We hypothesised that rehabilitation specifically addressing balance in Parkinson’s disease patients might improve not only balance but locomotion as well. Two balance-training protocols (standing on a moving platform and traditional balance exercises) were assessed by assigning patients to two groups (Platform,
Exercise is the foundation of physical rehabilitation in patients with movement disorders of various nature [
Treadmill training is a relatively popular type of exercise, because impaired locomotion is one of the most annoying signs of the disease. This training is easy to administer, can be repeated regularly, and can be practiced at home at the end of the inpatient period with relatively little cost. A recent Cochrane survey, based on data with an overall moderate quality of evidence, confirmed a mild improvement by treadmill training of gait speed and stride length [
One might envisage that selective training of locomotion by treadmill would not be enough for comprehensive rehabilitation of walking but that optimal success could be achieved when balance is also trained [
Here, we trained patients with PD with two different treatments, both specifically addressing balance. A platform onto which subjects stood moved in the anteroposterior, laterolateral, and diagonal direction in the horizontal plane. This platform protocol challenges both the anticipatory and the reactive capacities to the ongoing postural perturbations, thereby training dynamic balance control, aiming at the balance problems encountered during everyday activity. This protocol was based on a simpler moving-platform protocol that had been previously exploited for testing and enhancing balance capacities in patients with PD and with vestibular deficit [
Thirty-eight patients with mild to moderate idiopathic Parkinson’s disease (PD) (Hoehn–Yahr stage between 1.5 and 3) were recruited from the local association of Parkinson’s disease patients and from our laboratory database. All patients had a diagnosis of idiopathic PD based on defined criteria [
Demographics, clinical details, and medication of the 32 patients with PD who participated in the study, divided into the two training groups: PD-E (exercises) and PD-P (platform). The two groups were homogeneous at baseline, as shown by Student’s
Group | Sex | Age (years) | Body weight (kg) | Height (cm) | Duration (years) | MMSE | Hoehn–Yahr | UPDRS motor score | Medication (mg/die) | LED (mg) |
---|---|---|---|---|---|---|---|---|---|---|
PD-E | M | 70 | 76 | 160 | 12 | 27.03 | 3 | 12 | LD (937.5), PR (2.1), RA (1) | 1248 |
M | 75 | 83 | 178 | 4 | 23 | 2 | 20 | PR (3.15), RA (1) | 415 | |
M | 63 | 75 | 160 | 4 | 29 | 2 | 10 | LD (187.5), RA (1), RT (8) | 528 | |
M | 71 | 63 | 167 | 6 | 27 | 2 | 17 | LD (375), PR (0.52), RA (0.5) | 477 | |
W | 68 | 62 | 168 | 11 | 30 | 2.5 | 20 | EN (800), LD (500), RA (0.5), RO (8), RT (4) | 995 | |
M | 81 | 75 | 167 | 8 | 26.4 | 2 | 12 | LD (625), PR (0.52) | 677 | |
W | 58 | 53 | 167 | 11 | 30 | 2 | 21 | EN (800), LD (656.25), PR (1.05) | 978 | |
M | 66 | 69 | 166 | 3 | 27.9 | 2.5 | 11 | LD (437.5), RO (24) | 918 | |
