Increased foot strike variability during turning in Parkinson’s disease patients with Freezing of Gait

Abstract

Background:

Turning is a common trigger for freezing episodes in patients with Parkinson’s disease (PD). Freezing during turning can lead to falls and fractures and decreased quality of life.

Research question:

Does foot-strike contact variability also increase during turning, as previously reported in straight gait in PD patients with Freezing of Gait (FOG)?

Methods:

Subjects were instructed to walk on a gait mat, making “normal pivot” (180°) turns at each end. ProtoKinetics Movement Analysis Software (PKMAS) software was used for analysis. Video recordings and foot-pressure-prints were studied to identify and define turn segments. Spatiotemporal gait and turn measures were then determined only for the turn segments. A movement disorders neurologist determined clinical freezes.

Results:

100 subjects (28 controls, 38 noFOG and 34 FOG) were included. Compared to non-freezers (noFOG), FOG subjects had a smaller foot-strike during turning (a measure of completeness of foot contact with the mat) and increased foot-strike variability. FOG subjects also had a shorter stride-length, slower stride-velocity, and greater swing phase time and percentage during turns. After adjusting for turn direction, inner/outer leg dynamics showed heavier inner leg footsteps in FOG subjects. 38% of FOG subjects experienced freezes during turning. 69% of freezes occurred during the middle third of the turn. Turn-freezers had more severe spatiotemporal gait deficits.

Significance:

Developing targeted therapies to retrain subjects to plant their whole foot on the ground with more consistency could help decrease episodes of freezing of gait.

INTRODUCTION

Freezing of gait (FOG) in Parkinson’s disease (PD) occurs in 8% of patients early in disease [1, 2] and up to 92% by the time of death [3]. Gait initiation and turning most commonly provoke freezing [4], and freezing is a major cause of falls [5]. Turning difficulty and freezing have both been shown to significantly impact quality of life in PD [6]. In elderly subjects, falls during turning were 8 times more likely to lead to hip fractures than falls during straight walking [7] and PD alone increases the risk of fracture two fold [8]. During daily activities, we spend somewhere between 8% (walking to a car) to 50% (walking around a cafeteria) of our steps in turning behaviors [9], making turning a very important component of gait function.

Turning is an inherently asymmetric task, and bilateral coordination is impaired in PD FOG patients during straight walking [10], and during turning in smaller versus larger radius turns [11]. PD FOG subjects have also been reported to have increased turn time [12], narrower step width [12] and increasing step time variability with increasing sharpness of turning (90->120->180 degrees) with sharper turns eliciting greater freezing episodes [13]. During community walking freezers have been shown to have smaller turning angles [14] which could therefore elicit greater freezing episodes. Under dual-task conditions, turning speed in FOG and noFOG groups was reportedly similar during 180° turns [15], but slower in FOG during 360° turns [16]. Dual-task also increased step number during both 180° and 360° degree turns and increased cadence and freezing episodes during 360° turns in FOG subjects [16]. Turn duration and step number, have been reported to improve with levodopa more in PD FOG than noFOG patients [17]. Auditory cuing also appears to be beneficial, as it improved freezing [18], decreased cadence [18], and improved turn speed [19].

While there have been many small studies attempting to define features of turning, the ability to define objective spatiotemporal parameters has been limited by the lack of algorithms to define these features using accelerometers or VICON image capture (see Spildooren et al for recent review [20]). As our previous work suggested foot-strike variability was more impaired in freezers compared to non-freezers during self-paced, straight walks [21], we hypothesized that foot-strike would also be more impacted during turning in FOG. To test this hypothesis, we monitored gait during 180° turns from walking, using an instrumented gait mat time locked to a two-camera video capture system, in 100 subjects.

METHODS

People with Parkinson’s disease (PD) by UK brain bank criteria were recruited from the University of Arkansas for Medical Sciences (UAMS) Movement Disorders clinic, with family members asked to participate as age-matched controls (controls). Written informed consent was obtained after institutional review board approval (UAMS IRB# 203234) and in compliance with the Declaration of Helsinki. FOG presence was based on subjective report (Item 3 on the Freezing of Gait Questionnaire (FOG-Q) ≥1), and/or witnessed by a Movement disorders Neurologist (TV) during examination. Exclusion criteria included a Montreal Cognitive Assessment Score (MoCA) score <10, >1 fall/day and dopamine-receptor-antagonist medication use in the year prior to enrollment. One hundred subjects were recruited; 72 PD and 28 controls.

