tasks). In addition, generalizability of our findings to a group of individuals with Hoehn and Yahr scores 3 and 4 (moderate to severe symptoms) has to be investigated.

CONCLUSION

The results of the study demonstrate that the AM accurately records functional activities and transitions between activities; however, the accuracy of the AM is reduced with the recoding of activities lasting shorter than 5 seconds. The AM does not accurately record the transitions to and from brief activities or the duration of time spent in the short activity. At present, the AM provides a unique opportunity for physical therapists and other professionals involved with rehabilitation to quantify time spent in body positions and walking, and the number of transitions between body positions in individuals with PD.

ACKNOWLEDGEMENTS

The authors would like to thank the following individuals for their time and effort with this research project: Mary E. Del Olmo and Laura Safranski for assistance with data collection and video analysis, Dr. Philip White for his review of the manuscript, and the individuals who participated in the study. This research project is funded by the National Institutes of Health/National Institute on Aging as part of the grant award RO1AG021152 entitled, Rehabilitation for Self-Management in Parkinson Disease.

REFERENCES

1 Tanner CM, Aston DA. Epidemiology of Parkinson's disease and akinetic syndromes. Current Opinion in Neurology. 2000;13:427–430.
2 Gleb DJ, Oliver E, Gilman S. Diagnostic criteria for Parkinson's disease. Arch Neurol. 1999;56:33–39.
3 Karlesen K, Tandeberg E, Arsland D, Larsen JP. Health related quality of life in Parkinson's disease: a prospective longitudinal study. J Neurol Neurosurg Psychiatry. 2000;69:584–589.
4 Dural A, Atay M, Akbostanci C, Kucukdeveci A. Impairment, disability, and life satisfaction in Parkinson's disease. Disabil Rehabil. 2003;25:318–323.
5 Schenkman M, Cutson TM, Kuchibhatla M, Chandler J, Pieper C. Reliability of impairment and physical performance measures for person's with Parkinson's disease. Phys Ther. 1997;77:19–27.
6 Schenkman M, Hughes MA, Bowden MG, Studenski SA. A clinical tool for measuring functional axial rotation. Phys Ther. 1995;75:151–156.
7 Podsiadlo D, Richardson S. The timed “up & go”: a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc. 1991;39:142–148.
8 Light KE, Behrman AL, Thigpen M, Triggs WJ. The 2-minute walk test: a tool for evaluating walking endurance in clients with Parkinson's disease. Neurology Report. 1997;21:136–139.
9 Duncan PW, Werner DK, Chandler J, Studenski SA. Functional reach: a new clinical measure of balance. J. Gerontol. 1990;45: M192-M197.
10 Peto K, Jenkinson C, Fitzpatrick R. PDQ-39: a review of the development, validation and application of a Parkinson's disease quality of life questionnaire and its associated measures. J Neurol. 1998;245:S10-S14.
11 Post MWM, De Bruin A, De Witte L, Schrijvers A. The SIP68: A measure of health-related functional status in rehabilitation medicine. Arch Phys Med Rehab. 1996;77:440–445.
12 Schaafsma J, Giladi N, Balash Y, Battels A, Gurevich T, Hausdorff J. Gait dynamics in Parkinson's disease: relationship to Parkinsonian features, falls and response to levodopa. J. Neurol Set. 2003;212:47–53.
13 Sage J, Mark M. Basic mechanisms of motor fluctuations. Neurology. 1994;44:S10-S14.
14 Poewe W. Clinical aspects of motor fluctuations in Parkinson's disease. Neurology. 1994;44:S6-S9.
15 Beckley D, Panzer V, Remler M, Hog L, Bloem BR. Clinical correlates of motor performance during paced postural tasks in Parkinson's disease. J Neurol Sci. 1995;132:133–138.
16 MacKay-Lyons M. Variability in spatiotemporal gait characteristics over the course of the L-dopa cycle in people with advanced parkinson disease. Phys Ther. 1998;78:1083–1094.
17 Jankovic J, Schwartz K, Ondo W. Re-emergent tremor of Parkinson's disease. J Neurol Neurosurg Psychiatry. 1999;67:646–650.
18 Manson A, Brown P, O'Sullivan J, Asselman P, Buckwell D, Lees A. An ambulatory dyskinesia monitor. J Neurol Neurosurg Psychiatry. 2000;68:196–201.
19 Hoff J, Plas A, Wagemans E, van Hilten J. Accelerometric assessment of Levodopa-Induced dyskinesias in Parkinson's disease. Mov Disord. 2001;16:58–61.
20 Keijsers NL, Horstink MW, van Hilten JJ, Hoff JI, Gielen CC. Detection and assessment of the severity of Levodopa-Induced dyskinesias in patients with Parkinson's disease by neural networks. Mov Disord. 2000;15:1104–1111.
21 Bussmann JB, Tulen JH, Van Herel EC, Stam HJ. Quantification of physical activities by means of ambulatory accelerometry: A validation study. Psychophysiology. 1998;35:488–496.
22 Van den Berg-Emons H, Bussmanm J, Balk A, Stam H. Validity of ambulatory accelerometry to quantify physical activity in heart failure. Scand J Rehab Med. 2000;32:187–192.
23 Bussmann JBJ, Stam HJ. Techniques for measurement and assessment of mobility in rehabilitation medicine: a theoretical approach. Clin Rehabil. 1998;12:513–522.
24 Veltink P, Bussmann H, de Vires W, Martens W, Lummel R. Detection of static and dynamic activities using uniaxial accelerometers. IEEE Trans Rehabil Eng 1996;4:395–402.
25 Van den Berg-Emons H, Bussmann J, Balk A, Keijzer-Oster D, Stam H. Level of activities associated with mobility during everyday life in patients with chronic congestive heart failure as measured with an Activity Monitor.' Phys Ther. 2001;81:1502–1511.
26 Van den Berg-Emonds H, Bussmann J, Brobbel A, Roebroeck M, Van Meeteren J, Stam HJ. Everyday physical activity in adolescents andyoung adults with meningomelocele as measured with a novel activity monitor. J Pediatr. 2001;139:880–886.
27 Dunnewold R, Hoff J, van Pelt H, Fredrikze P, Wagemans E, van Hilten B. Ambulatory quantitative assessment of body position, bradykinesia, and hypokinesia in Parkinson's disease. Clin Neurophysiol. 1998;15:235–242.
28 Bussmann J, Reuvekamp P, Veltink P, Martens W. Validity and reliability of measurements obtained with an Activity Monitor' in people with and without a transtibial amputation. Phys Ther. 1998;78:989–998.
29 Bussmann J, van de Larr Y, Neeleman M, Stam H. Ambulatory accelerometry to quantify motor behaviour in patients after failed back surgery: A validation study. Pain. 1998;74:153–161.
30 Bussmann J, Martens W, Tulen J, Schasfoort F, Van Den Berg H, Stam H. Measuring daily behavior using ambulatory accelerometry: The Activity Monitor. Behav Res Methods InstrumComput. 2001;33:349–356.
31 Cohen J. Coefficient of agreement for nominal scales. Educ Psychol Meas. 1960;20:37–46.
Keywords:

ambulatory monitoring; physical activities; Parkinson disease

© 2006 Neurology Section, APTA