INTRODUCTION
Parkinson disease (PD) is a progressive neurological disorder with a rapidly increasing worldwide prevalence.1 Parkinson disease is characterized mainly by 4 cardinal motor symptoms: resting tremor, bradykinesia, rigidity, and postural instability (PI).2 Postural instability describes loss of balance control, therefore often referred to as balance dysfunction.3,4 It leads to loss of mobility, disability, and reduced quality of life.5,6 Recent studies have shown that PI begins early after disease onset,7,8 with around one third of individuals developing PI within the first 2 years after diagnosis.5
Postural instability affects balance control in PD mainly in 4 domains: (1) balance during quiet stance, (2) reactive postural adjustments to external perturbations, (3) anticipatory postural adjustments, and (4) dynamic balance.4 Because of these impairments, people with PD are at a particular risk of falling, with reports ranging from a 2- 9 to 9-fold10 increased risk compared with healthy older adults. Falls occur in approximately 50% of persons with PD, of which about 30% are injurious.11,12 Postural instability has been identified as an independent predictor of falls in persons with PD,13,14 and around one third of falls are caused by an unstable posture.15
Current evidence suggests that PI is poorly responsive to general treatment options in persons with PD, such as medication with L-dopa and deep brain stimulation16; some domains of balance dysfunction even worsen under pharmacological and surgical treatment.4 These findings provide an urgent need for other types of treatment, such as exercise therapy. Tomlinson et al17 demonstrated the beneficial effects of physical therapy interventions (including exercise therapy) on various outcome parameters in participants with PD, including gait, functional mobility, balance, overall disability, and activities of daily living. However, exercise therapy was not the primary focus in this review and the different types of physical interventions were not analyzed on the basis of their specific content. Other recently published meta-analyses focused on specific types of exercise, showing improvements in measures of postural stability in individuals with PD after performing tai chi,18 dancing,19 or aerobic exercises.20 Furthermore, on the basis of their literature review, Canning et al21 recommend to promote exercise for falls prevention in persons with PD, especially challenging balance exercises in a supervised setting. In a meta-analysis from 2011, Allen et al22 summarized the effect of exercise therapy on balance-related activities and falls in people with PD. Results showed an improvement of balance-related outcomes after exercising, and there was a trend for highly challenging balance exercises being most effective. These findings were derived from few preliminary studies in this field, particularly lacking evidence for robust practical recommendations regarding the adequate types of exercise interventions, the optimal level of postural challenge in exercises, and the required dosage (number and duration of sessions) of training. Aggregated research activities in the last years provide a more sophisticated body of literature today.
Therefore, the primary aim of this meta-analysis was to strengthen current evidence and to quantify the effectiveness of exercise therapy on PI in participants with PD. A secondary objective was to evaluate whether specific exercise interventions show superior effects, to identify the optimal mode of exercise therapy for improving postural stability in people with PD.
METHODS
Search Strategy
We conducted a systematic literature search including articles published until the end of January 2015 in the following databases: PubMed (Medline), Scopus, and PEDro (Physiotherapy Evidence Database). The search strategy was developed on the basis of the PICO framework (Population, Intervention, Comparison, Outcome) and was applied to each database with minor adaptations.23 An exploratory search with different combinations of key words was conducted in PubMed to identify the most precise search terms according to the research question. The final search is based on the following strategy: “Parkinson disease” or Parkinson, combined with (and) “exercise therapy” or exercise, combined with (and) balance. For PubMed searches, the following terms were included as MeSH terms: Parkinson disease, exercise therapy, and exercise. In addition, we hand searched reference lists to identify further relevant articles.
Selection Criteria
Three reviewers (S.K., S.S., and S.D.) independently screened identified studies, and inclusion was decided on the basis of consensus, reached by discussion among reviewers. For inclusion, studies had to be randomized controlled trials (RCTs), published in English or German language, and evaluating the effect of exercise therapy on PI in people with PD. Eligible studies had to compare any type of exercise therapy (eg, balance, aerobic, and resistance exercise) against either no intervention or sham exercise (stretching exercises, usual care with no specific exercise therapy, education, or interventions where authors stated that it was intended to be sham treatment). With regard to the secondary aim of this meta-analysis, trials were also accepted if they compared balance exercises against other types of exercise. To avoid heterogeneity and to assure high quality of data, eligible PI measures were defined before analysis. On the basis of a review of relevant literature, we decided about eligibility and priority of PI assessments according to (a) their frequency of usage, and (b) the test quality (reliability and validity).24–27 The ranking was as follows: (1) Berg Balance Scale (BBS),24 (2) postural sway in quiet stance (center of pressure sway area, Sensory Organization Test conditions 1 and 2),25 (3) the Timed Up and Go (TUG) test,26 and (4) the Functional Reach (FR) test.27 Studies were only included in the meta-analysis if they used one of these 4 PI measures. If studies used more than one PI measure, the one with the highest priority was included in the analysis.
