Temporal stability of gait in Parkinson's disease.

Background and Purpose. Evaluation of physical therapy for gait disorders in patients with Parkinson's disease (PD) requires an understanding of how the patients' medication cycle affects function. Four experiments were conducted to investigate stability of gait variables. Methods and Results. In experiment 1, 15 subjects with idiopathic PD and 15 subjects without PD performed two sets of gait trials spaced 30 minutes apart, with the initial trial conducted with the subjects at a peak dose of medication. Strong correlations, as determined by intraclass correlation coefficients (ICG[2,1]), occurred between repeat measures of speed (ICC=.92), cadence (ICC=.92), stride length (ICG=.94), and time spent ill double support (DS) (ICG=.93) . In experiment 2, 16 subjects with PD were tested at the same time on two consecutive days. There was good repeatability for speed (ICG=.88), cadence (ICG=.85), stride length (ICC=.84), and DS (ICC=.80). When we assessed the repeatability of measurements taken at peak dose and at end of dose, there were low correlations for speed (ICC=-.54), cadence (IGG=-.07), stride length (ICG=-.35), and DS (ICC=-.38). In a final experiment on 16 subjects with PD, we used time-series analysis to examine the stability of measurements taken every 15 minutes for 21/2 hours. Slopes of regression models, standard deviations, and residual autocorrelations were negligible, indicating that the measurements were stable. Conclusion and Discussion. The parkinsonian gait pattern is reproducible across either brief time intervals or 24 hours when peak medication prevails. At the end of dose, however, marked changes in gait occur, apparently related to depletion of medication. [Morris ME, Matyas TA, Iansek R, Summers JJ. Temporal stability of gait in Parkinson's disease. Phys Ther. 1996;76:763-789.]

Idiopathic Parkinson's disease (PD) is a chronic neurological condition characterized by a progressive depletion of dopaminergic neurons in the brain-stem nuclei and a reduction in dapamine receptor sites in the striatum.[1] Motor fluctuations frequently emerge as the disease advances, particularly as the response to pharmacological therapy becomes more variable? The majority of people with idiopathic PD initially respond very favorably to parkinsonian medication; however, it is common for motor fluctuations to occur 'after 5 years or more of dapamine-replacement therapy.[2-4] The swings in motor performance associated with changing levels of dapamine uptake can result in disabling bouts of akinesia, hypokinesia, rigidity, dyskinesia, and dystonia. In turn, these movement disorders can lead to disability and handicap.[5]

The variability in motor performance associated with advanced idiopathic PD is well documented for disorders of upper-limb movement and affects the individual's ability to perform handwriting, prehension, and pegboard tasks.[6-7] Studies on cortical brain potentials prior to upper-limb movement demonstrate that activity in the motor cortical regions varies according to levodapa dosage.[8] The degree to which locomotor control fluctuates throughout the day or from one day to the next, however, remains uncertain. In addition, the stability and repeatability of gait measurements at different stages of the medication cycle has been poorly documented. In order to evaluate the effects of service provision, clinicians require reliable and valid measures of change in the gait pattern. They also need to have a model of gait stability against which changes in performance with treatment can be compared.

The need for reliable measures of gait stability is particularly important for the spatiotemporal variables, which provide a direct measure of the short-stepped, high-cadence walking pattern that is characteristic of idiopathic PD.[9,10] Spatiotemporal measures quantify distance and time aspects of the footstep pattern such as the walking speed, cadence (steps per minute), stride length, and proportion of the gait cycle spent in the double-limb-support phase (DS), in which both feet are in contact with the ground. Additional measures that have been used to quantify parkinsonian gait include joint kinematics[11-13] and electromyographic activity of muscles.[14-15] Although these measures might eventually prove useful in explaining the underlying mechanisms that contribute to the parkinsonian gait pattern, they do not allow description of the footstep pattern that is the hallmark of PD. Furthermore, their measurement is dependent on highly sophisticated research laboratory tools that are not readily accessible for use in the clinical setting.

Spatiotemporal measures frequently have been used to quantify the gait pattern in neurological conditions such as stroke[16-18] and PD.[9-11, 19-22] There has been little attempt, however, to establish the repeatability and stability over time of these gait measures in subjects with PD. Two studies of the repeatability of spatio-temporal measures in PD[6,7] have yielded somewhat equivocal results. Bowes et al[6] reported low product-moment correlation coefficients for repeatability of measures of walking speed (r=.24), stride length (r=.24), cadence (r=.36), and mean DS (r=.25) in subjects with idiopathic PD. In contrast, Stern et al7 reported finding highly significant correlations between paired observations for walking speed and stride length but gave no values for these correlations. Different methodologies, however, were used for these two investigations. Stern et al made repeat determinations of the gait pattern between 1 and 4 weeks after the initial test using a metal walkway system. They did not report reliability coefficients or document the locus of the gait cycle in which measures were taken. Bowes et ale' examined the gait pattern six times using a gait-assessment trolley following omission of a morning dose of levodopa. A 3-m length of cotton. attached to the subjects' heels and attached to geared pinch wheels of the gait assessment trolley enabled the footstep variables to be documented, and the output was displayed on a chart recorder.. The encumbrance of the cotton might have increased the error variance of the gait measures. In addition, the gait traces from the chart recorder were measured by hand, which the authors acknowledged could have reduced reliability. Alternatively, omission of the morning levodopa dose might have produced motor fluctuations.

Related studies on the response of parkinsonian gait to levodopa have shown variability of performance according to the subjects' medication status.[20-23] Pedersen et al,[23] for example, found that gait speed and stride length decreased markedly on withdrawal of parkinsonian medication, whereas changes in the walking cadence (steps per minute) were negligible. Blin et al[20] tested subjects with PD before and 1 hour after the administration of levodopa and found that gait speed and stride length were dopa-sensitive and that cadence did not change. These results suggest that control of stride length and gait speed might be mediated by dopamine, whereas the regulation of walking cadence might be under the control of nondopaminergic systems.

The findings of Blin et al[20] and Pedersen et ales indicate that stride length and gait speed vary according to levodopa levels. Whether meaningful fluctuations in stride length, gait speed, or even cadence occur within a medication cycle is not known. Whether there is variation in the parkinsonian gait pattern from day to day also is not known.

