Website Edition: December 2009/January 2010

Editor's Choice Article

Movement Disorders
Volume 24, Issue 13, Pages 1934-1940
Published Online: 11 August 2009

Physical assessment as a predictor of mortality in people with Parkinson's disease: A study over 7 years
William K. Gray, PhD¹, Anthony Hildreth, MPhil¹, Julie A. Bilclough, MSc², Brian H. Wood, MD², Katherine Baker, PhD¹, Richard W. Walker, MD² *

¹School of Health, Community and Education Studies, University of Northumbria, Coach Lane Campus, Newcastle-upon-Tyne, United Kingdom
²Department of Medicine, North Tyneside General Hospital, North Shields, Tyne and Wear, United Kingdom

E-mail: Richard W. Walker (

*Correspondence to Richard W. Walker, Department of Medicine, North Tyneside General Hospital, Rake Lane, North Shields, Tyne and Wear, NE29 8NH United Kingdom

Potential conflict of interest: None reported.

Funded by:
Northumbria Healthcare NHS Foundation Trust
Parkinson's Disease Society
University of Northumbria
U.K. National Health Service


The primary aim of this study was to ascertain whether a battery of physical function measures in a Parkinson's disease (PD) patient cohort predicted mortality status at 7-year follow-up. Secondary aims were establishing which specific tests were the most useful, and whether PD phenotype was a predictor. A retrospective correlation design was used in this study. A cohort of 109 PD patients underwent baseline physiotherapy assessment of gait, balance, posture, muscle strength, and ability to change postural set. We compared mortality status at 7-year follow-up and baseline physical assessment tests. Tinetti gait and balance scores, UPDRS score, 10-m walk test (time, velocity, and number of strides), posture in standing, lying to sitting, sitting to standing, getting up from floor assessments, and time to ascend and descend four steps were found to be statistically significant physical predictors of mortality at 7-year follow-up. In addition, age, sex, and mini-mental state examination were significant nonphysical predictors of mortality. Using Cox regression, a survival model was constructed with age, sex, and Tinetti gait score as independent predictors of mortality. The results of this study suggest that there is a link between reduced physical function and an increased mortality risk in PD populations.

© 2009 Movement Disorder Society

Received: 19 January 2009; Revised: 30 March 2009; Accepted: 4 April 2009

Article Text

A number of primary studies and reviews have looked at predictors of disease progression in Parkinson's disease (PD).[1-5] These studies found strong evidence that a rapid disease course could be predicted by higher age at onset, but there was limited evidence for dementia, higher bradykinesia score, nontremor dominant phenotype, gait disturbance, symmetrical disease at baseline, and depression as predictors of rapid decline. A study over 20 years by Hely et al.[6-8] has indicated that mortality rates are significantly higher than those seen in the general population and that nondrug responsive symptoms predominate late in the disease cycle.

PD is predominantly a movement disorder, and physical assessment is a key component in diagnosis of individuals presenting with PD.[9] There is however a lack of information on physical ability as a predictor of mortality in PD patients. Within the general population, a number of studies have attempted to consider whether physical assessment can be used to predict mortality. Notable among these is a study from 1994.[10] The study looked at whether mortality risk in community dwelling elderly persons was predicted by three lower limb physical function tests and self-reported capability with activities of daily living (ADLs) requiring the use of lower limbs. The study of 5,174 individuals found that self-reported disabilities in ADLs and walking half a mile were significant predictors of death. Other researchers have suggested that reduced physical fitness and loss of upper limb muscle bulk and grip strength may also predict an increased mortality risk in the general population.[11][12]

The primary aim of this study was to understand whether physical assessment of individuals diagnosed with PD is a useful tool in predicting mortality rates at 7-year follow-up.