Studies were excluded from analysis if the intervention did not primarily focus on performing exercises (eg, whole body vibration without performing exercises), if exercise interventions (experimental or control) were not adequate for meta-analytic comparisons, if data for statistical analysis were missing, or if methodological quality of studies was poor (PEDro score < 5).
Quality Assessment
Assessment of methodological quality of all eligible studies was performed with the PEDro scale.28,29 The scale contains 11 items to assess risk of bias, with 1 criterion (item 1) addressing external validity. This criterion is not included in calculation of the total score and therefore a maximum of 10 points could be reached. A point was allocated to a criterion (yes) if it was clearly satisfied; otherwise, the item was rated as not fulfilled (no). In case of missing information in the publication, authors were contacted via e-mail (sending one reminder). Two reviewers (S.K. and S.S.) independently examined quality rating. Disagreements in ratings were discussed until consensus was found.
Data Extraction and Statistical Analysis
The main characteristics of each included study were extracted, including information about the sample (initial sample size, age, PD status, and dropout rate), intervention (experimental and control), and PI outcomes (assessment and follow-up). Only data assessed immediately after intervention period were extracted; data from longer follow-up periods were not taken into account because of the large variety of intervals. For studies with more than 2 intervention arms, data from several groups were combined if intervention groups were adequate for comparison.30 Authors were contacted via e-mail (sending one reminder) if descriptive information or data for statistical analysis were missing.
To investigate the overall effect of exercise therapy on PI in participants with PD, the primary analysis compared exercise (including various types of exercise) versus no exercise or sham treatment. To compare the effect of various exercise interventions with each other, different secondary analyses were performed. As a first step, subgroup analyses of the primary comparison were conducted. Subgroups were built according to the interventions' level of postural challenge for persons with PD. The rationale for this was based on a previously published framework for balance dysfunction in persons with PD. Accordingly, exercises including (a) balance during quiet stance, (b) reactive postural adjustments to external perturbation, (c) anticipatory postural adjustments in preparation for voluntary movements, and (d) dynamic balance during movements were intended to be most challenging for PI in this population.4
Four subgroups were defined for analysis:
- Balance-specific exercise: Interventions focusing on all 4 domains (a-d) of balance dysfunction in PD (conventional balance exercise),
- Exercise with balance components: Interventions incorporating exercises that challenge the participant's balance (tai chi and dancing),
- Exercise with no specific balance component: Exercises not primarily targeting at PI (aerobic, strength, walking, cycling, and yoga),
- Home-based multicomponent exercise programs: A combination of different types of exercises (eg, flexibility, endurance, strength, and balance), which were performed in a home-based setting.
Another exploratory subgroup analysis of the primary comparison was performed specifically for aerobic training (exercise primarily containing components to improve cardiorespiratory fitness, eg, walking or cycling exercise), on the basis of a framework suggesting that aerobic exercise enhances neuroplasticity and thereby motor performance in persons with PD.31 Our secondary analyses also included a separate statistical analysis evaluating the effect of balance training (including balance-specific exercise or exercise with balance component) when directly compared with other exercise programs.