M | 66 | 65 | 172 | 6 | 27.9 | 2.5 | 21 | LD (125), RA (1), RO (8) | 385 | |
M | 71 | 65 | 165 | 12 | 25.4 | 2.5 | 18 | EN (400), LD (250), PR (2.1), RA (1) | 643 | |
W | 51 | 90 | 173 | 10 | 30 | 2 | 19 | LD (312.5), PR (2.1), RA (1) | 623 | |
M | 70 | 76 | 180 | 9 | 30 | 2.5 | 22 | AM (50), EN (600), LD (375), PR (0.52), RA (0.5) | 651 | |
M | 54 | 68 | 167 | 7 | 30 | 2.5 | 31 | RA (1), RO (16) | 420 | |
W | 72 | 63 | 160 | 1 | 30 | 1.5 | 9 | PR (0.52) | 52 | |
M | 69 | 72 | 170 | 5 | 26.9 | 2 | 17 | LD (125), RA (1), RO (10) | 425 | |
M | 80 | 55 | 170 | 8 | 28.7 | 2.5 | 13 | LD (500), PR (1.57), RA (2) | 857 | |
M | 71 | 82 | 168 | 1 | 25.3 | 3 | 24 | PR (1.05), RA (1) | 205 | |
Mean | 68.0 | 70.1 | 168.1 | 6.9 | 27.9 | 2.3 | 17.5 | 617.5 | ||
SD | 8.0 | 9.9 | 5.6 | 3.6 | 2.1 | 0.4 | 5.8 | 307.4 | ||
PD-P | M | 66 | 85 | 179 | 13 | 26.9 | 2.5 | 14 | LD (125), PR (2.1), RA (1), RT (8) | 675 |
W | 67 | 75 | 152 | 10 | 23 | 2.5 | 13 | AM (200), LD (1000), PR (0.26), RA (1), RT (8) | 1566 | |
M | 79 | 62 | 170 | 12 | 23 | 2 | 20 | M (150), EN (800), LD (637.5), PR (2.1), RA (0.5) | 1258 | |
W | 66 | 49 | 170 | 6 | 27 | 2.5 | 19 | PR (3.41), RA (1) | 441 | |
M | 80 | 69 | 160 | 8 | 28.7 | 2.5 | 10 | LD (893.75), RA (1) | 994 | |
M | 67 | 81 | 169 | 3 | 30 | 2 | 14 | LD (500), PR (1.57) | 657 | |
M | 64 | 94 | 190 | 3 | 30 | 2.5 | 22 | LD (875), PR (0.52) | 927 | |
M | 73 | 92 | 172 | 12 | 28.3 | 2.5 | 27 | EN (800), LD (706.25), RA (1), RT (4) | 1159 | |
W | 62 | 60 | 157 | 5 | 30 | 1.5 | 9 | PR (0.52), RA (1) | 152 | |
W | 70 | 60 | 160 | 10 | 30 | 2 | 23 | EN (200), LD (137.5), PR (1.05), RT (8) | 528 | |
M | 75 | 80 | 170 | 10 | 27.4 | 2.5 | 24 | EN (600), LD (375), RO (16) | 819 | |
W | 66 | 65 | 168 | 2 | 27.9 | 2.5 | 41 | RA (1), RO (14) | 380 | |
W | 71 | 50 | 147 | 1 | 27.7 | 2.5 | 17 | RA (1), RT (6) | 280 | |
W | 64 | 75 | 160 | 5 | 30 | 2.5 | 24 | LD (125), PR (3.15), RA (1) | 540 | |
M | 72 | 99 | 180 | 4 | 27.4 | 2 | 24 | EN (600), LD (281.25), RT (6) | 554 | |
Mean | 69.5 | 73.1 | 166.9 | 6.9 | 27.8 | 2.3 | 20.1 | 728.7 | ||
SD | 5.5 | 15.6 | 11.2 | 4.0 | 2.3 | 0.3 | 8.0 | 390.7 | ||
Student’s |
0.56 | 0.58 | 0.69 | 0.92 | 0.38 | |||||
Mann–Whitney’s | 0.89 | 0.33 | 0.28 |
SD: standard deviation; M: man; W: woman; MMSE: Mini-Mental State Examination; UPDRS: Unified Parkinson’s Disease Rating Scale; AM: amantadine; EN: entacapone; LD: levodopa; PR: pramipexole; RO: ropinirole; RA: rasagiline; RT: rotigotine; LED: levodopa equivalent dose [
Patients were randomly assigned to two different groups of training: balance exercise training (PD-E,
Each of the ten sessions was composed of 45 minutes of balance exercises (PD-E) or mobile platform (PD-P) training, each treatment being followed by a 15 min final phase of lower limb stretching, performed with the assistance of a physiotherapist. Sessions were repeated two or three times a week, with at least one rest day between one session and the next, over four successive weeks. Each patient was treated on-phase, at the same time of the day across sessions.