The NHS trust involved in this study runs a comprehensive PD service for people with PD in their catchment area. A prevalence study determined that more than 85% of persons with idiopathic PD in the local catchment area are under the care of the service so these patients are thought to be representative of community dwelling PD patients.[13] All 141 patients registered with this service in January 2000 with idiopathic PD according to the UK brain bank criteria were considered for inclusion in a prospective study of falls.[14] The only exclusion criteria were being totally bedfast or major cognitive impairment. Thirty two patients declined to take part in the study.

A cohort of 109 (77%) individuals was eligible and consented, and was assessed using reliable and validated measures of physical function and ability from January to March 2000. Mortality status after 7 years (end of 2006) was then obtained, with retrospective assessment of the correlation between initial assessment and status at follow-up as the key aim of the study. National Research Ethics Service approval was granted for the use of both data sets used in this study.

Follow-up data on date of death and cause and place of death for individuals within the 2000 cohort who had died by 31st of December 2006 were obtained from the Office of National Statistics (ONS). Progression has been considered over 7 years, allowing a sufficiently long period of time for outcomes to be monitored. Because no patients are discharged from the service, ONS data on all those who were ever registered with the service were requested, including all persons who had moved away from the catchment area during the intervening 7 years. Thus, data from all 109 participants were included in this study, with no patients lost to follow-up.

The data collected in January 2000 can be divided into five main categories; general information and social history (e.g., sex, age, and domicile), previous medical history (e.g., blackouts, falls, and malignancies), medication and drug history, autonomic function, and physical function. Data within all these categories were collected using standard measuring devices, quality of life, and physical function rating scales, medical records or patient responses, as appropriate.

PD patients are known to present with widely varying physical signs and symptoms depending on which stage in their daily medication cycle they are at. A patient's physical ability when in an on state is often markedly better than when they are in an off state.[15][16] All patients were assessed in the morning after having taken their medication and, thus, were in the on state.

Data relating to physical function and ability were collected during a 30 minute objective and subjective assessment by a single senior physiotherapist with a special interest in PD at North Tyneside General Hospital. The specific tests carried out during the objective assessment are detailed in Table 1, included the 10-m walk test,[17] Tinetti balance and gait assessment,[18] Hoehn and Yahr rating,[19] and the Unified Parkinson's Disease Rating Scale (UPDRS) assessment.[20]

Table 1. Baseline physiotherapy assessment data

  Mean Minimum Maximum Range Standard deviation
Age at January 2000* 74.71 54 92 38 7.933
Years since diagnosis in 2000 5.42 0 31 31 5.563
Years since symptom onset in 2000 6.93 1 37 36 6.099
Total UPDRS score* 34.10 8 64 56 11.184
Hoehn and Yahr rating 2.023 1 4 3.0 0.7372
Total MMSE score* 26.33 15 30 15 3.594
Posture when standing rating (1-4)a* 3.11 2 4 2 0.637
Sitting to standing time (s) 3.03 1 16 15 2.282
Sitting to standing rating (0-3)b* 2.19 0 3 3 0.477
Lying to sitting time (s) 6.30 1 26 25 4.694
Lying to sitting rating (0-3)c* 2.08 1 3 2 0.367
Climb 4 stairs up time (s)* 7.11 3 62 59 6.774
Climb 4 stairs up rating (0-3)c 1.90 0 3 3 0.729
Climb 4 stairs down time (s)* 7.76 2 89 87 9.604
Climb 4 stairs down rating (0-3)c* 1.88 0 3 3 0.739
Getting up from floor time (s) 14.17 4 47 43 9.356
Getting up from floor rating (0-3)b* 1.24 0 3 3 0.956
Festination present in gait (0-1)d 0.14 0 1 1 0.350
Initiation difficulty (0-2)e 1.73 1 2 1 0.444
Gait description (0-3)f 2.55 1 3 2 0.573
Arm swing (0-4)g 1.88 0 4 4 1.305
10 m walk time (s)* 18.58 7 96 89 14.828
10 m walk number of strides* 26.05 13 82 69 13.801
10 m walk velocity (m/s)* 0.7362 0.10 1.43 1.33 0.32458
10 m walk seconds per stride 0.6691 0.16 1.82 1.66 0.21500
10 m walk stride length (m) 0.4666 0.01 1.64 1.63 0.22194
Tinetti Gait score (9-18)* 13.04 9 18 9 2.514
Tinetti Balance score (13-39)* 23.82 13 37 24 6.167
Total Tinetti Balance + Gait (22-57)* 36.86 22 54 32 8.220
* Correlation between mortality and predictors at P ≤ 0.05 significance level.
a 1, fully stooped posture; 2, significant flexion at hips, knees, trunk and shoulders; 2, some flexion at hips, knees, trunk and shoulders; 4, fully upright posture.
b 0, unable to assess; 1, unable to do; 2, able with use of an aid; 3, able without use of any aid.
c 0, unable to assess; 1, unable to do; 2, able with use of rail or stick; 3, able without use of rail or stick.
d 0, not present; 1, present.
e 0, unable to assess; 1, no initiation difficulty; 2, present.
f 0, unable to assess; 1, toe strike then heel; 2, flat footed; 3, heel then toe.
g 0, unable to assess; 1, bilateral loss; 2, unilateral loss; 3, unilateral reduction; 4, full bilateral.