The Cochrane Collaboration's software Review Manager (RevMan, version 5.3, Copenhagen) was used for the meta-analysis. All outcome variables used for analysis were continuous data and entered as mean values with standard deviation. Standardized mean differences (SMD = Hedge' adjusted g, defined as difference between the posttest means of treatment and control group divided by the pooled standard deviation) with 95% confidence intervals (CIs) were calculated to combine data of outcomes with different units. For scales indicating improvement by decrease (TUG, center of pressure sway area), mean values were adjusted by multiplication with −1.30 When data from multiple studies were available, they were pooled using a random-effects model. Heterogeneity was assessed by using I2 and χ2 statistics and 95% CI. To detect reporting bias for the primary comparison, the relationship between the effect estimates of the interventions and the standard error of these effect estimates was analyzed with a funnel plot.32
In addition, random-effects meta-regression was performed to investigate the influence of different types of exercises on PI. Specifically, we compared the effect of exercises containing balance components (subgroups 1 and 2) to exercises with no specific balance component (subgroups 3). A metaregression analysis was performed with SPSS (version 23, IBM Corp, Armonk) using the MetaReg macro (version 2005.05.23) published by Lipsey and Wilson.33,34
RESULTS
Study Characteristics
The selection process (Figure 1) revealed 22 RCTs35–56 being eligible for the meta-analysis. All included trials were published between 2003 and January 2015. A summary of the main characteristics is presented in Table 1. Sample size ranged from 10 to 195 participants, and mean age varied between 58 years and 73 years. The majority of participants were in mild to moderate disease stages (Hoehn & Yahr [H&Y] 1-3).57,58 Only 6 studies36,38,43,44,49,54 also included severely disabled participants (H&Y ≥ 4) and 5 articles35,48,50,51,53 did not report H&Y stage or only mean scores. In addition, the Unified Parkinson's Disease Rating Scale motor subscale was assessed in 15 trials35,37–40,42,43,45–47,49,51,53,55,56 ranging from 15 to 31. A large variability exists in the exercise interventions of the included studies. In general, these interventions included either a specific form of exercise (eg, balance, tai chi, and dancing) or a combination of various exercise components (typically strength, flexibility, aerobic, and balance). Nearly all of the participants exercised 2 to 3 times per week for 40 to 60 minutes per session. The duration of the intervention period varied widely among the RCTs from 4 weeks up to 2 years, but the majority of exercise therapies were conducted between 10 and 24 weeks. Ten of the included studies reported follow-up periods, which were all short term and ranged from 1 to 6 months.36,38,43,44,48–51,54,56 Except 2 trials,37,45 reported dropout rates were below 20% for short-term follow-up. Among the eligible balance assessments, the BBS and TUG were most frequently reported.
;)
Figure 1: PRISMA flow diagram showing the selection process.
Table 1-a: Characteristics of Included Studies
Table 1-b: Characteristics of Included Studies
Methodological Quality
The majority of studies showed a moderate risk of bias (PEDro score 5-6), and 8 out of the 22 included trials35,36,44,49,50,52,53,56 presented high methodological quality (PEDro score ≥ 7) (Table 2). Because only RCTs were accepted as being eligible, all trials fulfilled the criteria of randomization but only very few reported concealed allocation procedures for group assignment.35,36,44,50,55 Study groups were comparable at baseline in all trials, and point estimates as well as measures of variability were presented. None of the trials reported blinding of participants or care providers, but except for 4 trials38,42,44,48 all assessors were blinded. The majority of articles demonstrated an adequate follow-up, but only half of the 22 trials35,36,38–40,43,44,48–50,52–54,56 conducted their analyses on the basis of intention-to-treat principles.
Table 2: Quality Assessment (PEDro) of Included Studies
Effects of Exercise Therapy
For the primary comparison (exercise vs no exercise/sham exercise), 18 RCTs35–46,49–53,56 with a total sample size of 957 participants were included (Figure 2). The pooled effect estimates revealed a small significant positive effect of exercise therapy on PI (SMD, 0.23; 95% CI, 0.10-0.36; P < 0.001). There was no heterogeneity in the estimates of the effect of exercise therapy on PI in PD (P = 0.71; I2 = 0%).
Figure 2: Forest plot showing the effect of exercise therapy on postural instability compared with no exercise/sham treatment (with subgroup analyses).
Results of subgroup analyses on the basis of specific types of exercise are presented in Figure 2. Two trials38,56 (n = 62) investigated balance-specific training programs, showing a nonsignificant moderate effect on PI outcomes (SMD, 0.43; 95% CI, −0.10 to 0.96; P = 0.11). A subgroup analysis of 6 studies39,41,43,45,46,49 (n = 310) incorporating exercises with balance components showed a significant moderate effect on PI (SMD, 0.43; 95% CI, 0.21-0.66; P < .001). Data from 7 trials37,40,42,50–53 (n = 283) were available for subgroup analysis of exercise with no specific balance component. The pooled estimate of effect sizes revealed a nonsignificant small effect on PI (SMD, 0.20; 95% CI, −0.04 to 0.44; P = 0.11). However, meta-regression revealed that the difference between the effects of these subgroups was not statistically significant (P = 0.253). The pooled SMD of home-based multicomponent exercise programs35,36,44 (n = 302) showed no beneficial effect on PI in participants with PD (SMD, 0.02; 95% CI, −0.20 to 0.25; P = 0.86). No statistical heterogeneity was present in any of the subgroup analyses (balance-specific exercise: P = 0.42, I2 = 0%; exercise with balance component: P = 0.79, I2 = 0%; exercise with no specific balance component: P = 0.80, I2 = 0%; home-based multicomponent exercise programs: P = 0.89, I2 = 0%). The funnel plot (Figure 3) showed that smaller studies had a tendency to produce larger effect sizes, indicating some reporting bias in this comparison.
Figure 3: Funnel plot of SMD and SE (SMD) for primary comparison. SE, standard error; SMD, standardized mean difference.