The purpose of our four experiments was to investigate the stability over time of the locomotor pattern in patients with PD. the aim of the first experiment was to examine the repeatability of gait measurements taken 2 minutes and 30 minutes apart when subjects were at peak dose of medication (on average, 60 minutes alter the medication was taken). In file second experiment, we evaluated the difference between gait measurements taken at the same time on consecutive days when subjects were at peak dose. the purpose of the third experiment was to investigate the effects of medication status on the gait pattern by assessing the repeatability of gait measurements taken from peak dose to end of dose, when motor fluctuations are most likely to emerge. The aim of the final experiment was to determine the stability of gait measurements taken over a longer time period within a medication cycle. Because physical therapy is typically provided within a 2- to g-hour time frame when patients are in the "on" phase of the medication cycle, we measured the gait pattern approximately every 15 minutes over a 165-minute period at peak dose. A proportion of the data analyzed in these experiments was used previously to investigate the pathogenesis of gait hypokinesia in persons with PD.[9-10] In addition, the results of experiment 4 provide a more detailed analysis of individual subject data obtained from a recent study of stride-length regulation in patients with PD.[24]

Method

Subjects

Forty-four subjects were recruited from inpatient facilities, the Movement Disorders Clinic, and the Volunteer Services Unit of the Kingston Centre (Cheltenham, Victoria, Australia). There were 15 subjects with PD and 15 control subjects in the first experiment, 16 subjects with PD in the second experiment, 12 subjects with PD in the third experiment, and 16 subjects with PD in the final experiment. Tables 1 through 4 provide a summary of the characteristics of the subjects with PD and show that some subjects participated in more than one experiment.

To be included ill the investigation, subjects were required to be 60 to 85 years of age, to be able to walk a 10-m distance repeatedly without assistive devices or assistance from another person, and to be able to provide informed consent according to the Declaration of Helsinki (1964). Subjects were excluded if they had neurological conditions other than idiopathic PD, as determined by a neurologist. Subjects also were excluded if they had visual impairment or musculoskeletal or cardiovascular disorders that affected locomotion or if they scored less than 20 on the Short Test of Mental Status (STMS).[25] The STMS is a cognitive test that allows subjects to be screened tot dementia, short-term memory disorders, and problem-solving and planning deficits and has been validated previously by Kokmen and colleagues.[25] the total score obtainable on the scale is 38, and scores less than 20 are thought to indicate the presence of dementia. Subjects with severe lower-limb dyskinesia or dystonia were excluded because these movement disorders prevented using the footswitch system. The presence of dyskinesia or dystonia would be expected to lead to excessive movement disorders that would decrease the reliability of test scores. Thus, the retest reliability of measurements of gait variables in subjects with dyskinesia or dystonia is considered to warrant independent analysis in a subsequent investigation, when suitable instrumentation for enabling data collection in subjects with these conditions is available.

All subjects were tested while they were taking their parkinsonian medication, although the time in the medication cycle at which testing occurred varied according to the aims of the experiment. In general, subjects with PD were judged by a physical therapist and a neurologist to be in the "on" phase of the medication cycle when they were able to move freely, easily, and quickly without dystonia, excessive rigidity, or tremor and when they subjectively reported that the medication had taken effect. This response occurred between 30 and 70 minutes after administration of medication. The type and dosage of medication were appropriate to the needs of each subject and included levodopa/benserazide hydrochloride, levodopa/carbidopa, bromocryptine, and benzhexol hydrochloride (Tabs. 1-4).

Subjects with PD also were scored on the Webster scale[26] by a neurologist. The Webster scale provides an indication of the general level of functional disability and includes items on gait, tremor, balance, rigidity, hypokinesia, seborrhea, facial expression, and speech. the maximum score obtainable on the Webster scale is 36, and scores greater than 20 indicate marked disability.

The final sample for experiment 1 consisted of 15 subjects with PD (6 men, 9 women) and 15 control subjects (6 men, 9 women). The mean age was 72.2 years (SD=6.2, range=62-81) for the subjects with PD and 72.5 years (SD=6.46, range=63-85) for the control subjects. The mean height was 1.63 m (SD=0.09, range=1.49-1.8) for the subjects with PD and 1.64 m (SD=0.09, range=1.45-1.72) for the control subjects. The mean body weight was 62.5 kg (SD=12.0, range=46-81) for the subjects with PD and 65.0 kg (SD=12.0, range=49-80) for the control subjects. The mean Webster scale[26] score was 13.1 (SD=2.9, range=8-18) for the subjects with PD, and their mean score on the STMS was 28.1 (SD=5.23, range=21-36). In experiment 1, subjects with PD were tested on average 1.0 hours (SD=0.25, range=0.5-1.5) after their midmorning dose of medication, when they were judged by a physical therapist or a neurologist to have an effective response and to be in the "on" phase of the medication cycle.

In experiment 2, there were 6 women and 10 men with PD. The subjects' mean age was 74.1 years (SD=5.6, range=64-82), their mean height was 1.65 m (SD=0.85, range=1.51-1.78), and their mean body weight was 68.6 kg (SD=11.1, range=49-85). For this experiment, the mean Webster scale[26] score was 14.1 (SD=4.4, range=7-24), and the mean score on the STMS was 29.1 (SD=3.34, range=23-34). Subjects with PD were tested on average 1.0 hours (SD=0.8, range=0.5-1.25) after their midmorning dose of medication on the first day, and they were tested on average 1.0 hours (SD=0.7, range=0.5-1.25) after taking their medication on the second day.

In experiment 3, there were seven men and five women with PD. The subjects' mean age was 70.1 years (SD=5.9, range=63-81, their mean height was 1.66 m (SD=0.076, range=1.53-1.80), and their mean body weight was 63.3 kg (SD=9.0,range=53-78). The mean Webster scale[26] score was 13.3 (SD=2.6, range=7-18) in the "on" phase of the medication cycle and 15.33 (SD=2.4, range=12-19) in the "off" phase of the medication cycle. The mean SRMS score in the "on" phase was 27.3 (SD=3.6, range=21-33). The subjects were tested on average 1.0 hours (SD=0.1, range=0.5-1.5) after their midmorning dose of medication and then on average 0.5 hours prior to the next dose of medication.

In experiment 4, there were nine men and seven women with PD. The subjects' mean age was 72.75 years (SD=5.07, range=64-81), their mean height was 1.65 m (SD=0.08, range=1.54-1.78), and their mean body weight was 65.6 kg (SD--9.7, range=49-82). The mean Webster scale[26] score was 14.88 (SD=4.1, range=9-24), and the mean STMS score was 29.06 (SD=4.7, range=21-36).

Apparatus

The gait patterns of the subjects with PD and the control subjects were measured using the Clinical Stride Analyzer.* The stride analyzer enables measurement of temporal footstep variables such as walking speed (measured as the time taken to walk a set distance), cadence (steps per minute), and DS. It also allows the stride length (length of two consecutive steps) to be derived. This system, which has been described previously in more detail,[9,10] consists of a pair of footswitches worn as innersoles within the shoes, a lightweight portable data logger worn as a backpack, and a personal computer. At the beginning and end of each 10-m walking trial, the researcher (MEM) activated a hand-held trigger attached to the data logger to signal the commencement and termination of data collection. The gait data were analyzed using PCSA software, version 1.06.*

The gait walkway was 12 m long and tiled in linoleum. Because individuals with PD can experience motor blocks ("freezing" episodes during walking) in confined environments,[27] We ensured 2 m of clearance at each end and on the sides of the walkway.