The data were quantitative in nature and collected at a nominal, ordinal, and interval/ratio level. They were analyzed using standard statistical software, SPSS-16 for Windows (SPSS, Chicago, IL). With the exception of age, all predictor variables were found to be nonnormally distributed (Kolmogorov-Smirnov test) and so did not meet parametric assumptions.[21] Therefore, Spearman's correlation test was used to assess whether scores from one predictor were associated with scores from other predictors (multicollinearity) as part of a preliminary screening of the data. Because the outcome (mortality) was a dichotomy, a point biserial correlation test was used to assess correlation between predictor variables and mortality.[21] Finally, Cox proportional hazards regression analysis does not require data to be parametric and so was used to identify independent predictors of outcome and to obtain survival ratios. Time to event (death) was entered in months, with 1st January 2000 taken as baseline. A large value for the Wald statistic generally indicates that the predictor variable is a significant predictor of outcome and therefore makes a contribution to the predictive power of the model. Calculating Exp(B) from B, allows the relative change in the odds of the outcome occurring for a unit change in the predictor to be calculated.


Means, standard deviations, and spread of results for the baseline physiotherapy assessment data in January-March 2000 for all 109 individuals in the study are shown in Table 1. Of the 109 individuals in the original study, 46 (42.2%) were no longer alive by 31st December 2006. Table 2 shows demographic information on the data set with participants split into deceased or alive. The average age at death was 80.8 years (79.8 years for men and 82.4 years for women), with average years since onset of symptoms at death of almost 10 years.

Table 2. Demographic data split into deceased or alive

Mortality status Dec. 2006 Mean Minimum Maximum Range Std. Deviation
Age at January 2000 72.70 54 92 38 7.655
Years since diagnosis in 2000 5.08 0 27 27 5.629
Years since symptom onset in 2000 6.94 1 27 26 6.247
Age at January 2000 77.46 60 92 32 7.545
Years since diagnosis in 2000 5.89 1 31 30 5.498
Years since symptom onset in 2000 6.91 1 37 36 5.958
Age at death 80.80 63 99 36 7.398
Years from diagnosis to death 8.91 2.0 34.4 32.4 5.965
Years from symptom onset to death 9.93 3.3 40.4 37.1 6.446

Furthermore, of the 52 men in the study, 28 (53.8%) had died during the 7 year follow-up period, compared with only 18 (39.1%) of the 57 women. The ages in January 2000 for men and women were 74.0 and 75.3 years, respectively. The difference in mortality rates between men and women was significant ( = 5.528, P = 0.019). There was no significant difference between women and men when years since diagnosis (5.00 years for women, 5.88 years for men) or years since onset of symptoms (6.81 for women and 7.06 for men) were compared.