Four trials37,51–53 (n = 172) investigated the effect of aerobic training compared with no exercise or sham treatment (Figure 4). Pooled effect estimates indicated no beneficial effects of aerobic training on PI in participants with PD (SMD, 0.09, 95% CI, −0.22 to 0.40, P = 0.57). No statistical heterogeneity was observed in this comparison (P = 0.93, I2 = 0%).
Figure 4: Forest plots showing the effect of aerobic exercise on postural instability compared with no exercise/sham treatment and the effect of balance training (balance-specific exercise or exercise with balance component) on postural instability compared with multicomponent exercise programs.
Four studies47,48,54,55 directly compared balance training interventions with other forms of exercise therapy (all multicomponent) (Figure 4). Pooled data (n = 115) revealed a large significant effect on PI in favor of balance training (SMD, 1.31; 95% CI, 0.37 to 2.24; P = 0.006). Tests for heterogeneity suggested a substantial variability among estimates of effects (P = 0.007; I2 = 75%) in this comparison.
DISCUSSION
The aim of this meta-analysis was to strengthen current evidence regarding the effectiveness of exercise therapy on PI in participants with PD and to evaluate whether specific exercise interventions show superior effects. Our findings demonstrate that exercise therapy is an effective treatment option to improve postural stability in persons with PD, and provide first guiding principles for the optimal mode of exercise therapy in this field. Our findings suggest that exercises containing a balance component were most beneficial in improving postural stability in individuals with PD. The minimal and optimal amount of postural challenge during exercise currently remains unclear. Exercises not specifically targeting at balance dysfunction and programs conducted in a home-based setting seem to have no beneficial effect on PI.
Our primary analysis revealed a small positive effect of short-term exercise therapy on PI. These results support preliminary findings from another meta-analysis22; however, the larger amount of underlying data (22 trials; n = 1072) and the high consistency in the included data (low statistical heterogeneity) allow more robust conclusions from the present study. Allen et al22 reported a significant improvement (SMD, 0.33; 95% CI, 0.11-0.55; P = 0.003) of balance-related activities in participants with PD in favor of exercise, but their meta-analysis was based on 15 trials (n = 747) and a moderate level of statistical heterogeneity (P = 0.013; I2 = 47%) in the estimates. Furthermore, subgroup analyses revealed statistical significance only for turning time and gait-related outcomes (mainly gait velocity), but not for PI measures such as BBS, TUG, or FR. In another meta-analysis, Tomlinson et al59 evaluated the effect of physical therapy compared with no intervention/placebo in people with PD and reported significant superior effects of physiotherapy on PI for TUG (9 RCTs; n = 639), BBS (5 RCTs; n = 385) and FR (4 RCTs; n = 193). Tomlinson et al17 also investigated whether the various types of physical therapy show different treatment effects on PI in participants with PD, but no intervention presented superior results. However, they did not focus on the level of postural challenge included in the various physical therapy interventions.
A key feature of the present meta-analysis was the investigation of the most effective types of exercise therapy for improving postural stability in persons with PD. Our subgroup analyses demonstrated the most beneficial effects for exercises incorporating balance components, such as dancing or tai chi. Recently, published meta-analyses specifically investigating the effectiveness of tai chi18 and dancing19 on balance in participants with PD presented similar positive results in favor of these types of exercise therapy. Our findings are also in line with results from Allen et al22 who classified tai chi and dancing as “highly challenging” balance exercise and concluded that these interventions are most beneficial for balance-related activities in participants with PD. However, results from meta-regression revealed that effect estimates of exercise containing balance components (subgroups 1 and 2) and exercise with no specific balance component (subgroups 3) were not significantly different. Thus, there still remains some uncertainty regarding the optimal dosage of postural challenge during exercise. Trials comparing exercise interventions incorporating different amounts of balance components are needed to establish dose-response relationships in this particular target group. A consequent application of the framework suggested by Schoneburg et al4 might provide a methodological basis for a structured motor rehabilitation aiming to improve postural stability in PD.
