Marked alterations in the gait timing and rhythmicity of patients with de novo Parkinson s disease

Transcription

1 European Journal of Neuroscience, Vol. 24, pp , 2006 doi: /j x Marked alterations in the gait timing and rhythmicity of patients with de novo Parkinson s disease Rossitza Baltadjieva, 1 * Nir Giladi, 1,2 * Leor Gruendlinger, 1 Chava Peretz 1,2 and Jeffrey M. Hausdorff 1,2,3 1 Laboratory for Gait & Neurodynamics, Movement Disorders Unit and NPF Center for Parkinson s Disease, Department of Neurology, Tel Aviv Sourasky Medical Center, 6 Weizmann Street, Tel Aviv, 64239, Israel 2 Sackler School of Medicine, Tel-Aviv University, Tel Aviv, Israel 3 Department of Medicine, Harvard Medical School, Boston, MA, USA Keywords: basal ganglia, gait variability, motor programming Abstract Little is known about the gait characteristics of subjects with de novo Parkinson s disease (PD). We hypothesized that alterations in the spatio-temporal characteristics of gait will already be quantifiable in these patients. The gait of 35 patients with idiopathic PD (mean age 60 years) who were in the early stages of the disease (Hoehn and Yahr stage 1.8 ± 0.5, median 2.0, range ) and were not yet treated with any anti-parkinsonian medications were compared with the gait of age- and sex-matched healthy controls (n ¼ 22). The patients walked more slowly and with reduced swing times while also exhibiting increased left right swing asymmetry and marked inconsistencies in the timing of gait. By contrast, significant group differences in the peak forces at heel-strike and in the stride-to-stride variability of the ground reaction forces (a reflection of muscle output consistency) were not observed. These findings indicate that in de novo PD, an altered gait pattern is observed, even though dramatic changes in the gait pattern may not yet be apparent visually (e.g. fairly intact gait speed). Furthermore, the results demonstrate that the observed alterations are not just sideeffects of treatments or complications of the disease. Instead, there is evidence for motor programming deficits in gait, as revealed by increased gait variability and asymmetry in timing. PD apparently impinges on the regulation of a consistent gait rhythm, even early in the course of the disease when observed alterations are not the result of any pharmacologic treatment. Introduction The gait changes in advanced Parkinson s disease (PD) have been well studied; however, much less is known about the alterations in gait in the initial stages of the disease. Only a few quantitative investigations of patients with PD included patients who were not treated with antiparkinsonian medications (de novo PD) (Ebersbach et al., 1999b). To better understand the natural history of the gait changes in PD, it is important to identify the early alterations, prior to and without the influence of any modification by anti-parkinsonian treatments. Treatment side-effects (e.g. dyskinesias) may be quite significant in the advanced stages of the disease, making it difficult to distinguish between consequences of the disease and other factors, such as muscle disuse, immobility and the effects of treatment. The goal of the present study was to characterize quantitatively the gait pattern of de novo PD patients (i.e. untreated patients with relatively recent onset). To our knowledge, this is the first study to evaluate systematically the gait of de novo PD patients. Based on studies in advanced PD, we hypothesized that: (1) early in the disease, there will be alterations in the spatio-temporal characteristics of gait including reduced speed, reduced percentage swing time, increased stride time and increased left right asymmetry, compared with age-matched controls; (2) peak vertical ground reaction forces (VGRFs) will be reduced at heel-strike and toe-off in patients with de novo PD; and Correspondence: Dr J. M. Hausdorff, 1 Laboratory for Gait & Neurodynamics, as above. jhausdor@bidmc.harvard.edu *R.B. and N.G. contributed equally to this work. Received 1 February 2006, revised 25 June 2006, accepted 3 July 2006 (3) increased stride-to-stride variability and inconsistency of the temporal parameters of gait will be observed in PD and will be accompanied by inconsistency of force production. As we show below, surprisingly, the results support only a subset of these hypotheses. Materials and methods Experimental subjects Thirty-five patients with idiopathic PD, as defined by standard criteria (Gelb et al., 1999), were recruited from the outpatient clinic of the Movement Disorders Unit at the Tel Aviv Sourasky Medical Center. The diagnosis of idiopathic PD was confirmed by a general and neurological examination performed by a movement disorders specialist. All subjects displayed at least two of the cardinal features of PD (rest tremor, rigidity and or bradykinesia). Patients were excluded if they had dementia (as evaluated using DSM IV criteria), affective disorders or autonomic disturbances. Subjects with