Correlation between selected predictor variables is shown in Table 3. Tinetti gait and balance scores and all five measures of walking were correlated to various degrees with one another in all combinations (r = -0.324-0.924, P = 0.001-0.000). UPDRS score was highly correlated with Hoehn and Yahr score (r = 0.555, P = 0.000), 10 m walk test time (r = 0.484, P = 0.000), Tinetti balance score (r = 0.567, P = 0.000), and Tinetti gait score (r = 0.646, P = 0.000). Correlation coefficients between age and sex and physical predictors, although statistically significant in some cases, are relatively small.

Table 3. Correlations between selected predictor variables

  Sex Age at start Total UPDRS 10 m walk time 10 m walk time per stride Tinetti Gait score Tinetti Balance score
Correlation Coefficient 1.000 0.052 -0.057 0.281a 0.136 0.036 0.214b
Sig. (2-tailed) - 0.591 0.558 0.003 0.157 0.714 0.028
Age at start
Correlation Coefficient 0.052 1.000 0.035 0.443a 0.293a 0.318a 0.328a
Sig. (2-tailed) 0.591 - 0.720 0.000 0.002 0.001 0.001
Correlation Coefficient -0.057 0.035 1.000 0.484a 0.216b 0.567a 0.646a
Sig. (2-tailed) 0.558 0.720 - 0.000 0.024 0.000 0.000
10 m walk time
Correlation Coefficient 0.281a 0.443a 0.484a 1.000 0.729a 0.786a 0.777a
Sig. (2-tailed) 0.003 0.000 0.000 - 0.000 0.000 0.000
10 m walk time per stride
Correlation Coefficient 0.136 0.293a 0.216b 0.729a 1.000 0.439a 0.479aa
Sig. (2-tailed) 0.157 0.002 0.024 0.000 - 0.000 0.000
Tinetti Gait score
Correlation Coefficient 0.036 0.318a 0.567a 0.786a 0.439a 1.000 0.750a
Sig. (2-tailed) 0.714 0.001 0.000 0.000 0.000 - 0.000
Tinetti Balance score
Correlation Coefficient 0.214b 0.328a 0.646a 0.777a 0.479a 0.750a 1.000
Sig. (2-tailed) 0.028 0.001 0.000 0.000 0.000 0.000 -
a Correlation is significant at the 0.01 level (2-tailed).
b Correlation is significant at the 0.05 level (2-tailed).

All variables were screened to discover whether any variables had a statistically significant association with mortality status. The 16 predictors marked with an asterisk in Table 1, together with sex, showed significant correlations at the P ≤ 0.05 level of statistical significance.

Given that many of the apparent predictor variables correlate highly with each other, and can therefore be said to be measuring the same phenomenon (i.e., physical ability, muscle strength etc.), a Cox regression model was constructed in an attempt to adjust for interactions. The differences in sex and age between outcome groups can be partially explained by demographics within the general population. The different life expectancies of men and women and increased mortality rates with increasing age, mean that these two factors are likely to be significant predictors of outcome, hence their inclusion in the model.

The key outputs from this model are shown in Table 4. The value of Exp(B) indicates that a change of one unit in the Tinetti gait score (score, 9-18) results in an increase of 1.3 in the odds of death occurring. The cumulative survival plot based on the Cox regression model is shown in Figure 1, with patients split into men or women, the hazard for male patients is 2.86 that of female patients.

Figure 1Figure 1. Cumulative survival plot, for a hypothetical individual, at the mean of the covariates for men and women.
[Normal View 18K | Magnified View 44K]

Table 4. Cox regression model

  B SE Wald Df Sig Exp(B) 95.0% CI for Exp(B)
Lower Upper
Sex 1.052 0.324 10.505 1 0.001 2.863 1.515 5.407
Age at January 2000 0.064 0.024 7.129 1 0.008 1.066 1.017 1.117
Tinetti gait score 0.263 0.068 14.975 1 0.000 1.301 1.139 1.486

