Gait impairment is frequently a clinically significant feature of Parkinson disease (PD). Parkinsonian gait is characterized by small shuffling steps, stooped posture, and reduced arm swing. As disease progresses, these features worsen, treatment efficacy wanes, and gait impairment becomes increasingly disabling. Loss of lower extremity function has detrimental effects on self-reported motor and psychological quality of life1 and contributes to the deleterious decline in physical activity in persons with PD.2 Furthermore, as gait function worsens, there is frequently a concomitant loss of independence and increased mortality risk.3
For individuals with PD, restoration of walking ability is a primary concern.4 While many aspects of walking behavior garner the attention of clinicians, gait speed has been singled out as an important target symptom because of its relevance to community independence,5 and its predictive value for consequential health outcomes6,7 and mortality.8 Gait speed is often considered a clinical vital sign.9 In PD, gait hypokinesia is significantly related to clinical ratings of disease severity, including Unified Parkinson disease ratings (UPDRS), total and activities of daily living subscores, Columbia University Rating Scale bradykinesia score and UPDRS-III axial motor score,10 activity limitations,11 as well as increased overall disability.12 Furthermore, gait speed is reliably measured in PD12–14 and is a valid measure to predict community walking ability.15
The clinical relevance and value of gait speed as a measure are dependent (at least in part) on its responsiveness; that is, its ability to manifest real changes over time.16 If clinicians are to evaluate the effectiveness of interventions for treating gait impairment in PD, a better understanding of the inherent responsiveness of gait to treatment is required. Certainly, gait speed of an individual with PD can be referenced as a percentage of age- and sex-matched normative values.17 However, reference values that define clinically meaningful changes in gait performance in PD are lacking. In spite of this, gait speed may indeed be the most frequently used primary outcome measure in rehabilitation and pharmacological and surgical studies addressing gait dysfunction. Several indicators of a variable's potential responsiveness include distribution-based statistics such as a minimal detectable change and effect size. Distribution-based approaches rely on the statistical distribution of scores in the measurement of interest.18
Beyond distribution-based methods, responsiveness can be gleaned from standardized response mean and anchor-based statistics, such as the highly cited minimal clinically important difference (MCID). Anchor-based methods explore the relationship between the measurement of interest (in this case, gait velocity) and an independent, clinically relevant measure (or anchor) to explain the meaning of a particular degree of change in the original measure.18 Several studies have used these measures to describe the responsiveness of gait speed,4–6,19–23 but studies have not comprehensively addressed such a diverse population as PD. The MCID serves as a reference value that addresses the question, “How much improvement in gait speed is necessary to achieve a meaningful improvement in the level of disability?”19 Thus, the MCID represents the smallest change in an outcome measure that would be perceived as beneficial. The purpose of this study was to determine the MCID for gait speed using multiple methods of assessment in a large sample of persons with PD representative of various stages of disease severity.
Data were collected from 324 ambulatory individuals diagnosed with idiopathic PD confirmed by a fellowship-trained movement disorders neurologist. Data were collected “on” medication as part of a routine clinical examination at the Center for Movement Disorders and Neurorestoration from April 2011 to March 2012. Persons with deep brain stimulation, other Parkinson-related surgeries, or dementia (Modified Mini Mental State Exam score ≤ 26) were excluded. The participant demographics are presented in Table 1. Some of the participants' data, in part, have been previously reported.24 The study was approved by the University of Florida institutional review board and all participants provided written informed consent prior to data collection.
Gait performance was quantified while walking over the GAITRite instrumented walkway system (CIR Systems Inc, Havertown, Pennsylvania) which was centered within an isolated data collection hallway free from distraction. The 5.8×0.9-m walkway was composed of pressure-sensitive sensors that activated on foot contact with the walkway and deactivated as the foot left the ground. Data were collected at 120 Hz and processed within GAITRite Platinum software.