a history of stroke, head trauma, other neurological disease, orthopedic impairment or other diseases that could likely contribute to a gait disturbance or parkinsonism were also excluded as were patients who were suspected to have multiple system atrophy. In addition, patients suffering from any acute disease or illness other than PD or who were otherwise unstable were also excluded. Only subjects who had not yet started any dopaminergic or anti-parkinsonian medications (de novo PD) were invited to participate in the study. All subjects had a normal CT. The PD patients were compared with 22 healthy control subjects of similar age and gender distribution who met all of the above inclusion exclusion criteria, except for those regarding PD. Disease

2 1816 R. Baltadjieva et al. severity among the PD patients was characterized using the Hoehn and Yahr staging (Hoehn & Yahr, 1967). The study was approved by the Human Studies Committee of the Tel Aviv Sourasky Medical Center. All subjects provided informed written consent according to the declaration of Helsinki prior to entering the study. Protocol After providing informed consent, subjects were familiarized with the walkway. They walked four times along a 20-m-long, 2-m-wide, welllit, straight path (a hallway). Subjects walked on level ground at their preferred, self-selected usual walking speed for a total of 80 m. Apparatus A previously described computerized gait analysis system (CGA) was used to quantify gait (Hennerici et al., 1994; Bazner et al., 2000; Frenkel-Toledo et al., 2005; Yogev et al., 2005). The CGA consists of a pair of shoes that measure the forces underneath the foot as a function of time, together with a recording unit. Each shoe contains eight load sensors that cover the surface of the sole and measure the normal (vertical) forces under the foot. The recording unit ( cm; 1.5 kg) is carried on the waist. Plantar pressures under each foot are recorded at a rate of 100 Hz and are used to derive kinetic and temporal measures off-line. Measurements are stored in a memory card during the walk and, after the walk, are transferred to a personal computer for further analysis. Outcome measures include the temporal parameters of gait (e.g. stride time, swing time) and the VGRFs, a measure that reflects, in part, muscle activation (Anderson & Pandy, 2003). the intrinsic dynamics (Hausdorff et al., 1998, 2001; Schaafsma et al., 2003). The outliers were typically steps that occurred at the end of the hallway (i.e. turns). To adjust for any height, weight or speed influences on gait, we used the following normalization procedures. Time variables are presented as percentage of cycle duration, speed and stride length as percentage of height s and percentage body height, respectively, and VGRFs are normalized by dividing by body weight. The mean values for all the variables, both for the left and the right foot, were calculated. To quantify the (in)consistency of these measures and the stride-to-stride variability, we calculated the coefficient of variation (CV) of these measures. Thus, for each parameter, characteristics of each subject s typical, representative stride were summarized by determining the mean value from the 80-m walk while the stride-to-stride fluctuations around the mean were quantified by determining the variance with respect to the mean. For each walk, we also determined the average gait speed and stride length. The time required to walk 8 m was measured in the middle of the walkway four times and averaged to calculate average gait speed. Stride length was determined by average speed average stride time. To estimate temporal asymmetry, we determined the natural logarithm of the ratio between the mean value of percentage swing time of the right foot and the mean value of percentage swing time of the left foot, multiplied by 100 (Yogev et al., in press). In order to account for asymmetry that could be disease-dependent, we defined longer swing time (LST) as the foot with the longer average swing time and shorter swing time (SST) as the foot with the shorter average swing time and compared the groups with respect to LST and SST variables (instead of left and right). We determined asymmetry of all variables in the same way and analysed left right and LST SST foot with respect to average swing time. Data analysis The output of the eight sensors under each foot was summed to produce a measure of the VGRF as a function of time. The following gait parameters were calculated for each stride: the stride (or cycle) time, percentage swing time, percentage double support time, peak forces at heel-strike and toe-off, and the lowest force at mid-stance, the times from the beginning of heel strike to the point of peak force at heel-strike and to the point of lowest force during mid-stance, and the time from the point of peak force during toe-off until the end of toeoff. The VGRF is typically characterized by two peaks in a saddle form, one at heel-strike and one at toe-off, with