PD subjects were assessed in their usual levodopa-medicated state. The daily equivalent levodopa dose was calculated as previously described [22]. Subjects were administered a complete Unified Parkinson’s Disease Rating Scale (UPDRS) assessment (by T.V.), the FOG-Q [23], the Montreal Cognitive Assessment (MoCA), the Frontal Assessment Battery (FAB), and the Scales for Outcome in Parkinson’s Disease - Cognition (SCOPA-COG). We used the FOG-Q total score, a FOG-Q sub-score (sum of items 3–6 documenting freezing severity) and FOG-Q individual item scores, for correlation analysis.

Gait analysis:

Gait dynamics were collected using a 20’x4’ Zeno-walkway (Protokinetics, Haverton, PA) with two synchronized video cameras, with ProtoKinetics Movement Analysis Software (PKMAS). Subjects were instructed, using similar verbiage and visual demonstration, to walk at their comfortable pace and make “normal pivot” turns, at or before traffic cones placed approximately 1 foot from each end of the mat. Guidance was not given on the direction to turn. Walking 8 lengths of the mat, the goal was to obtain 7 turns each. Of 707 turns analyzed, only 8 turns with incomplete or absent footprints and 2 turns due to subject talking while turning, were excluded from analysis (1 control, 3 noFOG and 6 turns in 4 FOG subjects). When immediately identified, subjects were asked to perform additional turns to obtain 7 analyzable turns.

PKMAS’s algorithms provided initial footprint identification. Most turn segments required manual processing to mark unidentified or incorrectly identified footsteps. The start and end of each turn segment was identified based on an algorithm developed on control subjects, described in detail in supplementary methods. Straight-walk segment footprints were not included in the analysis.

The mean and percent coefficient of variation (CV), for each of the “spatiotemporal gait measures” defined in Table 1 were obtained for each subject using combined data from all turns. Footstep location and contact times were used to calculate the “turn measures” defined in Table 1. Turn measures were calculated for each turn and averaged to obtain subject values.

Table 1 –

Spatiotemporal gait and turn measure – sources and definitions

Measure	Source	Definition
Spatiotemporal gait measures:
Foot-strike-length (cm)	PKMAS	length of the major axis of the ellipse enclosing each footstep; PKMAS creates an ellipse around each footstep during footstep identification.
Foot-strike-width (cm)	PKMAS	length of the minor axes of the ellipse enclosing each footstep
Foot-strike-area (cm2)	PKMAS	the area of the ellipse using the major and minor axes as the radii.
Foot-strike-length-percent	PKMAS	a percentage measure of foot length relative to the longest foot length measured for all selected footsteps, for each foot.
Integrated-pressure (pressure x s)	PKMAS	the sum of pressure applied by a footstep at each sampling time (120 Hz sampling rate) in the area of its contact with the ground.
Stride-length (cm)	PKMAS	the distance between heel strikes of two consecutive footsteps of the same foot, i.e., two right or two left heel strikes.
Stride-time (s) also known as gait cycle time	PKMAS	the time difference (s) between the initial heel contacts with the mat of two consecutive footsteps of the same foot, i.e., two right or two left.
Stride-velocity (cm/s)	PKMAS	the stride length divided by stride time, calculated for each gait cycle.
Stance-time (s)	PKMAS	the time difference between the first and last contact of each footstep plus the sampling time and represents the time duration of foot contact with the ground.
Swing-time (s)	PKMAS	the difference between gait cycle time and stance time and represents the time duration when the foot is off the ground.
Stance-percent	PKMAS	percentage measure of time spent in stance phase of the gait cycle and calculated as stance time/gait cycle time.
Swing-percent	PKMAS	a percentage measure of time spent in swing phase of the gait cycle and calculated as stance time/gait cycle time.
Stance-COP-distance (cm)	PKMAS	stance Center of Pressure distance; the Pythagorean distance between the first and last contact points of the center of pressure (COP) waveform trail for a footstep in stance phase, based on X and Y coordinates corresponding to first and last contact times of a foot during stance.
Cadence (steps/minute)	PKMAS@	((total footsteps in each turn −1)*(total number of turns))/cumulative time difference (in minutes) between the initial contacts of first and last footstep each turn
Turn measures:
Turn-time (s)	author defined	the difference in time# between the first contact of the last normal-angled, pre-turn footstep, and the last contact of the first normal-angled, post-turn footstep.
Turn-length (cm)	author defined	the distance (difference) between the maximum and minimum value of the X coordinates# of foot placement on the mat.
Turn-width (cm)	author defined	the distance (difference) between the maximum and minimum value of the Y coordinates# of foot placement on the mat.
Turn-rectangular-area (cm2)	author defined	(π*[turn length/2]*[turn width/2])
Turn-ellipse-area (cm2)	author defined	(π*[turn length/2]*[turn width/2])
Step-count (no.)	author defined	sum of right and left steps in each turn segment