Procedure

The general procedure for gait analysis will be described, and then the procedural details of each experiment will be detailed. In each of the experiments, the subjects were tested on a modified Webster scale[26] by a neurologist and were given the STMS[25] by a physical therapist. Footswitches of appropriate size were fitted inside each subject's shoes, and the backpack was attached to the waist belt. The leads from the footswitches and the data-collection trigger were attached to the backpack. The subject was instructed to walk for a short period of time to accommodate to the apparatus and to identify any problems in footswitch fit. The subject was then instructed to "walk at a comfortable pace" down the gait walkway. the researcher walked 0.5 m behind the subject and activated the trigger at the beginning and end of the middle 10 m of each 12-m walk. After each trial, the subject walked back to the computer and the data were transferred from the backpack to the computer for further analysis. The subject was then instructed to sit down and have a brief rest. One 10-m practice trial preceded data collection for each of the conditions. After each reliability trial, the footswitches were removed from the shoes. the minimum number of steps over which the data were averaged was 15 steps for the subjects with PD and 20 steps for the control subjects.

In experiment 1, 15 subjects with PD and 15 control subjects matched for age, gender, and height performed two sets of two gait trials, with 30 minutes between sets. For this experiment, the first gait trial for subjects with PD was conducted on average 60 minutes after the midmorning dose of parkinsonian medication, when the subjects were judged to be at peak dose (the "on" phase of the medication cycle).

In experiment 2, a different group of 16 subjects with PD and matched control subjects were tested on average 60 minutes after the midmorning dose of medication, when they were judged to be in the "on" phase, and at the same time on the next day. The next dose of medication was scheduled on average 3 hours later. In this experiment, data from two consecutive gait trials performed 2 minutes apart on each day were analyzed.

The aim of the experiment 3 was to compare the retest reliability of gait measurements obtained at peak dose and at end of dose. In this experiment, 12 subjects walked at their preferred speed at optimal coverage of medication and 30 minutes prior to the next dose, when motor fluctuations should be most severe. On average, the interval between "on" and "off" measures was 21/2 hours.

In experiment 4, we examined the repeatability of gait measurements obtained every 15 minutes over a 2T/2hour period commencing 30 minutes after peak dose in 16 subjects with PD. This comparison enabled us to monitor the extent of motor fluctuations over a more extended period.

Spatial and Temporal Gait Variables

Measurements of the following gait variables were obtained for each subject on all trials: gait speed (in meters per minute), cadence (in steps per minute), stride length (in meters), and DS (expressed as a percentage of the gait cycle).

Statistical Analysis

The primary aim of the first three experiments was to investigate the repeatability of gait measures across two trials interspersed at different loci within the medication cycle. We therefore summarized the means and standard deviations for each of the dependent variables for each test and then examined the distribution of change scores over specified time intervals for each gait variable. Systematic trend was summarized by calculating mean change (D) over the relevant experimental interval. Paired t tests were used to determine whether systematic change in each gait variable was statistically significant. Variable change was estimated by obtaining the standard deviation of the change scores (SD.diff) and the 95% confidence interval (CI) for individual change scores (95% CI =D [+or-] Za=.05 X SD.diff).

Prior literature on PD gait has repeatedly reported Pearson product-moment correlations to summarize intertrial stability.[6,7] We therefore computed Pearson product-moment correlations for each variable to allow comparison of our results with those from previous studies. Because Pearson product-moment correlations are insensitive to systematic change, we also included intraclass correlation coefficients (ICC[2,1]) in the analyses. In addition to combining systematic and random variabilities in data, the ICC allows multiple sources of variability to be identified.[28]

In experiment 4, we used time-series analysis to investigate the stability of repeated measures of parkinsonian gait. Visual inspection of the results for each subject suggested that the data could be fitted using a linear model. After fitting each subject's data with a linear model, we assessed variability around the line of best fit by examining the standard deviation and autocorrelation function (ACF) of the residuals. The standard deviation was used as an index of overall magnitude of residual variability. The ACF assessed the presence of nonrandom components in the residuals, such as might arise from nonlinear or stochastic trends being present in addition to the linear trend accounted for by the regression model.[29]

Results

The means, standard deviations, and correlation coefficients for the pretest and posttest measures for each of the dependent variables in the first three experiments are presented in Tables 5 through 10. The change scores from the pretest to the posttest measures for each of the experiments are illustrated by a series of box plots in Figures 1 through 4. Box plots enable the median values of the change scores, the 10th, 25th, 75th, and 90tit percentiles, and the outliers to be displayed, which is important for the interpretation of data that are not normally distributed.

Experiment 1

The results of experiment 1 indicated that for repeat measures taken at 2-minute and 30-minute intervals, there was strong temporal stability of measures of walking speed, cadence, stride length, and DS (Figs. 1-4). As shown in Tables 7 and 8, the ICCs for the repeat determinations ranged from .71 to .94 for the subjects with PD and from .93 to .99 for the control subjects. For the subjects with PD, the differences between repeat determinations for each of the comparisons were not statistically significant. For the control subjects, paired t tests revealed an increase in cadence for the 2-minute test and increases in walking speed, cadence, and stride length for the 30-minute test. Fox' the 30-minute test, the standard deviations for the change scores for the control subjects were less than 3.5 steps/min for cadence, less than 0.03 m for stride length, and less than 2.5 m/min for speed.

We believe that the 95% CIs for each of the change scores for the 2-minute and 30-minute tests were within clinically acceptable limits, as shown by Figures 1 through 4. In addition, Figures 1 through 4 illustrate that the CI limits for all of the variables in both groups were narrower for the 30-minute test than for the 2-minute test. This finding was probably because only one gait trial was conducted for the pretest and posttest measures for the 2-minute test, whereas the mean of two consecutive gait trials was used for the pretest and posttest comparisons in the 30-minute test. The effect of averaging two trials was to reduce the standard deviation on the change scores. Therefore. even though the mean change was similar in the 2-minute and 30-minute tests, the limits of the CIs were smaller in the 30-minute test than in the 2-minute test.

Experiment 2

The results of experiment 2 showed stability in parkinsonian gait over a 24-hour period when subjects were tested at the same locus of the medication cycle. As with file first experiment, the means and standard deviations for walking speed and stride length in the subjects with PD indicated that these variables were stable from one day to the next (Tab. 9). None of the changes in gait variables were statistically significant.

Experiment 3

The ICCs for the repeat determinations from peak dose to end of dose in the subjects with PD ranged from -.54 to -.07 (Tab. 10), indicating an extremely poor degree of repeatability. The mean changes and standard deviations were large for each of the variables (Figs. 1-4). In addition, the upper and lower 95% Gis were very broad for walking speed, cadence, stride length, and DS.