UPDRS Phenotype Analysis

The UPDRS was developed in 1987[20] as an overall rating tool designed to follow the longitudinal course of PD. It is used to assess all aspects of disability and functional limitation resulting from PD.[22][23] Using the method described by Jankovic et al.,[24] it is possible to place patients into either a tremor predominant subgroup or a PIGD predominant subgroup. In total, 77 individuals (70.6%) were categorized as PIGD predominant, 12 (11.0%) fell into the tremor predominant subgroup, and 20 (18.4%) fell between the two cut-off scores for categorization and were thus of an indeterminate subgroup. In the study of 800 patients by Jankovic et al.,[24] 233 (29.1%) fell into the tremor phenotype, 441 (55.1%) into the PIGD phenotype, and 126 patients (15.8%) into the indeterminate group.

Of the 89 individuals categorized as either PIGD or tremor predominant, a total of 50 patients (56%) were still alive in December 2006. Sixty-seven percent (n = 8) of patients in the tremor predominant phenotype were still alive versus 55% (n = 42) in the PIGD group, P = 0.43. There was evidence of a small association between mortality status and mean UPDRS tremor subscore (r² = 0.04, P = 0.038).


For the most part, the cohort used in this study is significantly older and has a significantly higher percentage of women than those studied by other authors. In a study of disease prevalence in the North Tyneside general hospital catchment area, Porter et al.[13] observed the mean age of individuals with PD to be 74.1 years. Given this, and the broad inclusion criteria of this study, it is felt that the demographics of the population described here is more representative of the general population of PD patients than that used in similar studies.[24-31]

The average age at death was 79.8 years for men and 82.4 years for women, with an overall average of almost 81 years; the average years since onset of symptoms at death was almost 10 years. In comparison, Hoehn and Yahr's[19] study found an average age at death of 67.0 years and a mean disease duration at death of 9.4 years amongst 340 idiopathic PD patients. More recently, a community-based study in Sweden in 2003 found an average age at death of 81.9 years (83.2 years for women and 81.0 years for men) in 121 cases.[32] The relatively younger age at death in the 1967 study may be attributable to a number of factors such as increasing life expectancy within the general population, greater diagnostic accuracy, earlier diagnosis and treatment and improvements in therapy for PD patients since the mid-1960s.[33] The similarities in age at death between the Swedish study and this study may, in part, be due to broad inclusion criteria used in both studies.

A total of 17 predictors showed significant correlations with mortality and, of these, 13 are measures of either muscle strength or gait or balance or posture. Furthermore, sections II and III of the UPDRS score measure physical ability and ability with ADLs and so can be thought of as measuring the same phenomenon. Age at January 2000, sex, and total mini-mental state examination (MMSE) score were the only predictor variables that correlated highly with mortality status at 7 years follow-up that were not directly measuring physical function. The fact that MMSE score was a significant predictor of mortality suggests that cognitive decline, and other nonmotor factors, may also be important predictors of disease progression and mortality in PD, as noted by other authors.[6][8][34][35] Tinetti gait score, age, and sex appear to be independent predictors of mortality. More explicitly, the low multicollinearity between scores for these three variables suggests that they predict different parts of the variance within the survival model and therefore all contribute to its overall predictive power. The link between age, sex, and mortality is well established, but the influence of poor gait on mortality in PD has not previously been reported. Nevertheless, other variables may also be useful predictors of mortality, but do not appear in the model because they are measuring the same fundamental characteristic or phenomenon (i.e., physical ability). The strong association between Tinetti gait and Tinetti balance scores is of particular interest and emphasizes the link between poor balance and impaired walking ability.