Subjects wore their preferred, comfortable footwear. All “PKMAS” calculations were auto-generated by the PKMAS software and definitions based on our understanding of the PKMAS Measurements and Definitions manual Protokinetics provided with the software;

^@PKMAS definition altered to represent that this was auto-generated by PKMAS for turn segments;

^#time stamps and mat location coordinates for each footstep were auto-generated by PKMAS; all “author defined” turn measure calculations were done in Microsoft Excel.”

When turning leftward, the left foot is on the inner side of the turn and vice-versa for rightward turns. In order to study inner and outer foot dynamics we reclassified the “right” and “left” footsteps to “inner” or “outer” footsteps based on each turns direction. 10/38 noFOG and 11/34 FOG subjects had at least one turn requiring adjustment of the inner/outer legs. Ratios of the inner/outer gait measures were then calculated.

Using video footage, a Movement disorders Neurologist (T.V.) identified clinical freezing episodes during turn segments, and confirmed them with presence of a prolonged pressure trace. All visualized freezes met the pressure trace criteria. Each freezing episode was further classified as being in the inner, outer, or both feet (by L.P. and T.V.).

Statistical Analysis:

Statistical analyses were performed with SPSS Version 24 (IBM). Shapiro-Wilk was used to test normality of each measure by group. Significance of differences was assessed with one-way ANOVA for data that showed normal distribution and passed Levene’s test for homogeneity of variances. Kruskal-Wallis test (3 groups) or the Mann-Whitney U-test (2 groups) were used to assess significance for nonparametric data. A post-hoc Bonferroni correction was applied for all multiple comparisons and adjusted p-values reported. The test used for each demographic variable is indicated in the Tables. The ratio of freezing episodes to total turns was calculated for each FOG subject and used to test for Spearman’s Rank Order correlations with motor and non-motor features related to freezing. A linear step-wise backwards multivariate regression model, with FOG-Q as the dependent variable, was used to determine the gait and turn measures most related to freeze severity. A binary logistic regression model was applied to determine the gait and turn measures that best categorized PD freezers or non-freezers. Effect sizes (Cohen’s d) for group differences were calculated and provided in Supplementary tables.

Data availability statement:

As subjects in this study are participating in an ongoing longitudinal study, anonymized data sets can only be shared at the request of a qualified investigator.

RESULTS

One hundred subjects were enrolled in three groups - 28 controls, 38 PD without freezing of gait (noFOG), and 34 PD with freezing of gait (FOG), >95% of whom were Caucasian. FOG subjects were of the same age, had higher H&Y scores, UPDRS motor and total scores, longer disease duration, were on higher total daily levodopa dose and scored lower on the SCOPA-Cog and FAB compared to noFOG (Table 2). The time interval between last levodopa dose to gait assessments was not different between the PD groups (Table 2).