Patients in the PD-E group received a personalized exercise program developed by an expert physiotherapist. There was no predefined duration for each item of the set of exercises, but all patients underwent an overall 45 min period training per day according to the same schedule. This schedule (see Table
Balance exercises administered to the PD-E group, based on the Otago Exercise Program [
Exercise | Description & dose | Progression |
---|---|---|
Tandem | Place a foot straight in front of the other, with the heel touching the toe. |
Difficulty was raised according to patient’s skills. Starting with half-tandem (feet not near together), going on with tandem performed on different surfaces like foam and inclined ramp. |
One leg stance | Look straight ahead. Keep your hands on your hips. Lift on your leg without touching or resting your raised leg upon other standing leg. Stay standing on one leg 30 s long, then switch between one foot and the other. |
Difficulty was raised according to patient’s skills by utilizing different surfaces like foam or keeping one foot lifted up a step. |
Inclined ramp | Stand upon the inclined ramp with toes toward the top. Place feet shoulder-width apart. Maintain the position 45 s long, then turn around in order to have the top by your side. Repeat for each side and one more time with the top behind. | Difficulty was raised according to patient’s skills by closing up feet. Difficulty was further increased by keeping the eyes closed. |
Stance | Place your feet together until almost touching, looking straight ahead. Be as stable and still as possible. |
Difficulty was raised according to patient’s skills by performing exercises on foam surfaces, by closing up the feet or by keeping the eyes closed. |
Compensatory stepping correction | Stand in front of the physiotherapist and lean on his hands. When support is released, make a step to maintain balance. |
Difficulty was increased by keeping the eyes closed. |
(a) Distribution of the exercise subtypes in % of the total duration of the balance training sessions. The data originate from of all patients and all sessions collapsed. OLS: one leg stance. (b) Distribution of the platform perturbation subtypes in % of the total duration of the platform training sessions, all patients, and all sessions collapsed. Each patient was trained with from easy to difficult conditions.
Patients (PD-P) entered the mobile platform and put on a security harness (no weight unloading), which they wore during the entire training session. Their arms were free to move, but they were asked not to reach out for support. Each patient underwent 45 minutes of training (resting periods included), in which from 6 to 8 perturbation patterns were administered, each one lasting about 4 minutes. In order to improve balance control in different directions, in separate trials, subjects stood on the platform with different whole-body orientation with respect to the platform direction of motion. During training, the platform moved in the anteroposterior, laterolateral, and diagonal (45 deg) direction with respect to the body. The periodic platform displacement was 10 cm, regardless of the frequency, which could range from 0.3 to 0.6 Hz. Patients stood with eyes open and closed and feet together or 20 cm apart depending on the perturbation subtype. There was no predefined duration for each subtype of platform perturbation, but all patients were treated according to the same progressive schedule, from easy to difficult, based on the capacity of the patient to withstand the platform perturbation configuration. The actual mean distribution and duration of the sessions is depicted in Figure
All patients of both groups underwent a stretching exercise program as well, as recommended by several guidelines [
At baseline assessment, we recorded the patients’ clinical characteristics (gender, age, disease duration, body weight, and height) and disability (Mini-Mental State Examination, Hoehn–Yahr staging of Parkinson’s Disease, and the motor section of the Unified Parkinson’s Disease Rating Scale) (see Table
These were balance behaviour indexes, assessed by (1) the dynamic balance test on the mobile platform and (2) the Mini-BESTest.
We noted the number of cycles completed by each patient, both at baseline evaluation (T1) and at the end of the 10 training sessions (T2). Also, as an index of the average extent of back and forth displacement of the body segments in the sagittal plane (Index of Stability, IS), the standard deviation (SD) of head and hip markers’ traces along the anteroposterior axis over time was computed [
These were collected by (1) baropodometry and (2) the TUG test.
Seventeen patients were allocated in the PD-E (exercise) group and fifteen patients in the PD-P (platform) group. Sample size was chosen based on data from two previous studies performed in our laboratory employing the continuous platform perturbation as a means of training balance [
For all recorded variables, a test for normality (Shapiro–Wilk) was performed prior to statistical comparison of the differences. To detect differences between the clinical characteristics of the two groups at T1, nonpaired Student’s
The pre- to postrehabilitation differences of the normally distributed variables were assessed by repeated-measure ANOVA, with the groups (PD-E and PD-P) as independent factors. When ANOVA gave a significant result (
The distribution of the number of cycles in the dynamic test proved to be nonnormal by Shapiro–Wilk test. For this reason, we used the Wilcoxon signed rank test to compare total number of cycles between pre- and post-balance treatments within each group. The same was done for the total scores of the ordinal variables (Mini-BESTest, FES-I, and PDQ-8). To assess the difference in these variables between the two patient groups at baseline and after rehabilitation, the Mann–Whitney
Regression analysis was used for estimating relationships among variables, with a focus on the relationship between the value of a variable of interest posttreatment and pretreatment (considered as predictor). This analysis was made for speed and TUG duration. It was also applied to assess the degree of improvement as a function of the medication dosage.