Results from this meta-analysis revealed no evidence for multicomponent exercise programs being beneficial for PI in individuals with PD. The pooled SMD (subgroup 4, Figure 2) suggests that multicomponent exercise programs are not effective in improving postural stability when performed in a home-based setting. These results support current evidence from Canning et al,21 who conclude that there is only little evidence to promote home-based (minimally supervised) multicomponent exercise programs for the prevention of falls in persons with PD. Furthermore, our secondary analysis (Figure 3) showed that even if multicomponent exercise programs are performed in a supervised setting, they are less beneficial compared with balance training (balance-specific exercise and exercise with balance components). Thus, our findings suggest that multicomponent exercise programs, conducted at home or in a supervised setting, seem to have a reduced potential to improve postural stability in this specific population. Some caution needs to be taken because these findings are based on a small existing body of evidence and the underlying studies included participants with PD who were at risk of falling or had a history of falls. Nevertheless, our results support the advantage of taking task specificity into account as a key physical training principle in motor rehabilitation for people with PD.60–62
In a secondary analysis, we investigated the effect of aerobic exercise on PI in participants with PD. We based the rationale for this analysis on a previously published model by Petzinger et al,63 suggesting that regular aerobic exercise combined with goal-based motor training enhances neural plasticity via improvements in overall brain health and strengthening of neural circuitry. There was no evidence that aerobic training alone is effective in improving PI, compared with no exercise or sham treatment. Our findings are in contrast to a recent meta-analysis, which reported a large significant effect (SMD, 2.02; 95% CI, 0.45-3.59; P = 0.01) of aerobic exercise on PI in participants with PD.20 However, Shu et al20 included exercise interventions that incorporated not only aerobic but also balance components (eg, tai chi and dancing), whereas we only included exercises not specifically targeting at postural stability (eg, stationary bicycling,51 elliptical cross-trainer,53 aerobic walking training37,52). These findings support the idea that aerobic exercise alone is not sufficient in improving motor performance, but in combination with goal-based training (eg, gait or balance training) might be particularly beneficial for motor skill acquisition.63 Whether this holds true, particularly for PI, needs to be confirmed in future investigations.
The main study characteristics varied substantially among the 22 RCTs included in this meta-analysis. This specifically involves the types of exercise applied, duration of interventions, and exercise dosage (number of exercise sessions per week, duration of exercise session), as well as sample characteristics. Regarding the types of exercise therapies, we tried to overcome the present heterogeneity by building subgroups on the basis of the intervention level of postural challenge for persons with PD. The absence of heterogeneity in our subgroup analyses supports this categorization, which was based on a Parkinson-specific framework of balance dysfunction.4 In comparison, Allen et al22 classified balance exercise on the basis of findings from falls prevention in older adults (criteria: movement of center of mass, narrowing of the base of support, minimizing upper limb support).64 These different approaches should be taken into consideration when comparing the results of both studies, as they might lead to differences in interpretation.
The overall duration of interventions also varied widely among the 22 included trials (4 weeks to 2 years), but the majority of participants exercised 12 weeks or less. Therefore, the conclusions from this meta-analysis mainly represent short-term effects of exercise therapy. It remains unclear whether longer intervention periods would lead to comparable or even larger improvements. Studies in this meta-analysis also varied substantially in the intervention dosage, ranging from 1 to 7 sessions per week and from 30- to 90-minute training per session. Thus, it is difficult to provide robust recommendations for exercise prescription on the basis of this meta-analysis. It is noteworthy, though, that most of the trials conducted exercises for 40 to 60 minutes on 2 to 3 days a week, which seems to be an effective dose to improve postural stability in individuals with PD. According to the H&Y scale, only a few trials (n = 6) included participants with PD whose disease was in more advanced stages (H&Y ≥ 4), whereas the majority of studies included participants with PD who were mildly to moderately affected (H&Y 1-3). Thus, the present findings mainly apply for persons in earlier stages of PD, and it remains unclear whether people in advanced stages of PD might show different responses to exercise.
There are limitations to this meta-analysis that should be considered when interpreting the results. To counteract heterogeneity, eligible studies were required to use one of the predetermined PI measures (BBS, postural sway, TUG, FR test). Therefore, we might have omitted trials that also used other appropriate instruments, such as Mini-Balance Evaluation Systems Test (mini-BESTest)65 or Functional Gait Assessment.66 However, literature screening showed that studies using such newer or less widespread PI measures also included traditional tools such as the BBS or TUG, and therefore were also likely to be included in this meta-analysis. Furthermore, anti-Parkinson medication might have affected the results because 1 trial tested their participants during off-state of medication and 6 studies did not control for medication intake. The results of this meta-analysis should be interpreted with caution as they are mainly based on short-term exercise interventions and mainly represent effects in mildly to moderately affected persons with PD.
CONCLUSIONS
Exercise therapy is effective in improving postural stability in individuals with PD, particularly when exercises incorporate components of balance dysfunction. In contrast, interventions that are not specifically targeted at PI, as well as multicomponent exercise programs, seem to have a reduced potential to improve postural stability. Therefore, we highly recommend to integrate balance exercises in motor rehabilitation targeting at improvement of PI in individuals with PD, and to tailor these exercise interventions specifically to the different domains of balance dysfunction in persons with PD. These conclusions are mainly drawn from data of short-duration trials; thus, the long-term effects of exercise on PI remain unclear. An important future step is to evaluate the minimal and the optimal amount and intensity of balance exercises necessary to achieve optimal effects on PI. In addition, future trials should include participants with PD in different disease stages and with varying levels of balance impairment to investigate the potential of exercise therapy on PI throughout the course of the disease. Furthermore, there is an urgent need for long-term follow-up studies that ascertain the sustainability of the beneficial short-term effects of exercise therapy, and their potential to reduce falls in persons with PD.