Experiment 4

The time-series analyses conducted over a 2 1/2-hour period indicated little change in performance for any of the dependent variables during the "on" phase of medication in the majority of subjects. Visual inspection of the results for each individual suggested that a linear model could be fitted to the data. As documented in the group analyses by Morris et al,[24] the regression analyses obtained intercepts of 0.94+0.15 for stride length, 103.384-0.45 for cadence, and 36.204-5.60 for DS, using group mean values (+1 SD) to summarize the results. Similarly, the slopes were -0.00001 +0.001 for stride length, 0.224-0.85 for cadence, and 0.0054-0.02 for DS. Table 11 presents the slope and intercept values from the regression analysis for each subject with PD. Figure 5 illustrates the time series for stride length, cadence, and DS for subject 9, the characteristics of which were representative of 13 of the 16 subjects measured. The figure also illustrates the time series for each variable for an individual with greater variability of performance (subject 12). These data were representative of 3 of the 16 subjects. For this individual, the data could still be fitted using a linear model, and the standard deviation of the residuals was small for stride length (0.009), cadence (2.78), and DS (1.95). On occasions, the residuals appeared not to be entirely composed of random fluctuations around the line of best fit. The nonlinear residual, however, was of very small magnitude, in comparison with the differences in performance between subjects with PD and control subjects or in comparison with changes in patient performance previously reported for interventions such as physical therapy.[9,10] In addition, no autocorrelation in the residuals from linear regression was found.

Discussion

In this investigation, we documented the temporal stability of the walking pattern in subjects with idiopathic PD at different loci within the medication cycle and from one day to the next. The basic gait pattern exhibited by the subjects with PD in experiments 1, 2, and 4 was stable over time and was very similar to that previously reported for parkinsonian gait during the "on" phase of medication.[7,9-13,19,20] Compared with the gait pattern of the control subjects, the gait of the subjects with PD was very slow, with short strides and greater time spent with both feet in contact with the ground. The cadence values of the subjects with PD were similar to normal values in the majority of cases, despite reports in the literature that subjects with PD have a short-stride, high-cadence pattern.[7,9-13] Our results probably reflect the finding that the subjects with PD walked at slower free speeds. In comparison with reports of cadence rates for slow walking speeds in asymptomatic elderly people,[30,31] the cadence rates for the subjects with PD are higher than normal when speed of walking is taken into account.

The results of experiment 1 indicated that parkinsonian gait is stable from the period commencing on average 60 minutes postmedication to periods 2 minutes and 30 minutes later. In the subjects with PD, the ICCs for each of the dependent variables ranged from .71 to .94, indicating moderate to good repeatability. In the control subjects, the ICCs for repeat determinations obtained at 2 minutes and 30 minutes from the pretest were high, ranging from .93 to .99. As indicated by Figures 1 through 4, the median values and CIs for the change scores were relatively small for each of the variables hi both groups.

The 95% CIs for the change scores also suggested that the gait patterns of the subjects with PD were stable over time, provided that the subjects were in the "on" phase of medication. The CIs account for both the systematic error and the random error of measurement and provide the clinician with the upper and lower limits of the range of change scores due to measurement error. To give an example, the 95% CIs for the 30-minute test in the first experiment indicated that for subjects with PD, the stride length would need to increase by more than 0.10 m between two consecutive measurements for the change to represent real improvement rather than measurement error. Likewise, the stride length would need to decrease by more than 0.09 m between two consecutive measurements to reflect a decrease in stride length that is not simply due to measurement error. Given that changes in stride length in subjects with PD with interventions such as physical therapy have been reported in the range of 0.5 to 1.0 m,[9,10] the level of measurement error indicated by the Gis could be considered to be negligible.

In agreement with the reliability study conducted by Stern et al,[7] the results of experiment 2 indicated a high degree of repeatability for measures of walking speed, stride length, and DS from one day to the next when subjects with PD were measured at the same time of day during the "on" phase of the medication cycle. Intraclass correlation coefficients ranged from .80 for DS to .88 for walking speed.

In experiment 3, the ICCs for repeat determinations from peak dose to end of dose ("on" versus "off" values) showed a low degree of correlation (ICC=-.54 to -.07) and the change scores for each of the variables were large. Given the consistency of results of the other three experiments, it seems likely that this variability in test results was primarily the result of levodopa status rather than error introduced by the measurement apparatus or the procedures used. These results are similar to those of Bowes et al,[6] who reported product-moment correlation coefficients of .24 to .86 following omission of a morning dose of levodopa medication. Both investigations suggest that with apparent depletion of medication, more variable changes occur in subjects with PD.

In experiment 3, the 95% CIs about the change scores for stride length from peak dose to end of dose indicated that changes greater than 0.50 m would be needed to conclude that interventions such as physical therapy have an effect on stride length in addition to the effects of medication and measurement error. Similarly, stride length would have to decrease by more than 0.90 m with training to reflect deterioration not attributable to the effects of pharmacological therapy and measurement error. Mirroring these results, the 95% CIs about the change scores for walking speed and cadence for repeat determinations obtained from peak dose to end of dose were very large. Persons with PD would have to increase their speed of walking by more than 33 m/min for changes to be greater than the effects of measurement error and medication. Similarly, subjects would have to decrease their walking speed by more than 62 m/min to demonstrate a change not attributable to the effects of medication and measurement error. These very large values probably indicate that medication had a considerable impact on gait performance.

In experiment 4, we demonstrated that during a 165-minute period after levodopa was administered, the majority of subjects with PD maintained a very stable gait pattern. There were negligible slopes on linear regression analysis, and examination of the degree of autocorrelation in the residuals from linear regression revealed only minor fluctuations in performance that were not statistically significant. These findings indicate that in the period when physical therapy is usually provided, the gait pattern remains stable. Marked changes that occur concomitant with physical therapy during this period, therefore, are likely to be due to therapeutic processes rather than random variability associated with the locus of the medication cycle.

There are, however, a number of limitations in our investigation. Most notably, the sample sizes used in each of the experiments were relatively small, ranging from a minimum of 12 to a maximum of 16. In addition, only subjects who could perform a series of 10-m gait trials without assistive devices or without any help from another person were included. The sample, therefore, did not include people who were severely affected by PD, as the Webster scale[26] scores indicate. Subjects with dystonia and dyskinesia were also excluded from the sample. Furthermore, the focus of the investigation was on impairment variables such as gait speed and stride length rather than on functional variables such as gait endurance, maximum walking distance, or energy expenditure during ambulation. These factors limit the generalizability and application of the findings.

A number of implications for clinical practice, however, can be derived from our study. The results suggest that when evaluating the effects of interventions such as physical therapy for persons with PD, repeat assessments of the walking pattern at peak dose will be much more reliable than assessments that incorporate a pretest measure during the first 2 1/2 hours after levodopa medication and a posttest measure in the half hour prior to the next dose. The high level of random variability in repeat measures from the "on" phase to the "off" phase of the medication cycle suggests that repeat measures of this type should be avoided.