Participants performed 4 walking trials, beginning and terminating their walks at least 1.5 m before and after the walkway to minimize acceleration and deceleration effects. Participants were directed to walk at their comfortable pace. A verbal cue was provided to begin walking. No other cues were provided, nor was any feedback provided regarding their performance. None of the participants performed the walks using assistive devices. In the event of a freezing episode, additional walks were conducted. All trials in the analyses were free from any episodes of freezing and data from all 4 walks were combined prior to analysis. After completing the walking trials, the neurologist administered part III of the Unified Parkinson disease Rating scale and determined the participant's modified Hoehn and Yahr (H&Y) staging.25 In addition, the Schwab and England (S&E) Disability Scale26 was completed.
We used multiple methods to determine the MCID in gait speed, using distribution-based statistics in conjunction with anchor-based methods. For the distribution-based approach, we tested gait speed to determine the deviation, if any, from a normal distribution. Subsequently, we used the mean and standard deviation measured in the current sample of participants to determine effect sizes relative to 1 standard deviation. The magnitude of the effects was based on the Cohen effect size recommendations whereby we considered 0.2 as small, 0.5 as moderate, and 0.8 as large.27 These effects correspond to the number of standard deviations a value lies away from the mean of the distribution.
We chose 3 commonly used clinical measurement tools to provide anchors for the analyses. Two (UPDRS motor scale and H&Y) are clinician-centric scales. The third (S&E) is based on the patient perspective of the ability to perform activities of daily living. The first step in these analyses was to determine what changes in the clinical measurements (anchors) would represent minimal, moderate, and maximal changes in the clinical measure. We used cutoff points of 2.7 for a minimal difference, 6.7 for a moderate difference, and 10.8 as a large difference for the UPDRS motor score.28 Our data included participants who used only the upper half of the S&E scale (50-100); therefore, we did not use the sample standard deviation but used the cutoff point described by Shulman et al.28 In that study, 10% was considered to represent a moderate effect. For H&Y, the standard deviation within our sample was 0.7; therefore, a change of 1 category within H&Y would constitute a large effect. For the UPDRS gait item (question 29), a 1-point change would also represent a large effect.
Descriptive statistics were first generated for all variables of interest and distributional assumptions tested. Bivariate, zero-order associations between gait speed and the anchor-based statistics were assessed and curve estimation was performed to identify the appropriate follow-up analyses. Curve estimation procedures were performed on the univariate relationships between gait speed and each of the proposed predictor variables to determine the best fit models using linear, quadratic, cubic, and logarithmic functions. Where no differences were noted between model fit parameters, the simplest model was chosen.
General linear models were constructed to calculate means and confidence intervals for gait speed at each level of the predictor variables (H&Y, S&E, UPDRS III motor, UPDRS III-Gait item). Average differences across categories were also calculated. The UPDRS motor score was normally distributed (K-S z score 1.26, P = 0.088) and treated as a continuous variable in follow-up analyses. All analyses were adjusted for age on the basis of quartiles established from the distribution of age within the sample.
Age was significantly associated with gait speed, S&E, H&Y, and UPDRS motor score (r = −0.33, −0.25, 0.30, and 0.26, respectively). Gait speed was significantly associated with H&Y (−0.45), UPDRS (−0.45), and S&E (0.51). In each of the models used, there was no significant interaction noted between age and the predictor of primary interest to this study.
Gait speed was normally distributed (K-S z score 1.22, P = 0.099). The average speed of participants was 0.98 m/s (range = 0.22-1.73 m/s; SD = 0.27 m/s).
When considering the S&E scale, the simplest significant association between S&E and gait speed was linear. Gait speed ranged from means of 0.67 ± 0.24 m/s in participants reporting some dependency (S&E score of 50) to 1.15 ± 0.25 m/s in participants who were completely independent (Table 2). The participants in the 2 highest scored categories (S&E 90 or 100) walked significantly faster than participants in S&E categories 50, 70, and 80. Participants in S&E 50 walked significantly slower than participants in S&E 80, 90, and 100.