Table 2 –

Demographics and basic disease characteristics of subjects included in analysis

	Results			Statistics
	controls (n=28)	PD no-FOG (n = 38)	PD FOG (n = 34)	controls vs. PD no-FOG	controls vs. PD FOG	PD no-FOG vs. PD FOG
Age (years)#	62.45 ± 9.08	66.58 ± 6.30	67.45 ± 9.51	n.s.	0.034	n.s.
Gender (percent female)	60.71%	47.37%	32.35%	-	-	-
Right handed	93.10%	89.47%	91.18%	-	-	-
Disease duration (years)#	-	7.21 ± 5.98	9.18 ± 5.12	-	-	0.029
Hoehn & Yahr score#	-	1.80 ± 0.54	2.49 ± 0.62	-	-	0.000
UPDRS Part III Score (motor)#	1.54 ± 1.65	12.34 ± 5.26	18.68 ± 9.45	0.000	0.000	0.010
Total UPDRS Score#	3.46 ± 2.47	22.09 ± 7.56	35.88 ± 13.53	0.000	0.000	0.000
FOG-Q score#	0.07 ± 0.38	2.34 ± 2.59	11.03 ± 4.53	0.000	0.000	0.000
MoCA score#	27.36 ± 1.75	25.87 ± 3.16	23.82 ± 4.08	n.s.	0.001	n.s.
FAB score#	16.5 ± 1.43	16.03 ± 2.49 (n = 36)	13.97 ± 2.91	n.s.	0.001	0.006
SCOPA-Cog score*	29.07 ± 4.7	24.94 ± 4.87 (n = 36)	20.94 ± 4.97	0.003	0.000	0.003
Subjects with falls	-	21.05% (8/38)	47% (16/34)	-	-	-
Fall frequency (/month)#	-	0.09 ± 0.25 (n = 8)	1.91 ± 5.75 (n = 16)	-	-	0.018
Total daily levodopa dose (mg)#	-	477 ± 345 (n = 33)	832 ± 537 (n = 32)	-	-	0.003
Last levodopa dose (mg)#	-	150 ± 65	215 ± 130	-	-	0.024
Time to last levodopa dose (hrs)#	-	2.79 ± 2.35	1.89 ± 1.16	-	-	n.s.

^*ANOVA;

^#Mann-Whitney U. Bonferroni adjusted p-values reported for all multiple comparisons.

Combined foot analysis:

The foot-strike-length-percent, a measure of completeness of foot contact with the mat, was significantly lower in FOG compared to both noFOG and controls. FOG subjects also took significantly shorter (vs noFOG) and wider (vs controls) turns (Figure 1, Supplementary Table 1) but the overall rectangular and elliptical-turn-area, was similar between all groups. FOG subjects took the greatest number of steps and longest time to turn (FOG>noFOG>controls). Stride-length, stride-time and swing-time were shorter, and stride-velocity slower in FOG compared to noFOG (Figure 1, Supplementary Table 1). Turn-width and stance-COP-distance CV were greater in FOG compared to noFOG (Figure 1, Supplementary Table 1).

Figure 1:

Combined foot analysis of gait and turn measures during turning in PD subjects with (FOG) and without (noFOG) freezing of gait as the percentage difference from results in controls. The zero percent line demarcates controls. Mean percent difference +/− standard error in mean are shown in the graphs.

Inner and outer leg analysis:

We analyzed the inner and outer leg gait measures for PD subjects based on turn direction. Other than inner leg stance-COP-distance CV not being statistically significant, the results were the same as the combined foot analysis shown in Figure 1, and are therefore not shown again. Since asymmetry has been suggested to play a role in freezing we took the ratio of the inner/outer leg measures and found that FOG subjects applied greater integrated-pressure on the inner compared to outer leg than noFOG (noFOG: 0.90 ± 0.02; FOG: 0.98 ± 0.02; p=0.006; Supplementary Figure 1, Supplementary Table 2).

Freezing episodes:

38% (13/34) FOG subjects experienced freezing during at least one turn on the mat (FOG turn-freezer); a combined total of 19.7% (50/254) turns had freezes. The distribution of freezes for each FOG turn-freezer is shown in Supplementary Table 3. 69% of the freezing episodes occurred in the middle third of the turn and 9% and 22% occurred in the first or last third, respectively. Outer leg freezing (54.9%) was more prevalent than inner leg (33.3%), or both legs (11.7%) freezing. Turn-freezers had a more severe gait and turn phenotype with shorter, slower steps, increased step number, with decreased foot contact area and length and decreased pressure applied over that area, with less proportional time spent swinging the foot (Figure 2, Supplementary Table 4). FOG-Q score was significantly different between the two groups (Supplementary Table 5).

Figure 2:

The gait and turn measures of the subgroup of FOG subjects that had witnessed freezing during turns (turn-freezer; n=13) are shown as the percent difference from the subgroup that did not have freezing during of turns (turn-non-freezer; n=21). The zero percent line demarcates the FOG turn-non-freezer subgroup. Mean percent difference +/− standard error in mean are shown in the graphs.