Mortality status was not significantly correlated with PD phenotype, although the weak association between UPDRS tremor subscore and mortality status merits further investigation. Those alive in December 2006 had had higher mean tremor scores, indicating a greater degree of tremor or a more global bilateral tremor. Jankovic and Kapadia[28] have suggested that the slower rate of decline to disability in individuals with greater tremor may be due to a distinct biochemical or degenerative pathway of the disease not seen in PIGD predominant subjects. Strong evidence for a more rapid rate of progression to disability within the PIGD phenotype was also noted in a recent systematic review.[2]

Gait and balance impairment, together with a more flexed and stooped posture in standing, are the strongest physical predictors of mortality and are manifestations of the complex neurological changes associated with PD.[36] It may be that, ultimately, it is the rate of these neurological changes that predict mortality. If this is the case, then improving gait and balance may have little or no overall impact on mortality.

In contrast, it may also be suggested that improving gait and balance would reduce falls risk and therefore improve mobility, ability to perform ADLs, and social interaction. This would lead to an improved quality of life and reduced risk from conditions, which can be exacerbated by being house bound or immobile for long periods of time, such as depression and chest infection.[19] Indeed, pneumonia has long been recognized as a major cause of death in PD patients and of the 46 patients in this study who were deceased by December 2006, 21 (46%) had pneumonia or chest infection listed as the primary cause of death on their death certificate.[32][37]

Physiotherapy aimed at improving gait, balance, and posture is recommended by a number of clinical guidelines.[38][39] Recent systematic reviews concluded that therapy aimed at improving balance, physical capacity, and gait can be beneficial, although a lack of high quality evidence to support specific treatment approaches was noted.[40][41] Nevertheless, there is evidence to suggest that improving physical fitness can be beneficial in preventing immobility in PD populations, and that patients who exercise have a lower mortality rate.[42] In conclusion, this study suggests a link between diminished performances in commonly used tests of physical function, in particular, measures of gait, balance, posture, ability to climb and descend stairs and ability to change postural set, and an increased risk of mortality amongst PD patients. Furthermore, age, sex, and Tinetti gait score are independent predictors of mortality.

Having indicated which tests of physical function may be the most useful predictors of an increased mortality risk, it would be useful for future studies to consider the rate at which these scores decline and whether such measures can be useful in predicting the rate of decline of physical function and impending terminal decline. Interestingly, an association between a decreased mortality risk and higher scores in the tremor items of the UPDRS was observed.

Author Roles:

Research Project conception: William K. Gray, Katherine Baker, Richard W. Walker, Julie A. Bilclough; Organization: William K. Gray, Katherine Baker, Richard W. Walker; Execution: William K. Gray; Primary data collection: Brian H. Wood, Julie A. Bilclough; Statistical Analysis Design: William K. Gray, Anthony Hildreth; Execution: William K. Gray; Review and critique: Anthony Hildreth; Manuscript Writing of first draft: William K. Gray, Richard W. Walker; Review and critique: William K. Gray, Richard W. Walker, Anthony Hildreth, Julie A. Bilclough, Brian H. Wood, Katherine Baker.


The British Geriatric Society supported the initial physiotherapy data collection in 2000, via a start-up grant. This was part of a larger prospective study of falls in PD funded by Northumbria Healthcare NHS Foundation Trust. Follow-up data was obtained from a study of mortality in PD funded by the Parkinson's Disease Society. The study described here formed part of research undertaken by William K. Gray in part fulfillment of the requirements of a Masters degree in physiotherapy at the University of Northumbria, funded by the U.K. National Health Service.