The results indicate that patient performance fluctuates within a medication cycle and that even at peak dose of medication the walking pattern is not completely normal. This finding might signal the need for movement rehabilitation strategies to augment the beneficial effects of pharmacological therapy, particularly at the end of dose of medication when the walking pattern can be markedly hypokinetic and akinetic. Current strategies for movement rehabilitation in persons with PD are documented in the summaries by Schenkman,[32] Turnbull,[33] Banks,[34] and Morris et al.[35]

The results of experiment 1 highlighted that the standard deviation, and thus the CIs about file change scores, are reduced when the mean of two or more gait trials is used to evaluate change in performance, rather than comparing the data from single gait trials. In this respect, we recommend that when physical therapists use a 10-m gait test, they perform a minimum of two gait trials and average the data. Alternatively, therapists could consider measuring the gait pattern over more than 10 m (perhaps 15 or 20 m) to minimize variability in test results. Further research is needed to determine which of these procedures would be most effective in reducing error variance. Further research also is needed to determine the reliability of repeat determinations over longer periods, such as 1 week, 1 month, or 6 months, to assist clinicians in interpreting their gait assessment findings over these longer time intervals.

Conclusions

The results of this investigation have shown a high degree of stability over time and good repeatability within a session and from one day to the next for gait speed, stride length, and DS measured using a footswitch system, provided that subjects with PD were in the "on" phase of the medication cycle. In marked contrast, the repeatability of measures Dom peak dose to end of dose was very poor, and there was considerable deterioration in performance at the end of dose. These findings may not be surprising to physical therapy clinicians who specialize in the management of PD; however, they provide the first comprehensive, quantitative, and controlled confirmation of the temporal stability of the walking pattern at different loci within the medication cycle.

Acknowledgments

We thank the subjects who gave so 'freely of their time for this investigation, the physical therapists at Kingston Centre for their comments on the manuscript, and Cameron Grant for technical assistance.

Invited Commentary and Author Response follow on pages 778-780.

References

1 Marsden CD. Parkinson's disease. J Neurol Neurosurg Psychiatry. 1994; 57:672-681.

2 Marsden CD, Fahn S. Problems in Parkinson's disease and other akinetic-rigid syndromes. In: Marsden CD, Fahn S, eds. Movement Disorders, Volume 3. Oxford, England: Butterworth-Heinemann; 1994: 117-123.

3 Nutt JG, Woodward WR, Hammerstad JP, et al. The "on-off" phenomenon in Parkinson's disease: relation to levodopa absorption and transport. N Engl J Med. 1984;310:483-488.

4 Obeso JA, Luquin MR, Grandas L, et al. Motor response to repeated dopaminergic stimulation in Parkinson's disease. Clin Neuropharmacol. 1992; 15: 75-79.

5 Schenkman M, Donoran J, Tsubota J, et al. Management of individuals with Parkinson's disease: rationale and case studies. Phys Ther. 1989;69:944-955.

6 Bowes SG, Clark PK, Chadett A, et al. Objective outcome criteria in trials of anti-Parkinson therapy in the elderly: sensitivity, specificity, and reliability of measures of brady- and hypo-kinesia. Br J Clin Pharmacol. 1991;31:295-304.

7 Stern GM, Franklyn SE, Imms FJ, Prestidge SP. Quantitative assessments of gait and mobility in Parkinson's disease. J Neural Transm Suppl. 1983;19:201-214.

8 Dick JPR, Rothwell JC, Day BL, et al. The bereitschafts-potential is abnormal in Parkinson's disease. Brain. 1989;112:233-244.

9 Morris ME, Iansek R, Matyas TA, Summers JI. The pathogenesis of gait hypokinesia in Parkinson's disease. Brain. 1994;117:1161-1182.

10 Morris ME, Iansek R, Matyas TA, Summers JI. The ability to modulate walking cadence remains intact in Parkinson's disease. J Neurol Neurosurg Psychiatry. 1994;57:1532-1534.

11 Murray P, Sepic SB, Gardner GM, Dowries WJ. Walking patterns of men with parkinsonism. Am J Phys Med. 1978;278-294.

12 Knutsson E. An analysis of parkinsonian gait. Brain. 1972;95:475486.

13 Knutsson E, Martensson A. Quantitative effects of L-dopa on different types of movements and muscle tone in parkinsonian patients. Scand J Rehabil Med. 1971;3:121-130.

14 Buchthal F, Ferandez-Ballesteros ML. Electromyographic study of the muscles of the upper arm and shoulder during walking in patients with Parkinson's disease. Brain. 1965;88:875-896.

15 Richards CL, Cioni M, Malouin F, et al. Changes in the gait of patients with Parkinson's disease induced by sensory cues and L-dopa. In: Proceedings of the Sixth Biennial Conference, Canadian Society for Biomechanics. 1990:199-202.

16 Brandstater M, deBruin H, Gowland C, Clark B. Hemiplegic gait: analysis of temporal variables. Arch Phys Med Rehabil. 1983;64:583-587.

17 Hill KD, Goldie PA, Baker PA, Greenwood KM. Retest reliability of the temporal and distance characteristics of hemiplegic gait using a footswitch system. Arch Phys Med Rehabil. 1994;75:577-583.

18 Wall J, Turnbull GI. Gait asymmetries in residual hemiplegia. Arch Phys Med Rehabil. 1986;67:550-553.

19 Blin O, Ferrandez AM, Serratrice G. Quantitative analysis of gait in Parkinson patients: increased variability of stride length. J Neurol Sci. 1990:98:91-97.

20 Blin O, Ferrandez AM, Pailhouse J, Serratrice G. Dopa-sensitive and dopa-resistant gait parameters in Parkinson's disease. J Neurol Sci. 1991;103:51-54.

21 Bowes SG, Clark PK, Leeman AL, et al. Determinants of gait in the elderly parkinsonian on maintenance levodopa-carbidopa therapy. Br J Clin Pharmacol. 1990;30:13-24.

22 Bowes SG, Charleft A, Dobbs Rj, Lubel DD, et al. Gait in relation to ageing and idiopathic parkinsonism. Scand J Rehabil Med. 1992;24:181186.

23 Pedersen SW, Eriksson T, Oberg B. Effects of withdrawal of antiparkinson medication on gait and clinical score in the Parkinson patient. Acta Neurol Scand. 1991;84:7-13.

24 Morris ME, Iansek R, Matyas TA, Summers JJ. Stride length regulation in Parkinson's disease: normalization strategies and underlying mechanisms. Brain. In press.

25 Kokmen E, Smith GE, Petersen RC, et al. The Short Test of Mental Status: correlations with standardized psychometric testing. Arch Neurol. 1991;48:725-728.

26 Webster DD. Critical analysis of the disability in Parkinson's disease. Modern Treatment. 1992;5:257-282.

27 Giladi N, McMahon D, Przedborski S, et al. Motor blocks in Parkinson's disease. Neurology. 1992;42:333-339.

28 Polger S, Thomas SA. Introduction to Research in the Health Sciences. Melbourne, Victoria, Australia: Churchill Livingstone; 1991.

29 Gottman JM. Time Series Analysis: A Comprehensive Introduction for Social Scientists. New York, NY: Cambridge University Press; 1981.