Total daily dose of levodopa showed a weak-positive correlation (r_s(32)=0.362, p=0.035), while the amount of, and time from the last levodopa dose showed no relation to presence of freezing episodes. FOG-Q score and FOG-Q sub-score (for both r_s(32)=0.524, p=0.001) and all FOG-Q individual item scores (r_s(32) range: 0.423 to 0.588, all p<0.02), except item 2, had a moderate positive correlation with witnessed freezing during trials. Presence of falls in the prior 3 months (r_s(32)=0.431, p=0.011) and fall frequency/month (r_s(32)=0.353, p=0.041) were also weakly-moderately positively related to presence of witnessed freezing. Age, gender, disease duration, cognitive scores, visuospatial cognitive sub-scores, and motor and total UPDRS scores did not show any relation to presence of freezing episodes. Mean stride-length (r_s(32)=−0.641, p=<0.001), foot-strike-length-percent (r_s(32)=-0.601, p=<0.001), and swing-time (r_s(32)=−0.602, p=<0.001) all showed strong negative correlation with the presence of freezing episodes.

A multivariate regression model with FOG-Q as the dependent variable and gait and turn measures as independent variables was built in a step-wise manner to avoid collinearity in measures (Supplementary Table 6). In the final model, mean stance-COP-distance had the greatest significance (p<0.001) followed by the mean foot-strike-width and turn ellipse-area. A binary logistic regression was also performed to determine which features best-differentiated freezers from non-freezers (Supplementary Table 7). The final model had an overall 76.4% predictive value, and mean stance-COP-distance (p=0.003) again had a high statistical significance, along with mean integrated-pressure, and stance-time.

DISCUSSION

In this study we explore in detail the dynamics of turning in a relatively large population of 72 PD subjects (34 FOG, 38 noFOG) compared to prior studies (see Spildooren et al., 2018 [20] for review). We find that foot-strike contact (as measured by foot-strike-length-percent and stance-COP-distance) was shorter and more variable in PD FOG compared to noFOG subjects. Stance-COP-distance was also related to freezing severity and an important feature in categorizing freezers and non-freezers in regression models. Due to the comprehensive nature of our evaluations using an instrumented gait mat, we had other novel findings that are discussed below.

Basic turn parameters

FOG subjects, while ON levodopa, took significantly more steps, took longer to turn and had a greater cadence (steps/min) than noFOG. This is consistent with previous studies that showed increased turn time in FOG both OFF [15–18, 24, 25] and ON levodopa [19, 26], increased turn steps both OFF [16–18, 25] and ON levodopa [26] and greater cadence OFF [16, 18, 25] and ON levodopa [13]. While turn-to-turn variability was increased in our PD group compared to controls, there were no significant differences between the FOG and noFOG groups.

Objective spatiotemporal gait parameters

As we had hypothesized based on our previous work in straight walking [21], FOG subjects had significantly shorter foot contact with the ground (measured as reduced stance-COP-distance and foot-strike-length-percent) than noFOG and controls (Figure 1), and increased foot-strike variability (measured as increased stance-COP-distance CV) during turns. This suggests that FOG subjects have a smaller base of foot contact to the ground during turns, which could contribute to increased instability. The increased foot-strike variability could be due to an increase in behaviors such as sliding on toes, toe/heel touches, pivoting on heels, and shuffling feet that were observed on video but difficult to quantify. This increased variability could also contribute to freezing. As there was no significant difference in the integrated-pressure applied to the mat between the groups (which measures pressure over the total area), with shorter foot contact, FOG subjects place heavier footsteps in a smaller area. This finding could possibly explain the often described sensation of the “feet sticking to the ground”.

Compared to noFOG, subjects with FOG also showed shorter stride-length and stride-time, slower stride-velocity, and decreased absolute and proportional time spent in the swing phase during turning. Using the ratio of inner/outer leg to determine asymmetry, only mean pressure applied on the inner leg was significantly higher in FOG compared to noFOG. This is in contrast to work by Petersen et al. [11] in which they calculated asymmetry differently than us, as a phase coordination index of the relative timing of stepping of one leg compared to consecutive steps of the other leg. While we did not repeat their calculations, stride-time variability was approximately 25% higher in our FOG group, but due to inter-subject variability was not statistically significant (p=0.883).