For the H&Y, the simplest significant association was linear. Gait speed ranged from a mean low of 0.70 m/s for participants in H&Y stage 4 to a mean of 1.11 m/s for participants with unilateral disease (H&Y stage 1) (Table 3). Participants in H&Y1 and 2 walked significantly faster than those in H&Y 2.5, 3, and 4. Participants in H&Y 2.5 walked significantly faster than those in H&Y 3 and 4.
As with the H&Y, the association between UPDRS motor score and gait speed was also linear. The resulting average age-adjusted change in gait speed per unit change in the UDPRS-motor score was −0.01 m/s, indicating that each 1-point increase in UPDRS motor score could be associated with an average reduction in speed of 0.01 m/s. When considering only the gait item from the UPDRS, gait speed ranged from 0.67 m/s for participants with severe gait disturbance to 1.16 m/s in participants with “normal” gait.
Regression modeling indicated that the average difference across categories of S&E was 0.13 m/s and 0.15 m/s across stages on H&Y. Across each level of the UPDRS gait item, the average change was 0.18 m/s. These coefficients indicate the difference in gait speed expected when considering individuals with PD in different categories of the clinical measurement.
Using distribution-based analyses and effect size metrics, the “small” important difference calculated using the sample variability within gait speed was 0.06 m/s, moderate was 0.14 m/s, and large was 0.22 m/s. Applying established cut-points for small, moderate, and large change in the UPDRS-motor scale (2.7, 6.7, and 10.8 points), the associated differences in gait speed that might be expected were 0.02, 0.06, and 0.09 m/s, respectively.
This study provides initial estimates of the MCID magnitude in gait speed for persons with PD. Clinically meaningful difference has been previously published for values of the UPDRS;18 however, these values have not been particularly useful for the actual changes observed in gait because of the scale's simplicity. The data from this study will be useful for clinicians, therapists, and researchers attempting to evaluate differences in gait speed in response to disease progression and/or intervention. Estimates of meaningful clinical difference have been provided from a large and diverse cohort of participants with PD and are based on sensitive and objective measures of gait performance.
In clinical trials, sample size and performance variability heavily influence the ability to detect a statistically significant improvement. Thus, others have suggested that interventional effects on performance measures can be better interpreted by comparison to an accepted meaningful difference.29 Recently, Tomlinson et al30 evaluated 33 physical therapy intervention trials, including 1518 participants. The authors concluded that physical therapy, when compared with no intervention, was shown to significantly improve gait speed (mean difference of 0.05 m/s). Our results suggest that this difference represents a small but clinically important difference in gait speed.
Recently, Shulman et al31 observed 0.05, 0.08, and 0.01 m/s improvements in comfortable walking speed following high-intensity treadmill exercise, low-intensity treadmill exercise, and stretching and resistance exercise, respectively. Similarly, 6 months of tai chi or resistance training increased gait speed by 0.1 m/s.32 These magnitudes of change approached what would be considered small to moderately important differences. The MCID documented herein can also be used to evaluate effects of short-term interventions. For example, within a single session, transcranial pulsed current stimulation enhanced gait speed by 0.085 m/s,33 which is a small but important difference. Last, our data can also be used to evaluate differences in performance over time that may be reflective of disease progression. Cavanaugh et al34 recently reported that gait speed in the medicated state declined by 0.03 m/s over 1 year in a subset of 33 participants. This magnitude would not be considered a clinically meaningful difference in gait speed. However, levodopa equivalent daily dose increased by 28% during this time period. Thus, one could interpret that gait speed declined in the face of more aggressive pharmacological management.
Many physical therapy or behavioral intervention trials (as described previously) evaluate the efficacy of interventions on gait speed when in the on-medication dopaminergic state, or a typical state for a given individual with PD. Conversely, pharmacological and DBS trials more commonly evaluate the intervention with respect to gait performance in the off-medication state to isolate the size of the effect. Thus, the values used for the meaningful differences reported here are lower than the magnitude of improvements observed typically. Pharmacological treatment using levodopa has been shown to improve gait speed in the range of 0.17 to 0.31 m/s when compared with the off-medicated state.35–38 Reports have also suggested that DBS enhances gait speed by as much as 0.6 m/s from the off-medication state.39–46 Thus, it is important not to extrapolate the difference values reported here to judge the meaningfulness of change in these types of trial.