30 Leiper CI, Craik RL. Relationships between physical activity and temporal-distance characteristics of walking in elderly women. Phys Ther. 1991;71:791-803.

31 Murray MP, Kory RC, Clarkson BH. Walking patterns in healthy old men. J Gerontol. 1969;24:169-178.

32 Schenkman M. Physical therapy intervention for the ambulatory patient. In: Turnbull GI, ed. Physical Therapy Management of Parkinson's Disease. New York, NY: Churchill Livingstone Inc; 1992:137-192.

33 Turnbull GI. The role of physical therapy intervention. In: Turnbull GI, ed. Physical Therapy Management of Parkinson's Disease. New York, NY: Churchill Livingstone Inc; 1992:91-111.

34 Banks MA. Physiotherapy. In: Caird FI, ed. Rehabilitation in Parkinson's Disease. London, England: Chapman & Hall Ltd; 1991:45-65.

35 Morris ME, Iansek R, Summers JI, Matyas TA. Motor control considerations for the rehabilitation of gait in Parkinson's disease. In: Glencross DI, Piek JP, eds. Motor Control and Sensory Motor Integration: Issues and Directions. Amsterdam, the Netherlands: Elsevier Science Publishers BV; 1995:61-93.

Key Words: Gait, Movement disorders, Parkinson's disease, Physical therapy. TA Matyas, PhD, is Associate Professor, School of Behavioural Health Sciences, La Trobe University:

R Iansek, PhD, FRACP, is Neurologist, Geriatric Research Unit, Kingston Centre.

JJ Summers, PhD, is Professor, School of Psychology, University of Southern Queensland, Toowoomba, Queensland, Australia 4350.

This study was approved by the ethics committees at La Trobe University and Kingston Centre.

ME Morris, PhD, PT, is a physiotherapist and Manager, Geriatric Research Unit, Kingston Centre, Warrigal Rd, Cheltenham, Victoria, Australia 3192, and Senior Lecturer, Scimol of Physiotherapy, La Trobe University, Bundoora, Victoria, Australia 3083. She was a postgraduate student in the School of Behavioural Health Sciences, La Trobe University, at the time of this study. Address all correspondence to Dr Morris at the Kingston Centre address.

This research was funded by a grant from the Sandoz Foundation for Gerontological Research and by the Victorian Neurology Special Interest Group of the Australian Physiotherapy Association.

  Table 1.
    Characteristics of Subjects With Parkinson's
Disease (PD) in Experiment 1

                                             Webster
Subject  Age              Height    Weight    Scale
No.      (y)     Gender    (m)       (kg)     Rating

1        81        M        1.75       81        13
2        73        F        1.64       76        8
3        63        F        1.57       56        11
4        62        F        1.57       59        12
5        76        M        1.7        67        14
6        76        M        1.69       67        16
7        74        F        1.57       57        11
8        78        M        1.68       60        13
9        72        F        1.56       46        18
10       64        M        1.8        80        14
11       67        M        1.67       80        10
12       68        F        1.67       58        11
13       80        F        1.59       53        12
14       78        F        1.49       50        18
15       71        F        1.53       48        16

Subject          Duration
No.      STMS    of PD                 Dosage
         [a]      (y)    Medication    (mg/d)

1        29        13    Parlodel          10
                         Sinemet       300/75
                         Artane             2
2        29              Madopar      600/150
3        27              Madopar      400/100
                         Sinemet CR   1,000/250
4        31        17    Sinemet       250/62.5
                         Parlodel         7.5
                         Artane             4
5        21        12    Madopar M     300/75
                         Sinemet CR    200/50
                         Motilium          60
6        21              Madopar M    400/100
                         Madopar HB   500/125
7        26              Madopar HB    300/75
                         Parlodel          15
                         Eldepryl           5
                         Motilium          20
8        31        6     Sinemet       750/75
9        34        10    Madopar Q   150/37.5
                         Parlodel          15
10       23        5     Madopar HB   600/150
                         Parlodel          15
11       36        6     Sinemet      400/100
                         Sinemet CR    200/50
12       35        15    Sinemet      400/100
                         Motilium          30
13       23        5     Sinemet      500/125
                         Motilium          30
14       22        10    Madopar      800/200
                         Sinemet CR    200/50
15       33        10    Sinemet      500/125
                         Parlodel         7.5

"STMS=Short Test of Mental Status.

  Table 2.
    Characteristics of Subjects With Parkinson's Disease (PD)
in Experiment 2

Subject
No.
          Other        Age       Gender        Height       Weight
       Experiments     (y)
           [a]

1          1,4          76          M           1.71           68
2                       76          M            1.7           80
3                       80          F           1.51           58
4           3           65          M           1.52           62
5           4           77          F           1.54           57
6                       71          M           1.68           85
7           3           81          M            1.7           80
8                       82          F            1.6           57
9          3, 4         70          M            1.7           70
10          3           68          M           1.72           82
11       1, 3, 4        75          F           1.57           56
12                      81          M           1.69           72
13       1, 3, 4        71          F           1.54           49
14          4           72          M           1.68           70
15       1, 3, 4        64          M           1.78           81
16                      76          F           1.69           70

          Webster             Duration
           Scale                of        Medi-          Dosage
          Rating      STMS[b]  PD(y)      cation          (mg/d)

1           24          34      12       Madopar M       300/75
                                         Motilium         60
2           16          28      2        Madopar        900/225
3           11          30      6        Sinemet M       100/10
                                         Madopar       HBS400/100
                                         Eldepryl          5
4           21          26      8        Parlodel         15
                                         Sinemet CR     800/200
                                         Sinemet M       100/10
5           17          25      4        Parlodel         15
                                         Sinemet         250/25
                                         Sinemet CR     500/125
6           7           30      5        Sinemet M       500/50
7           13          32      8        Sinemet CR     500/125
                                         Sinemet        150/37.5
8           10          28               Madopar        450/112.5
                                         Motilium         20
9           16          24      12       Sinemet CR     500/125
                                         Parlodel         12.5
10          9           30      12       Madopar        600/150
                                         Sinemet CR      200/50
                                         Eldepryl         10
11          11          33      4        Sinemet CR     800/200
                                         Sinemet CR      300/75
                                         Eldepryl         10
                                         Madopar HBS    400/100
                                         Pergolide      100 mcg
12          12          32      10       Sinemet         250/25
                                         Sinemet CR     800/200
                                         Pergolide       375 mg
13          13          33      10       Sinemet       1,250/125
                                         Parlodel        7.5
                                         Motilium         60
14          17          28      12       Sinemet CR      200/50
                                         Madopar        600/150
15          14          23      5        Madopar M       100/25
                                         Madopar HBS     300/75
                                         Parlodel         15
16          14          30     0.5       Madopar Q      150/37.5

"Other experiments in which the subject participated.

"STMS=Short Test of Mental Status.