For Mini-BESTest and gait speed, the response rate was the percentage of patients that improved after treatment, estimated by using the minimal detectable change (MDC). Cut-off for determining improvement was based on the values reported in [
Results are reported in the text and figures as mean ± SD. Statistical analysis was performed using Statistica (StatSoft Inc., Tulsa, OK, USA).
Figure
Flowchart for participant inclusion, allocation, evaluations, intervention, and analysis. Abbreviations: MMSE: Mini-Mental State Examination; UPDRS III: Unified Parkinson’s Disease Rating Scale; TUG: Timed Up and Go Test; FES-I: Falls Efficacy Scale-International; PDQ-8: Parkinson’s Disease Questionnaire, 8 items.
Table
Training effects on body stabilization assessed by the moving-platform test in the two groups of patients (PD-E and PD-P) at baseline (T1, yellow columns) and after treatment (T2, pink columns). All subjects were tested at 0.4 Hz perturbation frequency, eyes closed. At T2, patients endured longer periods on the platform than at T1 (a). Head (b) and hip (c) displacement (Index of Stability (IS)) improved significantly after both platform and exercise training, indicating a decrease in body segment oscillation. IS at T2 was better in the PD-P than in the PE-E group for both the head and hip. (d) shows that feet position on the platform was substantially unvarying for patients in both groups. Asterisks (
Figures
Training effect on balance, measured by the total score of the Mini-BESTest. Yellow columns represent pretraining and pink columns posttraining evaluation. A significant difference was found between T1 and T2 within each group (Wilcoxon test;
Analysis of the training effects assessed by the baropodometric and clinical measures collected in the two groups at baseline (T1, yellow columns) and after the treatment (T2, pink columns). Gait speed (a) significantly improved in both groups, while cadence (b) and step length (c) increased only slightly (significantly so in PD-E). Dashed lines indicate the limits of normality. Time to perform the TUG test (d) slightly diminished in both groups; cut-off score for fall risk is indicated by the dashed line. Asterisks (
As to gait cadence (Figure
There was a modest but significant effect of training on step length (ANOVA,
Figure
(a) This shows the correlation between gait speed pre- and posttreatment, as assessed by baropodometry. Red and blue circles represent all single subjects of the PD-E and PD-P groups, respectively. Most data points lay above the identity, indicating increased walking speed in most patients. (b) The patients with a lower gait speed at T1, belonging to both treatments groups, did not show a statistically significant disproportionate improvement after training.
We have checked this by an ANOVA, separately run for the two treatments. For each treatment, the gait speeds at T1 were sorted into slow and fast subgroups. Accordingly, ANOVA showed a significant difference in speed between subgroups (PD-E,
Figure
(a) The scatterplot shows the changes in TUG time (T2–T1) plotted against the TUG time at T1 for each patient of both groups. For most patients, TUG time at T1 was close to the normal values of age-matched healthy subjects. The decrease in time was limited (or absent) in most cases, except for three patients, who improved much their initial performance. (b) Percent changes in TUG time after rehabilitation were not related to the percent improvement of gait speed assessed by baropodometry.
Based on the value of the minimal detectable change (MDC) for the Mini-BESTest, published in [
Figure
The regression lines are drawn through the data points representing the percent changes in gait speed at T2 against medication. There was no effect of total medication (expressed as levodopa equivalent dose) on changes in gait speed, in either treatment group.