The values reported here represent differences that are meaningful in the medicated state only. Thus, the change represents decay or additional benefit beyond that of pharmacological management. Within this context, we can evaluate potential effects of long-term DBS on gait. For example, Fasano et al41 reported that after 1 year of stimulation, DBS of the subthalamic nucleus led to a 0.4 reduction on the UPDRS gait item when on medication. Combined with our data, this magnitude of difference would correspond to a 0.07 m/s improvement in gait speed. This magnitude of change would be considered a small important change based on the data presented herein.
Pharmaceutical trials in PD often apply 20% to 30% cutoffs for defining meaningful improvement. Previously, Shulman et al28 documented that a 3-to 5-point change on the UPDRS motor scale represented a minimal to moderate clinically important difference. This difference tended to coincide with the 20% to 30% benchmark that has been considered meaningful. In our sample, a 20% to 30% improvement would necessitate a 0.19 to 0.29 m/s improvement in gait speed. This difference would represent moderate to large improvement. Based on MCID methods, minimal to moderate change more reasonably would resemble a change of 0.05 m/s to 0.15 m/s. A moderate change in UPDRS motor scores (6-point change) would correspond to a 0.06 m/s improvement in gait speed. The discrepancy in these values highlights the difficulty in interpreting gait speed in PD based on select items of clinical scales.
Clinically important differences likely vary across ages, diseases, and disease progression.23,29,47,48 Kwon et al23 reported a minimally meaningful change in gait speed of 0.03 to 0.05 m/s and a substantial change of 0.08 m/s from The Lifestyle Intervention and Independence for Elders Pilot (LIFE-P) multicenter study of sedentary older adults (70-89 years) without PD. Perera et al29 provided estimates of small meaningful change (0.04-0.06 m/s) and substantial change estimates of 0.1 m/s in older adults with mobility disability. Barthuly et al47 reported that an MCID ranged from 0.1 to 0.18 m/s in patients (73.6 years) undergoing short-term rehabilitation following acute care hospitalization. Our data extend the findings of these previous studies by adding benchmarks for persons with PD. As mentioned, in rehabilitation settings, meaningful difference is often considered with respect to improvement due to therapy (behavioral or surgical). However, in degenerative conditions such as PD, meaningful differences should also be considered with respect to impairment.
Another important consideration for the clinically important difference is the range of findings estimated from the 3 different anchors. Clinicians will have to consider the value of a 1 category change on the H&Y scale versus a 6-point change on the UPDRS motor scale. For example, an intervention that leads to an improvement in gait speed of 0.05 m/s is not likely to result in a category change in H&Y or S&E but one may expect some adjustment in UPDRS scores. Similarly, the differences in gait speed associated with a stage difference in H&Y (0.15 m/s) and a 10% difference in S&E score (0.13 m/s) would be associated with a 17.5-point difference in UPDRS motor score. Importantly, using the different scales as the “currency” for comparison may also differentially represent the source of disability.
Our study is not without limitations. The MCID estimates presented may not reflect populations that are different from our study sample, but our large participant sample is representative of the types of patients seen at a tertiary care movement disorder center. The data are from a large cross-sectional cohort rather than a longitudinal evaluation. Gait speed was measured only over the short walking distance across an instrumented mat in an isolated hallway. Previous work by Rolland et al49 has shown that gait speed during a 4-m walk is significantly related to gait speed obtained during a 400-m walk. Furthermore, clinician assessments can be readily arrived at over short distances during routine office visits.