  1. Gasparoli E, Delibori D, Polesello G, et al. Clinical predictors in Parkinson's disease. Neurol Sci 2002: 23: S77-S78. Links
  2. Post B, Merkus MP, De Haan RJ, Speelman JD. Prognostic factors for the progression of Parkinson's disease: a systematic review. Mov Disord 2007: 22: 1839-1851. Links
  3. Suchowersky O, Reich S, Perlmutter J, Zesiewicz T, Gronseth G, Weiner WJ. Practice parameter: diagnosis and prognosis of new onset Parkinson disease (an evidence-based review) report of the quality standards subcommittee of the American academy of neurology. Neurology 2006: 66: 968-975. Links
  4. Marras C, Rochon P, Lang AE. Predicting motor decline and disability in Parkinson disease - a systematic review. Arch Neurol 2002: 59: 1724-1728. Links
  5. Alves G, Wentzel-Larsen T, Aarsland D, Larsen JP. Progression of motor impairment and disability in Parkinson disease - a population-based study. Neurology 2005: 65: 1436-1441. Links
  6. Hely MA, Morris JGL, Reid WGJ, Trafficante R. Sydney multicenter study of Parkinson's disease: non-L-dopa-responsive problems dominate at 15 years. Mov Disord 2005: 20: 190-199. Links
  7. Hely MA, Morris JGL, Traficante R, Reid WGJ, O'sullivan DJ, Williamson PM. The Sydney multicentre study of Parkinson's disease: progression and mortality at 10 years. J Neurol Neurosurg Psychiatry 1999: 67: 300-307. Links
  8. Hely MA, Reid WGJ, Adena MA, Halliday GA, Morris JGL. The Sydney multicenter study of Parkinson's disease: the inevitability of dementia at 20 years. Mov Disord 2008: 23: 837-844. Links
  9. National Institute for Health and Clinical Excellence. Parkinson's disease: diagnosis and management in primary and secondary care. London: NICE Clinical Guideline 35: 2006.
  10. Guralnik JM, Simonsick EM, Ferrucci L, et al. A short physical performance battery assessing lower-extremity function - association with self-reported disability and prediction of mortality and nursing-home admission. J Gerontol 1994: 49: M85-M94. Links
  11. Sandvik L, Erikssen J, Thaulow E, Erikssen G, Mundal R, Rodahl K. Physical-fitness as a predictor of mortality among healthy, middle-aged Norwegian men. N Engl J Med 1993: 328: 533-537. Links
  12. Gale CR, Martyn CN, Cooper C, Sayer AA. Grip strength, body composition, and mortality. Int J Epidemiol 2007: 36: 228-235. Links
  13. Porter B, Macfarlane R, Unwin N, Walker R. The prevalence of Parkinson's disease in an area of North Tyneside in the North-East of England. Neuroepidemiology 2006: 26: 156-161. Links
  14. Wood BH, Bilclough JA, Bowron A, Walker RW. Incidence and prediction of falls in Parkinson's disease: a prospective multidisciplinary study. J Neurol Neurosurg Psychiatry 2002: 72: 721-725. Links
  15. Bond JM, Morris M. Goal-directed secondary motor tasks: their effects on gait in subjects with Parkinson disease. Arch Phys Med Rehabil 2000: 81: 110-116. Links
  16. O'sullivan JD, Said CM, Dillon LC, Hoffman M, Hughes AJ. Gait analysis in patients with Parkinson's disease and motor fluctuations: influence of levodopa and comparison with other measures of motor function. Mov Disord 1998: 13: 900-906. Links
  17. Wade DT. Measurement in neurological rehabilitation. Oxford medical publications. Oxford: Oxford University Press: 1992.
  18. Tinetti ME. Performance-oriented assessment of mobility problems in elderly patients. J Am Geriatr Soc 1986: 34: 119-126. Links
  19. Hoehn M, Yahr M. Parkinsonism: onset, progression and mortality. Neurology 1967: 17: 427-442. Links
  20. Fahn S, Elton RL. UPDRS development committee, Unified Parkinson's Disease Rating Scale. In: Fahn S, Marsden CD, Calne DB, Goldstein M, editors. Recent developments in Parkinson's disease. New Jersey: Macmillan: 1987. p 153-163.