ACKNOWLEDGMENTS
The study was supported by the Emerging Fields Initiative (EFIMoves project) of the Friedrich-Alexander-University Erlangen-Nürnberg (Germany) and by the Deutsche Stiftung Neurologie. We further thank Ms Lyusyena Novokreshchenova for her support in the initial phase of literature screening.
REFERENCES
1. Wirdefeldt K, Adami H, Cole P, Trichopoulos D, Mandel J. Epidemiology and etiology of Parkinson's disease: a review of the evidence. Eur J Epidemiol. 2011;26(Suppl 1):S1–58.
2. Lees AJ, Hardy J, Revesz T. Parkinson's disease. Lancet. 2009;373:2055–2066.
3. Horak FB. Clinical assessment of balance disorders. Gait Posture. 1997;6:76–84.
4. Schoneburg B, Mancini M, Horak F, Nutt JG. Framework for understanding balance dysfunction in Parkinson's disease. Mov Disord. 2013;28:1474–1482.
5. Kim SD, Allen NE, Canning CG, Fung, Victor S C. Postural instability in patients with Parkinson's disease. Epidemiology, pathophysiology and management. CNS Drugs. 2013;27:97–112.
6. Soh S, Morris ME, McGinley JL. Determinants of health-related quality of life in Parkinson's disease: a systematic review. Parkinsonism Relat Disord. 2011;17:1–9.
7. Mancini M, Horak FB, Zampieri C, Carlson-Kuhta P, Nutt JG, Chiari L. Trunk accelerometry reveals postural instability in untreated Parkinson's disease. Parkinsonism Relat Disord. 2011;17:557–562.
8. Maetzler W, Mancini M, Liepelt-Scarfone I, et al. Impaired trunk stability in individuals at high risk for Parkinson's disease. PLoS One. 2012;7:e32240.
9. Ganz DA, Bao Y, Shekelle PG, Rubenstein LZ. Will my patient fall? JAMA. 2007;297:77–86.
10. Bloem BR, Grimbergen YA, Cramer M, Willemsen M, Zwinderman AH. Prospective assessment of falls in Parkinson's disease. J Neurol. 2001;248:950–958.
11. Rudzińska M, Bukowczan S, Stożek J, et al. The incidence and risk factors of falls in Parkinson disease: prospective study. Neurol Neurochir Pol. 2013;47:431–437.
12. Rudzińska M, Bukowczan S, Banaszkiewicz K, Stozek J, Zajdel K, Szczudlik A. Causes and risk factors of falls in patients with Parkinson's disease. Neurol Neurochir Pol. 2008;42:216–222.
13. Kerr GK, Worringham CJ, Cole MH, Lacherez PF, Wood JM, Silburn PA. Predictors of future falls in Parkinson disease. Neurology. 2010;75:116–124.
14. Pickering RM, Grimbergen, Yvette A M, et al. A meta-analysis of six prospective studies of falling in Parkinson's disease. Mov Disord. 2007;22:1892–1900.
15. Michałowska M, Fiszer U, Krygowska-Wajs A, Owczarek K. Falls in Parkinson's disease. Causes and impact on patients' quality of life. Funct Neurol. 2005;20:163–168.
16. Poewe W. Clinical measures of progression in Parkinson's disease. Mov Disord. 2009;24(Suppl 2):S671–S676.
17. Tomlinson CL, Patel S, Meek C, et al. Physiotherapy versus placebo or no intervention in Parkinson's disease. Cochrane Database Syst Rev. 2012;7:CD002817.
18. Yang Y, Li X, Gong L, Zhu Y, Hao Y. Tai Chi for improvement of motor function, balance and gait in Parkinson's disease: a systematic review and meta-analysis. PLoS One. 2014;9:e102942.
19. Sharp K, Hewitt J. Dance as an intervention for people with Parkinson's disease: a systematic review and meta-analysis. Neurosci Biobehav Rev. 2014;47:445–456.
20. Shu H, Yang T, Yu S, et al. Aerobic exercise for Parkinson's disease: a systematic review and meta-analysis of randomized controlled trials. PLoS One 2014;9:e100503.
21. Canning CG, Paul SS, Nieuwboer A. Prevention of falls in Parkinson's disease: a review of fall risk factors and the role of physical interventions. Neurodegener Dis Manag. 2014;4:203–221.