All participants were tested while on anti-Parkinsonian medication. We did not however, standardize where in the medication cycle participants were at the time of gait testing. All participants self-reported that they were adequately responding to their medicine at the time of testing. The 4 walking trials were averaged to represent a single value eliminating any potential influence intertrial variation on the analyses. Previous work has shown intraclass correlation coefficients as high as 0.96 for walking speed collected in multiple sessions, and in pilot work we have detected no significant differences in gait speed among the 4 walks. In addition, differences in gait speed between levels of the S&E (eg, 50%-80%) were not statistically significant. This type of finding is not unique to our study but suggests that primary differences in gait velocity may be reached with larger changes in disability than the 10% change suggested in prior work establishing change in the S&E scale. Finally, it is worth consideration that the MCID is calculated as a single value for the homogenous population, but it is likely that the MCID varies along the course of disease.
As clinical trials of behavioral, pharmacological, and surgical interventions for gait dysfunction in persons with PD are published, it is critical to establish the clinically meaningful difference in gait speed. Our data revealed that the clinically important differences in gait speed among persons with PD in the on-medication state ranged from 0.05 m/s to 0.22 m/s by distribution-based analysis and ranged from 0.02 m/s to 0.15 m/s per level (ie, H&Y stage, S&E) through the use of anchor-based metrics. The data presented in this study serve as an important tool for benchmarking treatment effects in persons with PD “on” medication. Having data on the clinically meaningful difference for gait in PD will aid clinicians in translating and interpreting statistically significant differences reported in large clinical trials.
1. Stewart KC, Fernandez HH, Okun MS, Jacobson CE, Hass CJ. Distribution of motor impairment influences quality of life in Parkinson's disease. Mov Disord. 2008;23(10):1466–1468.
2. Ellis T, Motl RW. Physical activity behavior change in persons with neurologic disorders: overview and examples from Parkinson disease and multiple sclerosis. J Neurol Phys Ther. 2013;37(2):85–90.
3. Lo RY, Tanner CM, Albers KB, et al. Clinical features in early Parkinson disease and survival. Arch Neurol. 2009;66(11):1353–1358.
4. Nisenzon AN, Robinson ME, Bowers D, Banou E, Malaty I, Okun MS. Measurement of patient-centered outcomes in Parkinson's disease: what do patients really want from their treatment? Parkinsonism Relat Disord. 2011;17(2):89–94.
5. Andrews AW, Chinworth SA, Bourassa M, Garvin M, Benton D, Tanner S. Update on distance and velocity requirements for community ambulation. J Geriatr Phys Ther. 2010;33(3):128–134.
6. Friedman PJ, Richmond DE, Baskett JJ. A prospective trial of serial gait speed as a measure of rehabilitation in the elderly. Age Aging. 1988;17(4):227–235.
7. Rabadi MH, Blau A. Admission ambulation velocity predicts length of stay and discharge disposition following stroke in an acute rehabilitation hospital. Neurorehabil Neural Repair. 2005;19(1):20–26.
8. White DK, Neogi T, Nevitt MC, et al. Trajectories of gait speed predict mortality in well-functioning older adults: the health, aging and body composition study. J Gerontol A Biol Sci Med Sci. 2013;68(4):456–464.
9. Studenski S, Perera S, Wallace D, et al. Physical performance measures in the clinical setting. J Am Geriatr Soc. 2003;51(3):314–322.
10. Vieregge P, Stolze H, Klein C, Heberlein I. Gait quantification in Parkinson's disease—locomotor disability and correlation to clinical rating scales. J Neural Transm. 1997;104:237–248.
11. Tan D, Danoudis M, McGinley J, Morris ME. Relationships between motor aspects of gait impairments and activity limitations in people with Parkinson's disease: a systematic review. Parkinsonism Relat Disord. 2012;18(2):117–124.
12. Brusse KJ, Zimdars S, Zalewski KR, Steffen TM. Testing functional performance in people with Parkinson disease. Phys Ther. 2005;85(2):134–141.
13. Lim LI, van Wegen EE, de Goede CJ, et al. Measuring gait and gait-related activities in Parkinson's patients own home environment: a reliability, responsiveness and feasibility study. Parkinsonism Relat Disord. 2005;11(1):19–24.