  21. Field A. Discovering statistics: using SPSS. 3rd ed. London: Sage Publications Ltd: 2009.
  22. Olanow CW, Watts RL, Koller WC. An algorithm (decision tree) for the management of Parkinson's disease (2001): treatment guidelines. Neurology 2001: 56: S1-S88. Links
  23. Richards M, Marder K, Cote L, Mayeux R. Reliability of symptom onset assessment in Parkinsons-disease. Mov Disord 1994: 9: 340-342. Links
  24. Jankovic J, McDermott M, Carter J, et al. Variable expression of Parkinsons-disease - a base-line analysis of the DATATOP cohort. Neurology 1990: 40: 1529-1534. Links
  25. Diamond SG, Markham CH, Hoehn MM, McDowell FH, Muenter MD. An examination of male-female differences in progression and mortality of Parkinson's Disease. Neurology 1990: 40: 763-766. Links
  26. Goetz CG, Stebbins GT, Blasucci LM. Differential progression of motor impairment in levodopa-treated Parkinson's disease. Mov Disord 2000: 15: 479-484. Links
  27. Hely MA, Morris JG, Reid WG, Williamson PM, Broe GA, Adena MA. Age at onset: the major determinant of outcome in Parkinson's disease. Acta Neurol Scand 1995: 92: 455-463. Links
  28. Jankovic J, Kapadia AS. Functional decline in Parkinson disease. Arch Neurol 2001: 58: 1611-1615. Links
  29. Scigliano G, Musicco M, Soliveri P, et al. Progression and prognosis in Parkinson's disease in relation to concomitant cerebral or peripheral vasculopathy. Park Dis 1996: 69: 305-309. Links
  30. Starkstein SE, Mayberg HS, Leiguarda R, Preziosi TJ, Robinson RG. A prospective longitudinal-study of depression, cognitive decline, and physical impairments in patients with Parkinsons-disease. J Neurol Neurosurg Psychiatry 1992: 55: 377-382. Links
  31. Uitti RJ, Baba Y, Wszolek ZK, Putzke DJ. Defining the Parkinson's disease phenotype: initial symptoms and baseline characteristics in a clinical cohort. Park Relat Disord 2005: 11: 139-145. Links
  32. Fall PA, Saleh A, Fredrickson M, Olsson JE, Granerus AK. Survival time, mortality, and cause of death in elderly patients with Parkinson's disease: a 9-year follow-up. Mov Disord 2003: 18: 1312-1316. Links
  33. Trail M, Protas EJ, Lai EC. Neurorehabilitation in Parkinson's disease: an evidence based treatment model. Thorofare, New Jersey: Slack Incorporated: 2008.
  34. Chaudhuri KR, Healy DG, Schapira AHV. Non-motor symptoms of Parkinson's disease: diagnosis and management. Lancet Neurol 2006: 5: 235-245. Links
  35. Global Parkinson's Disease Survey Steering Committee. Factors impacting on quality of life in Parkinson's disease: results from an international survey. Mov Disord 2002: 17: 60-67. Links
  36. Braak H, Del Tredici K. Invited article: nervous system pathology in sporadic Parkinson disease. Neurology 2008: 70: 1916-1925. Links
  37. Beyer MK, Herlofson K, Aarsland D, Larsen JP. Causes of death in a community-based study of Parkinson's disease. Acta Neurol Scand 2001: 103: 7-11. Links
  38. Keus S, Hendriks H, Bloem B, et al. KNGF guidelines for physical therapy in patients with Parkinson's disease. Dutch J Physiother 2004: 114( Suppl): S1-S42. Links
  39. Plant R, Jones D, Thomson J, et al. Guidelines for physiotherapy practice in Parkinson's disease. Newcastle: Northumbria University, Institute of Rehabilitation: 2001.
  40. de Goede CJ, Keus SH, Kwakkel G, Wagenaar RC. The effect of physical therapy in Parkinson's disease: a research synthesis. Arch Phys Med Rehabil 2001: 82: 509-515. Links
  41. Deane KHO, Ellis-Hill C, Jones D, et al. Systematic review of paramedical therapies for Parkinson's disease. Mov Disord 2002: 17: 984-991. Links
  42. Gage H, Storey L. Rehabilitation for Parkinson's disease: a systematic review of available evidence. Clin Rehabil 2004: 18: 463-482. Links