22. Allen NE, Sherrington C, Paul SS, Canning CG. Balance and falls in Parkinson's disease: a meta-analysis of the effect of exercise and motor training. Mov Disord. 2011;26:1605–1615.
23. Huang X, Lin J, Demner-Fushman D. Evaluation of PICO as a knowledge representation for clinical questions. AMIA Annu Symp Proc. 2006:359–363.
24. Berg KO, Wood-Dauphinee SL, Williams JI, Maki B. Measuring balance in the elderly: validation of an instrument. Can J Public Health. 1992;83(Suppl 2):S7–S11.
25. Ford-Smith CD, Wyman JF, Elswick RK, Fernandez T, Newton RA. Test-retest reliability of the sensory organization test in noninstitutionalized older adults. Arch Phys Med Rehabil. 1995;76:77–81.
26. Podsiadlo D, Richardson S. The timed “Up & Go”: a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc. 1991;39:142–148.
27. Duncan PW, Weiner DK, Chandler J, Studenski S. Functional reach: a new clinical measure of balance. J Gerontol. 1990;45:M192–M197.
28. Maher CG, Sherrington C, Herbert RD, Moseley AM, Elkins M. Reliability of the PEDro scale for rating quality of randomized controlled trials. Phys Ther. 2003;83:713–721.
29. Macedo LG, Elkins MR, Maher CG, Moseley AM, Herbert RD, Sherrington C. There was evidence of convergent and construct validity of Physiotherapy Evidence Database quality scale for physiotherapy trials. J Clin Epidemiol. 2010;63:920–925.
30. Higgins JPT, Green S, eds. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0. The Cochrane Collaboration; 2011.
31. Petzinger GM, Fisher BE, McEwen S, Beeler JA, Walsh JP, Jakowec MW. Exercise-enhanced neuroplasticity targeting motor and cognitive circuitry in Parkinson's disease. Lancet Neurol. 2013;12:716–726.
32. Sterne JA, Egger M. Funnel plots for detecting bias in meta-analysis: guidelines on choice of axis. J Clin Epidemiol. 2001;54:1046–1055.
33. Lipsey MW, Wilson DB. Practical Meta-analysis. Thousand Oaks, California: Sage Publications; 2001.
34. Wilson DB. Meta-analysis macros for SAS, SPSS, and Stata. Version 2005.05.23, 2005.
http://mason.gmu.edu/∼dwilsonb/ma.html. Accessed August 18, 2015.
35. Allen NE, Canning CG, Sherrington C, et al. The effects of an exercise program on fall risk factors in people with Parkinson's disease: a randomized controlled trial. Mov Disord. 2010;25:1217–1225.
36. Ashburn A, Fazakarley L, Ballinger C, Pickering R, McLellan LD, Fitton C. A randomised controlled trial of a home based exercise programme to reduce the risk of falling among people with Parkinson's disease. J Neurol Neurosurg Psychiatry. 2007;78:678–684.
37. Cakit BD, Saracoglu M, Genc H, Erdem HR, Inan L. The effects of incremental speed-dependent treadmill training on postural instability and fear of falling in Parkinson's disease. Clin Rehabil. 2007;21:698–705.
38. Capecci M, Serpicelli C, Fiorentini L, et al. Postural rehabilitation and kinesio taping for axial postural disorders in Parkinson's disease. Arch Phys Med Rehabil. 2014;95:1067–1075.
39. Choi H, Garber CE, Jun T, Jin Y, Chung S, Kang H. Therapeutic effects of tai chi in patients with Parkinson's disease. ISRN Neurol. 2013;2013:548240.
40. Colgrove YS, Sharma N. Effect of yoga on motor function in people with Parkinson's disease: a randomized, controlled pilot study. J Yoga Phys Therapy. 2012;02.
41. Duncan RP, Earhart GM. Are the effects of community-based dance on Parkinson disease severity, balance, and functional mobility reduced with time? A 2-year prospective pilot study. J Altern Complement Med. 2014;20:757–763.
42. Ganesan M, Sathyaprabha TN, Gupta A, Pal PK. Effect of partial weight-supported treadmill gait training on balance in patients with Parkinson disease. PMR. 2013;6:22–33.
43. Gao Q, Leung A, Yang Y, et al. Effects of Tai Chi on balance and fall prevention in Parkinson's disease: a randomized controlled trial. Clin Rehabil. 2014;28:748–753.
44. Goodwin VA, Richards SH, Henley W, Ewings P, Taylor AH, Campbell JL. An exercise intervention to prevent falls in people with Parkinson's disease: a pragmatic randomised controlled trial. J Neurol Neurosurg Psychiatry. 2011;82:1232–1238.