14. Morris ME, Matyas TA, Iansek R, Summers JJ. Temporal stability of gait in Parkinson's disease. Phys Ther. 1996;76(7):763–777; discussion 778-780.
15. Elbers RG, van Wegen EE, Verhoef J, Kwakkel G. Is gait speed a valid measure to predict community ambulation in patients with Parkinson's disease? J Rehabil Med. 2013;45(4):370–375.
16. Beaton DE, Bombardier C, Katz JN, Wright JG. A taxonomy for responsiveness. J Clin Epidemiol. 2001;54(12):1204–1217.
17. Hass CJ, Buckley TA, Pitsikoulis C, Barthelemy EJ. Progressive resistance training improves gait initiation in individuals with Parkinson's disease. Gait Posture. 2012;35(4):669–673.
18. Norman GR, Sridhar FG, Guyatt GH, Walter SD. The relation of distribution- and anchor-based approaches in interpretation of changes in health related quality of life. Med Care. 2001;39:1039–1047.
19. Tilson JK, Sullivan KJ, Cen SY, et al. Meaningful gait speed improvement during the first 60 days poststroke: minimal clinically important difference. Phys Ther. 2010;90(2):196–208.
20. Kesar TM, Binder-Macleod SA, Hicks GE, Reisman DS. Minimal detectable change for gait variables collected during treadmill walking
in individuals post-stroke. Gait Posture. 2011;33(2):314–317.
21. Hollman JH, Beckman BA, Brandt RA, Merriwether EN, Williams RT, Nordrum JT. Minimum detectable change in gait velocity during acute rehabilitation following hip fracture. J Geriatr Phys Ther. 2008;31(2):53–56.
22. Steffen T, Seney M. Test-retest reliability and minimal detectable change on balance and ambulation tests, the 36-item Short-Form Health Survey, and the Unified Parkinson Disease Rating Scale in people with parkinsonism. Phys Ther. 2008;88(6):733–746.
23. Kwon S, Perera S, Pahor M, et al. What is a meaningful change in physical performance? Findings from a clinical trial in older adults (the LIFE-P study). J Nutr Health Aging. 2009;13(6):538–544.
24. Hass CJ, Malczak P, Nocera J, et al. Quantitative normative gait data in a large cohort of ambulatory persons with Parkinson's disease. PLoS One. 2012;7(8):e42337.
25. Goetz CG, Poewe W, Rascol O, Sampaio C, Stebbins GT, Counsell C, Giladi N, Holloway RG, Moore CG, Wenning GK, Yahr MD, Seidl L (2004). “Movement Disorder Society Task Force Report on the Hoehn and Yahr Staging Scale: Status and Recommendations. The Movement Disorder Society Task Force on Rating Scales for Parkinson's Disease.” Movement Disorders 19(9):1020–1028. doi: 10.1002/mds.20213.PMID 15372591
26. Schwab JF, England AC. Projection Technique for Evaluating Surgery in Parkinson's Disease. In: Billingham FH, Donaldson MC, eds. Third Symposium on Parkinson's Disease. Livingstone, Edinburgh; 1969: 152–157.
27. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale, NJ: Lawrence Earlbaum Associates; 1988.
28. Shulman LM, Gruber-Baldini AL, Anderson KE, Fishman PS, Reich SG, Weiner WJ. The clinically important difference on the Unified Parkinson's Disease Rating Scale. Arch Neurol. 2010;67(1):64–70.
29. Perera S, Mody SH, Woodman RC, Studenski SA. Meaningful change and responsiveness in common physical performance measures in older adults. J Am Geriatr Soc. 2006;54(5):743–749.
30. Tomlinson CL, Patel S, Meek C, et al. Physiotherapy versus placebo or no intervention in Parkinson's disease. Cochrane Database Syst Rev. 2012;8:CD002817.
31. Shulman LM, Katzel LI, Ivey FM, et al. Randomized clinical trial of 3 types of physical exercise for patients with Parkinson disease. Arch Neurol. 2012:1–8.