  Table 3.
    Characteristics of Subjects With Parkinson's
Disease (PD) in Experiment 3

Subject
No.    Other    Age   Gender    Height    Weight    Webster
       Experi-                   (m)       (kg)      Scale
       ments                                        Rating
                                                      On

1               80      F       1.58       60          15
2               69      F        1.6       53          13
3      2, 4     71      M        1.7       70          14
4       1       63      F       1.66       60          13
5    1, 2, 4    72      F       1.53       48          12
6       2       66      M       1.64       56          15
7               63      M       1.68       65          7
8       2       68      M       1.72       70          12
9       4       69      M        1.7       70          18
10   1, 2, 4    74      F       1.57       57          15
11      2       81      M        1.7       72          13
12   1, 2, 4    65      M        1.8       78          12

         Webster   STMS   Duration
          Scale     (b)    of        Medicaiton      Dosage
         Rating            PD (y)                    (mg/d)
           Off

1           17      25      10       Sinemet          800/200
2           16      26      6        Sinemet         6 00/150
3           18      24      13       Sinemet CR     1,200/300
4           14      31      17       Madopar          800/200
                                     Motilium           20
5           12      33      10       Sinemet         750/67.5
                                     Sinemet CR       100/25
                                     Motilium           60
6           17      26      8        Sinemet CR     1,800/450
                                     Sinemet          200/50
7           15      28      8        Sinemet          800/200
8           12      30      12       Madopar          600/150
                                     Eldepryl            5
                                     Sinemet CR       300/75
9           19      21      6        Sinemet          200/20
10          17      26      3        Madopar HBS      300/75
                                     Parlodel           15
                                     Eldepryl            5
                                     Motilium           20
11          15      32      8        Sinemet CR     1,500/375
                                     Sinemet          300/75
12          12      25      5        Madopar HBS      600/150
                                     Parlodel           15

[a] "Other experiments in which the subject participated. b STMS=Short Test of Mental Status.

* B&L Engineering, Santa Fe Springs, CA 90670.

  Table 4.
    Characteristics of Subjects With Parkinson's
Disease (PD) in Experiment 4

Subject
No.                                                    Webster
          Other       Age    Gender   Height  Weight    Scale
       Experiments    (y)                              Rating
           [a]

1           1         81       M       1.75      82      16
2                     64       F       1.6       53      10
3        1, 2, 3      64       M       1.78      81      12
4           1         67       M       1.76      79      9
5                     79       F       1.58      60      18
6          1, 2       76       M       1.71      68      24
7        1, 2, 3      71       F       1.54      49      14
8          2, 3       70       M       1.73      70      16
9           1         75       M       1.69      65      20
10                    78       M       1.66      66      11
11                    73       F       1.61      64      16
12                    73       F       1.55      60      15
13          2         77       F       1.54      57      12
14          3         69       M       1.7       70      13
15       1, 2, 3      75       F       1.57      56      12
16          2         72       M       1.68      70      20

              Duration
                of       Medication         Dosage
      STMS[b]  PD(y)                        (mg/d)

1       32      13       Sinemet          600/150
2       36               Madopar          500/125
3       23      5        Madopar M         100/25
                         Madopar HBS       100/25
                         Parlodel           15
4       33      6        Sinemet        1,000/250
                         Sinemet CR        200/50
5       31      12       Sinemet        1,000/250
                         Madopar HBS       300/75
6       34      12       Madopar M         300/75
                         Domperidone        20
7       33      10       Sinemet        1,200/300
                         Parlodel          7.5
                         Motilium           60
8       24      12       Sinemet          500/125
                         Parlodel         12.5
9       24      6        Madopar          800/200
                         Sinemet CR     2,000/500
10      26      6        Madopar          900/225
                         Sinemet CR       900/225
                         Sinemet CR       600/150
                         Eldepryl            5
11      29      7        Parlodel            5
                         Madopar           300/75
12      33      10       Sinemet           200/50
                         Sinemet CR     1,200/300
13      25      4        Sinemet           100/25
                         Sinemet CR     1,800/450
14      21      6        Sinemet           300/75
                         Parlodel           7.5
15      33      4        Madopar HBS      400/100
                         Parlodel           15
                         Eldepryl            5
16      28      12       Madopar        1,250/312
                         Sinemet CR        200/50
                         Motilium           30

"Other experiments in which the subject participated.

+, STMS:Short Test of Mental Status.

Table 5.

Means and Standard Deviations for Pretest and Posttest Scores for Subjects With Parkinson's Disease in Experiments 1 Through 3

  Pretest        Posttest

Experiment                  X       SD      X       SD
1 (2 min)
Walking speed (m/min)      48.1    10.6    47.9    9.3
Cadence (steps/min)       104.61  10.09   103.53   8.97
Stride length (m)         0.913    0.16   0.919   0.135
DS[a] (% of gait          32.59    3.4    32.27    2.84
cycle)

1 (30 min)
Walking speed (m/min)     47.98    9.52   48.58    8.36
Cadence (steps/min)       104.07   9.26   104.88   8.56
Stride length (m)         0.916   0.142   0.924   0.135
DS (% of gait cycle)       32.6    2.84   32.89    2.55

2 Walking speed (m/min)    51.7    10.4    53.5    12.4
Cadence (steps/min)       104.33  11.75   104.22  13.55
Stride length (m)         0.992   0.152   1.018   0.161
DS (% of gait cycle)      36.38    5.7     35.3    6.1

3 Walking speed (m/min)    61.9    10.4   47.15    17.2
Cadence (steps/min)       114.31   8.55   105.03  15.59
Stride length (m)         1.085   0.184   0.878   0.258
DS (% of gait cycle)      33.53    2.69   36.48    5.9

"DS=double-limb-support phase of gait.

  Table 6.
    Means and Standard Deviations for Pretest and
Posttest Scores for Control Subjects in
Experiment 1

                             Pretest            Posttest

Experiment                  X       SD         X       SD

1 (2 min)
Walking speed (m/min)    70.17     10.59     71.41     10.89
Cadence (steps/min)      110.47     9.28     112.23     8.94
Stride length (m)        1.272     0.163     1.272     0.155
DS[a] (% of gait         32.19     32.27     32.27      4.83
cycle)

1 (30 min)
Walking speed (m/min)    70.79      10.6     71.92     10.86
Cadence (steps/min)      111.38     8.98     111.86     8.83
Stride length (m)        1.272     0.158     1.287     0.164
DS (% of gait cycle)     32.23      4.8      32.45      4.68

"DS-double-limb--support phase of gait.