Based on the sequence-effect model of freezing [27], we previously showed that stride-length successively shortened in FOG compared to noFOG subjects prior to entering a 180° turn, especially when freezing occured during turning [28]. In support of this model, we now show that during turning, stride-length is shorter in FOG compared to noFOG, and in FOG turn-freezers mean stride-length was shorter and CV stride-length was higher than in FOG turn-non-freezers. Freezing severity (as measured by FOG-Q scores) was also inversely correlated with turn stride-length and foot-strike-length-percent similar to that seen in straight line gait [21].

Turns and witnessed freezing events:

FOG subjects applied greater pressure to the inner leg when turning (Figure 2). Despite this, the outer leg appeared to have greater proportion of freezing (approx. 60% to 40%; outer vs inner) in contrast to a prior study that showed more freezing on the inner leg (35% to 65%; outer vs inner) [18]. The difference could be accounted for by the experimental protocol; while we studied 180° single-task turning, Spildooren et al. [18] combined 180° and 360° turns under both single and dual-task conditions, with turns required in both directions.

The majority of witnessed freezing occurred during the middle 1/3 of the turn (69%), which would make sense, as during a 180-degree turn, the maximal angular changes occur during that turn segment. In comparison, a prior study showed the majority (47.6%) of freezing in the last turn quarter, with 14.3% and 28.6% in the 2^nd and 3^rd quarters respectively [12]. The differing results could be due subjects in our study turning at a marker placed on the side as opposed to turning around a marker and turning in random rather than forced bilateral directions, in addition to the larger number of turns with freezing we analyzed (48 vs 9).

In our study the FOG turn-freezers subgroup took significantly more steps and time to complete turns, had significantly shorter steps and decreased swing phase percent and had shorter foot-strike length. 46% of our FOG turn-freezers reported daily freezing episodes compared to 10% of FOG turn-non-freezers. The FOG turn-freezers subgroup also had more individuals reporting more frequent and longer duration freezing based on the FOG-Q individual questions scores. Perhaps future studies could use items 5 and 6 of the FOG-Q to select individuals with higher propensity for freezing in research settings.

Limitations:

Patients in our study were in the levodopa-medicated state. However, the effect of levodopa on turn parameters is unclear as studies have reported both improvement [29], and no improvement [17] in step-number and turn time while making 180° turns in place. Even in the levodopa medicated state 38% of our FOG group had freezing episodes during “normal” turning which is comparable to other studies performed in the OFF-levodopa state (31–38%) [12, 30]. We also included freezing episodes in the analysis, as this is the real world situation for these patients when they are walking at home. This could partly account for the differences seen between the noFOG and FOG groups as the turn-freezers had a more severe gait phenotype than turn-non-freezers. However, in our cohort 5/13 turn-freezers had only 1–2 freezes and only 3/13 had freezes in each of the 7 turns, which would suggest less of an impact.

Conclusions:

In our study, FOG subjects had a smaller foot-strike contact and increased variability in foot-strike to the ground while turning. In addition, they took shorter, slower strides with decreased swing-time and applied more pressure to the inner turn leg. These parameters were more affected in subjects with witnessed freezing, suggesting that they could contribute to the freeze event. Freezing episodes were most common in the middle third of the turn when maximum angular changes are occurring. These turn features could be potential targets for intervention during physical therapy. Future investigation into the force trajectories of foot-strike could better define no-FOG and FOG differences in foot placement. Determining in the future, which changes occur earliest could also help define cohorts more prone to severe gait freezing.

Highlights.

FOG subjects had shorter mean foot-strike contact during turns

• FOG subjects had greater variability in foot-strike contact during turns

• FOG subjects had shorter stride-length and slower stride-velocity during turns

• FOG subjects placed greater weight on the inner leg during turns

• Freezing occurred most commonly (69%) during the middle third of the turn

Glossary of Terms:

COP

center of pressure

CV

Coefficient of Variation

FAB

Frontal Assessment Battery

FOG

freezing of gait

FOG-Q

Freezing of Gait Questionnaire

MoCA

Montreal Cognitive Assessment Score

noFOG

no freezing of gait

PD

Parkinson’s disease

SCOPA-COG

Scales for Outcome in Parkinson’s Disease - Cognition

UPDRS

Unified Parkinson’s Disease Rating Scale