32. Li F, Harmer P, Fitzgerald K, et al. Tai chi and postural stability in patients with Parkinson's disease. N Engl J Med. 2012;366(6):511–519.
33. Alon G, Yungher DA, Shulman LM, Rogers MW. Safety and immediate effect of noninvasive transcranial pulsed current stimulation on gait and balance in Parkinson disease. Neurorehabil Neural Repair. 2012;26(9):1089–1095.
34. Cavanaugh JT, Ellis TD, Earhart GM, Ford MP, Foreman KB, Dibble LE. Capturing ambulatory activity decline in Parkinson's disease. J Neurol Phys Ther. 2012;36(2):51–57.
35. Bryant MS, Rintala DH, Hou JG, Lai EC, Protas EJ. Effects of levodopa on forward and backward gait patterns in persons with Parkinson's disease. NeuroRehabilitation. 2011;29(3):247–252.
36. Moore ST, MacDougall HG, Gracies JM, Ondo WG. Locomotor response to levodopa in fluctuating Parkinson's disease. Exp Brain Res. 2008;184(4):469–478.
37. Morris M, Iansek R, McGinley J, Matyas T, Huxham F. Three-dimensional gait biomechanics in Parkinson's disease: evidence for a centrally mediated amplitude regulation disorder. Mov Disord. 2005;20(1):40–50.
38. Shan DE, Lee SJ, Chao LY, Yeh SI. Gait analysis in advanced Parkinson's disease—effect of levodopa and tolcapone. Can J Neurol Sci. 2001;28(1):70–75.
39. Carpinella I, Crenna P, Marzegan A, et al. Effect of L-dopa and subthalamic nucleus stimulation on arm and leg swing during gait in Parkinson's disease. Conf Proc IEEE Eng Med Biol Soc. 2007;2007:6665–6668.
40. Chastan N, Westby GW, Yelnik J, et al. Effects of nigral stimulation on locomotion and postural stability in patients with Parkinson's disease. Brain. 2009;132(Pt 1):172–184.
41. Fasano A, Romito LM, Daniele A, et al. Motor and cognitive outcome in patients with Parkinson's disease 8 years after subthalamic implants. Brain. 2010;133(9):2664–2676.
42. Ferrarin M, Rizzone M, Bergamasco B, et al. Effects of bilateral subthalamic stimulation on gait kinematics and kinetics in Parkinson's disease. Exp Brain Res. 2005;160(4):517–527.
43. Kelly VE, Israel SM, Samii A, Slimp JC, Goodkin R, Shumway-Cook A. Assessing the effects of subthalamic nucleus stimulation on gait and mobility in people with Parkinson disease. Disabil Rehabil. 2010;32(11):929–936.
44. Liu W, McIntire K, Kim SH, et al. Quantitative assessments of the effect of bilateral subthalamic stimulation on multiple aspects of sensorimotor function for patients with Parkinson's disease. Parkinsonism Relat Disord. 2005;11(8):503–508.
45. Bastian AJ, Kelly VE, Revilla FJ, Perlmutter JS, Mink JW. Different effects of unilateral versus bilateral subthalamic nucleus stimulation on walking
and reaching in Parkinson's disease. Mov Disord. 2003;18(9):1000–1007.
46. Faist M, Xie J, Kurz D, et al. Effect of bilateral subthalamic nucleus stimulation on gait in Parkinson's disease. Brain. 2001;124(Pt 8):1590–1600.
47. Barthuly AM, Bohannon RW, Gorack W. Gait speed is a responsive measure of physical performance for patients undergoing short-term rehabilitation. Gait Posture. 2012;36(1):61–64.
48. Palombaro KM, Craik RL, Mangione KK, Tomlinson JD. Determining meaningful changes in gait speed after hip fracture. Phys Ther. 2006;86(6):809–816.
49. Rolland YM, Cesari M, Miller ME, Penninx BW, Atkinson HH, Pahor M. Reliability of the 400-m usual-pace walk test as an assessment of mobility limitation in older adults. J Am Geriatr Soc. 2004;52(6):972–976.