  Table 7.
  Means and Standard Deviations of Change Scores, 95% Confidence
Intervals (CI), t Values, and Correlation Coefficients for
Repeat Measurements Taken at 2-Minute Intervals

                            Change Scores
                                               Lower
Group                       D   [SD.sub.diff]  95% CI

Subjects with PDb
Walking speed (m/min)    -0.267     5.625      11.292
Cadence (steps/min)       -1.08     4.661     -10.216
Stride length (m)         0.006     0.083      -0.157
DS[c] (% of gait cycle)   -0.65     2.345      -5.25
Control subjects
Walking speed (m/min)     1.247     3.154      -4.935
Cadence (steps/min)        1.76     3.146      -4.406
Stride length (m)       -0.0001     0.037       0.073
DS (% of gait cycle        0.08     0.723      -1.34

                                           Product
Group                    Upper              Moment        ICC[a]
                         95% CI     t     Correlation    (2,1)
                                          (test 1/2)    (test 1/2)

Subjects with PD[b]
Walking speed (m/min)    10.758   0.184      0.85        0.85
Cadence (steps/min)       8.056   0.897      0.89        0.88
Stride length (m)         0.169   0.273      0.86        0.85
DS[c] (% of gait cycle)   3.95    1.037      0.72        0.71
Control subjects
Walking speed (m/min)     7.429   1.531      0.96        0.95
Cadence (steps/min)       7.926   2.166[d]   0.94        0.93
Stride length (m)         0.073   0.014      0.98        0.98
DS (% of gait cycle       1.5     4.28       0.99        0.99

[a] ICC=intraclass correlation coefficient.

[b] PD-Parkinson's disease.

[c] DS-double-limb--support phase of gait.

[d] Statistically significant on two-tailed t test, P<.05.

  Table 8.
    Means and Standard Deviations of Change Scores, 95% Confidence
Intervals (CI), t Values, and Correlation Coefficients for
Repeat Measurements Taken at 30-Minute Intervals

                             Change Scores
                                             Lower   Upper
Group                          D   [SD.sub.   95%     95%
                                    diff]     CI      CI
Subjects with PDb
Walking speed (m/min)        0.597   3.743  -6.739   7.933
Cadence (steps/min)          0.813   3.487   6.021   7.647
Stride length (m)            0.008   0.048   0.086   0.102
DS[c] (% of gait cycle       0.039   1.056  -2.031   2.109
Control subjects
Walking speed (m/min)        2.118   2.487   2.757   6.993
Cadence (steps/min)          1.586    2.54   3.392   6.564
Stride length (m)            0.019   0.027  -0.034   0.072
DS (% of gait cycle)         0.011    0.57   1.106   1.128

                                   Product
Group                              Moment            ICC[a]
                           t     Correlation         (2,1)
                                  (test 1/2)       (test 1/2)

Subjects with PD[b]
Walking speed (m/min)    0.617       0.92            0.92
Cadence (steps/min)      0.903       0.93            0.92
Stride length (m)        0.632       0.94            0.94
DS[c] (% of gait cycle   0.139       0.93            0.93
Control subjects
Walking speed (m/min)    3.187[d]    0.97            0.96
Cadence (steps/min)      2.336[d]    0.96            0.94
Stride length (m)        2.71[d]     0.99            0.98
DS (% of gait cycle)     0.07        0.99            0.99

[a] ICC=intraclass correlation coefficient.

[b] PD-Parkinson's disease.

[c] DS-double-limb--support phase of gait.

[d] Statistically significant on two-tailed t test, P<.05.

  Table 9.
    Means and Standard Deviations of Change Scores, 95%
Confidence Intervals (CI), t Values, and Correlation
Coefficients for Repeat Determinations Taken From Day
1 to Day 2 for Subjects With Parkinson's Disease

                            D        [SD.
                                     sub.      Lower      Upper
Group                                diff]      95%        95%
                                                CI         CI

Walking speed (m/min)      0.084     1.801     -3.446     3.614
Cadence (steps/min)       -0.117     7.205    -14.239    14.005
Stride length (m)          0.026     0.088     -0.146     0.198
DS[b] (% of gait cycle    -1.344     3.792     -8.776     6.088

                                   Product
Group                              Moment            ICC[a]
                           t     Correlation         (2,1)
                                  (test 1/2)       (test 1/2)

Walking speed (m/min)     -1.279     0.9             0.88
Cadence (steps/min)        0.065     0.85            0.85
Stride length (m)         -1.21      0.84            0.84
DS[b] (% of gait cycle     1.373     0.8             0.80

[a] ICC=intraclass correlation coefficient.

[b] PD-Parkinson's disease.

  Table 10.
    Means and Standard Deviations of Change Scores, 95%
Confidence Intervals (CI), t Values, and Correlation
Coefficients for Repeat Determinations Taken at Peak
Dose and at End of Dose for Subjects With Parkinson's
Disease
                             D       [SD.
                                     sub.      Lower      Upper
Group                                diff]      95%        95%
                                                CI         CI

Walking speed (m/min)      -14.72    24.109   -61.974     32.534
Cadence (steps/min)        -9.283    17.601   -43.781     25.215
Stride length (m)          -0.207     0.353    -0.899      0.485
DS[b] (% of gait cycle      2.958     7.543   -11.822     17.738

                                    Product
Group                               Moment           ICC[a]
                           t      Correlation        (2,1)
                                   (test 1/2)      (test 1/2)

Walking speed (m/min)      2.115c   -0.51             -0.54
Cadence (steps/min)        1.827c    0.02             -0.07
Stride length (m)          2.024c   -0.26             -0.35
DS[b] (% of gait cycle    -1.359    -0.47             -0.38

[a] ICC=intraclass correlation coefficient.

[b] PD-Parkinson's disease.

[c] DS-double-limb--support phase of gait.

  Table 11.
    Slope and Intercept Values From Linear Regression for
Each Subject in Experiment 4

Sub-     Stride        Stride        Cadence        Cadence
ject     Length        Length        Slope         Intercept
No.      Slope        Intercept

1        -0.001         1.034         -0.008         89.9
2        -0.0002        1.04           0.006        110.5
3        -0.0001        1.093          0.008        113.5
4        -0.0004        0.982          0.003        100.5
5        -0.00002       0.887         -0.009         81.9
6        -0.00009       1.075          0.023        100.9
7         0.001         0.751         -0.09         115.9
8        -0.001         0.968         -0.026        117.7
9        -0.000003      0.997         -0.0003       115.7
10       -0.0003        1.004         -0.0001        96.7
11       -0.00005       0.676          0.151         95.4
12        0.001         0.816         -0.022         110
13       -0.00003       0.689          0.05          95.5
14       -0.00002       1.109         -0.036        108.3
15       -0.0002        0.87           0.051        106.3
16       -0.00002       1.114         -0.012         95.2

         Double        Double
         Limb            Limb
         Support        Support
         Slope           Slope

1         0.037          29.1
2        -0.003          33.9
3         NA[a]           NA
4        -0.001          38.1
5         0.009          49.2
6         0.003          32.1
7         0.03           32.9
8         0.011          33.9
9         0.0001         30.4
10        0.009          30.7
11        0.006          39.9
12       -0.041          41.9
13        0.004          42.3
14        0.003          39.8
15       -0.004          36.9
16        0.008          32.1

"NA: not available.

[FIGURES HAVE BEEN OMITTED]