45. Hackney ME, Earhart GM. Tai Chi improves balance and mobility in people with Parkinson disease. Gait Posture. 2008;28:456–460.
46. Hackney ME, Earhart GM. Effects of dance on movement control in Parkinson's disease: a comparison of Argentine tango and American ballroom. J Rehabil Med. 2009;41:475–481.
47. Hackney ME, Kantorovich S, Levin R, Earhart GM. Effects of tango on functional mobility in Parkinson's disease: a preliminary study. J Neurol Phys Ther. 2007;31:173–179.
48. Hirsch MA, Toole T, Maitland CG, Rider RA. The effects of balance training and high-intensity resistance training on persons with idiopathic Parkinson's disease. Arch Phys Med Rehabil. 2003;84:1109–1117.
49. Li F, Harmer P, Fitzgerald K, et al. Tai chi and postural stability in patients with Parkinson's disease. N Engl J Med. 2012;366:511–519.
50. Paul SS, Canning CG, Song J, Fung, Victor S C, Sherrington C. Leg muscle power is enhanced by training in people with Parkinson's disease: a randomized controlled trial. Clin Rehabil. 2014;28:275–288.
51. Qutubuddin A, Reis T, Alramadhani R, Cifu DX, Towne A, Carne W. Parkinson's disease and forced exercise: a preliminary study. Rehabil Res Pract. 2013;2013:375267.
52. Reuter I, Mehnert S, Leone P, Kaps M, Oechsner M, Engelhardt M. Effects of a flexibility and relaxation programme, walking, and nordic walking on Parkinson's disease. J Aging Res. 2011;2011.
53. Sage MD, Almeida QJ. Symptom and gait changes after sensory attention focused exercise vs aerobic training in Parkinson's disease. Mov Disord. 2009;24:1132–1138.
54. Smania N, Corato E, Tinazzi M, et al. Effect of balance training on postural instability in patients with idiopathic Parkinson's disease. Neurorehabil Neural Repair. 2010;24:826–834.
55. Volpe D, Signorini M, Marchetto A, Lynch T, Morris ME. A comparison of Irish set dancing and exercises for people with Parkinson's disease: a phase II feasibility study. BMC Geriatr. 2013;13:54.
56. Yen C, Lin K, Hu M, Wu R, Lu T, Lin C. Effects of virtual reality-augmented balance training on sensory organization and attentional demand for postural control in people with Parkinson disease: a randomized controlled trial. Phys Ther. 2011;91:862–874.
57. Hoehn MM, Yahr MD. Parkinsonism: onset, progression and mortality. Neurology. 1967;17:427–442.
58. Goetz CG, Poewe W, Rascol O, et al. Movement Disorder Society Task Force report on the Hoehn and Yahr staging scale: status and recommendations. Mov Disord. 2004;19:1020–1028.
59. Tomlinson CL, Patel S, Meek C, et al. Physiotherapy intervention in Parkinson's disease: Systematic review and meta-analysis. BMJ. 2012;345.
60. Jöbges M, Heuschkel G, Pretzel C, Illhardt C, Renner C, Hummelsheim H. Repetitive training of compensatory steps: a therapeutic approach for postural instability in Parkinson's disease. J Neurol Neurosurg Psychiatry. 2004;75:1682–1687.
61. Lehman DA, Toole T, Lofald D, Hirsch MA. Training with verbal instructional cues results in near-term improvement of gait in people with Parkinson disease. J Neurol Phys Ther. 2005;29:2–8.
62. Morris ME, Iansek R, Matyas TA, Summers JJ. Stride length regulation in Parkinson's disease. Normalization strategies and underlying mechanisms. Brain. 1996;119(Pt 2):551–568.
63. Petzinger GM, Fisher BE, McEwen S, Beeler JA, Walsh JP, Jakowec MW. Exercise-enhanced neuroplasticity targeting motor and cognitive circuitry in Parkinson's disease. Lancet Neurol. 2013;12:716–726.
64. Sherrington C, Whitney JC, Lord SR, Herbert RD, Cumming RG, Close, Jacqueline C T. Effective exercise for the prevention of falls: a systematic review and meta-analysis. J Am Geriatr Soc. 2008;56:2234–2243.
65. Franchignoni F, Horak F, Godi M, Nardone A, Giordano A. Using psychometric techniques to improve the Balance Evaluation Systems Test: the mini-BESTest. J Rehabil Med. 2010;42:323–331.
66. Wrisley DM, Marchetti GF, Kuharsky DK, Whitney SL. Reliability, internal consistency, and validity of data obtained with the functional gait assessment. Phys Ther. 2004;84:906–918.