For each patient, the percent change in gait speed is plotted as a function of the levodopa equivalent dose. Regression lines were drawn through the data points. For both groups, the lines were almost superimposed, and their slope was not significant (
There was a marginal effect of training on the total score of FES-I (not shown). At baseline, the average score of FES-I was 23.5 ± 7.1 and 26.3 ± 10.7 (Mann–Whitney
At baseline, the average score of PDQ-8 was 5.9 ± 6.1 and 8.9 ± 7.1 (Mann–Whitney
The interplay between balance and gait disorders in Parkinson’s disease is a matter of debate. Poor postural control is common in PD patients, is not always improved by dopaminergic drugs [
Beyond the time-honoured notion that physical activity of different nature and intensity improves activities of daily life in PD [
Here, we studied two populations of patients with PD on stable medication, matched for age and severity of disease. They were trained with two protocols specifically aimed at improving dynamic balance and trunk control. In no case was specific exercises for gait rehabilitation or treadmill training included. One group underwent a program based on classified physical exercises, targeting balance. Another rode a platform, which moved 10 cm back and forth sinusoidally in different directions, a treatment that challenges dynamic balance by forcing subjects to maintain equilibrium and adapt to the continuous postural perturbation [
Protocols partly differ, though. During the exercises, the base of support and the feet position are being deliberately changed by using different materials (solid, foam), inclinations (flat, tilted), features (feet parallel, semi-tandem, tandem, and one leg stance), and intervention (push and release). During the mobile platform protocol, the feet are parallel and their position in relation to the platform does not change. The periodic platform displacement trains adaptation to balance perturbation by a mix of ankle and hip strategies of different relative weight and by learning to anticipate postural adjustments [
Both balance-training protocols decreased the periodic oscillations of the head and hip, as tested by means of continuous and periodic anteroposterior displacement of the support base. The PD-P group reduced head and hip oscillations slightly more than the PD-E group. In particular, the centre of mass (the hip marker) appeared to be better controlled in the patients to whom platform training was administered (PD-P). Most likely, this is connected with the PD-E patients being naïve to the platform, while the PD-P patients were already familiar with it because the treatment was based on a mobile-platform protocol and were likely less susceptible to the startling effects of the platform displacement [
The improvement in locomotion was hypothesised, but not necessarily expected. In both groups, walking velocity increased, suggesting that balance exercises or an instrumental protocol aimed at improving equilibrium has a positive effect on walking speed. The baropodometric findings showed that, at T1, both groups’ mean step length was below the lower limit of normality. At T2, step length increased and rose above the lower limit of normality in both groups. Cadence increased to a limited extent, coherent with [
The time to perform the TUG test also diminished concurrently with the increase in step length in both groups. However, the decrement was modest in both groups. In this connection, Podsiadlo and Richardson [
Even though the period of training was relatively short, there was a slight tendency of reduction of the fear of falling of patients during the execution of activity of daily living, as evaluated with the FES-I scale. Again, no differences between patient groups were found. Similarly, in the PDQ-8 scale that evaluates the health-related quality of life, there was a slight tendency to reduction, without the prevalence of one approach over the other.
All in all, the findings are in keeping with the hypothesis that treatments aimed at improving the dynamic control of balance improve locomotion in PD patients. In a different group of PD patients, Arcolin et al. [
Of note, all our patients were on medication, with the drug dosage spanning in an ample range within and across groups. On the one hand, it is unclear whether medication affects the capacity of the patients to improve in response to rehabilitation. On the other hand, the effect of levodopa on balance and gait is controversial [
All in all, we would therefore argue that dynamic-balance rehabilitation is sufficient for improving locomotion in PD patients. We would also put forward the notion that walking problems in PD depend on, or are very closely related to, balance impairment. This extends to PD the idea that walking velocity is affected by postural instability, very much as it has been suggested for cerebellar and neuropathic diseases [
The relatively unexceptional improvement in the spatiotemporal variables of gait, in both groups, may have been limited by the values of these variables at baseline. Patients recruited in this study had an average Hoehn–Yahr stage of just 2.3, featuring spatiotemporal variables of gait at baseline only slightly off the normal range. The effects on gait of these exclusive balance trainings are based on a small sample size and should be confirmed in a larger cohort, also including more severely affected PD patients. In this line, we would also note that our balance treatments did not overly improve the Mini-BESTest scoring, either, most likely because the patients’ scores at T1 were close to those of normal subjects of the same age.
We do not know whether longer treatments (e.g., [
A four-week balance treatment, not containing any gait-rehabilitation exercises, is sufficient for producing considerable improvement in walking velocity in mildly to moderately affected patients with PD. Although these conclusions are based on small patient numbers, the data are in keeping with the hypothesis that balance control is paramount for locomotion in Parkinsonian patients. One could be even justified to posit that locomotion is degraded in PD
The authors declare that there is no conflict of interests regarding the publication of this paper.
This work was supported in part by “Ricerca Finalizzata” grants (RF-2010-2312497 and RF-2011-02352379) from the Italian Ministry of Health and “PRIN 2010-2011” grants (2010MEFNF7 and 2010R277FT) from the Italian Ministry of University and Research.