a local minimum in the force during mid-stance. Heel-strike force was defined as the first peak in the VGRF after the foot touches the ground. Toe-off force was defined as the last peak in the VGRF before toe-off, i.e. the peak during forefoot contact. Minimum mid-stance force was defined as the lowest force between these two peaks, i.e. during mid-stance. The time to heel-strike was defined as the time from initial contact to maximum heel force, presented as percentage of the stride cycle time. The time to toe-off was defined as the time from the peak at toe-off to the end of contact (terminal contact), presented as percentage of the cycle time. The time to mid-stance force was defined as the time from initial contact to the time of the lowest force at mid-stance, presented as percentage of the stride cycle time. If only one peak was identified, it was assigned as heel-strike peak or toe-off peak according to whether it occurred closer to the beginning or to the end of the stance phase, respectively. These parameters were calculated for each stride cycle during the 80-m walk. An automated filtering procedure was applied to eliminate outliers with respect to the mean of each time series in order to study Statistical analysis The two groups were compared using a Student s t-test (univariate analysis). To build a multivariate, parsimonious model of the characteristic features of the de novo PD subjects, we used logistic regression models to compare de novo PD cases and controls (dependent variable) with respect to the different features of the gait (covariates). Initially, univariate models were applied to compare one covariate at a time to determine if this parameter identified (i.e. was different in) the PD patients. Subsequently, the results of these univariate tests were used to establish a multivariate model, first for each aspect of gait (e.g. temporal parameters, ground reaction forces, inconsistency of timing, inconsistency of forces) and then, using the results of each aspect of gait, to establish a final multiple logistic regression model. At each step, multivariate models that initially included any covariates that had at least borderline significance (P ¼ 0.10) in the previous steps (e.g. univariate comparisons) were initially included. All models took into account the correlations between the covariates. The resulting multivariate model is a parsimonious description of the independent characteristics of the patient group. Descriptive statistics are reported as mean ± SD or using Pearson s correlation coefficients to show the bivariate association between variables. P-values reported are based on a two-sided comparison. All statistical analyses were performed using SAS (Version 8.2). Results Demographic and anthropometric characteristics of the patient and control groups are summarized in Table 1. Both groups were similar

3 Gait changes in de novo PD 1817 Table 1. Demographic and anthropometric characteristics of the two subject groups Table 3. Measures based on average values of the peak forces at heel-strike and toe-off (n ¼ 22) P-value* (n ¼ 22) P-value Age (years) 59.9 ± ± 11 ns No. of men women ns Height (cm) 169 ± ± 7 ns Weight (kg) 71 ± ± 11 ns Patients with de novo PD: Duration of symptoms 1.5 ± 0.9 (range years); Hoehn and Yahr Stage 1.8 ± 0.5 (median 2.0, range ). *P-value (according to t-test). ns, non-significant, P > 0.15; PD, Parkinson s disease. with respect to age, height, weight and gender. The Hoehn and Yahr stage of all patients was less than or equal to 2.5. The stage was 1.0, 1.5, 2.0 and 2.5 in eight, four, 20 and three PD subjects, respectively. Spatio-temporal parameters: univariate analyses Summary measures of the spatio-temporal parameters of gait are shown in Table 2. All variables were significantly (P <0.05) different between the PD patients and controls except for a borderline difference of percentage swing time LST (P ¼ 0.057). The patients with de novo PD walked significantly more slowly and with shorter strides. Stride time was significantly increased in the PD patients. The PD patients also spent more time with their feet on the ground and in double support. This was primarily due to a reduction in the percentage swing time of one leg. Consistent with this, the asymmetry measure based on swing time was significantly increased in the patient group. VGRFs and timing of peak forces: univariate analyses Significantly lower peak forces were observed in the forces at toe-off, but not at heel-strike, for both legs (Table 3). Mid-stance forces were not significantly different (P > 0.264; not shown). The results were similar when we compared the left and right feet or LST and SST for all ground reaction forces and times to peak forces. There were no significant group differences in the asymmetry of the forces at heelstrike or of toe-off between the LST and SST foot. While in controls Table 2. Spatio-temporal characteristics of gait (n ¼ 22) P-value Temporal parameters Speed (m s) 1.10 ± ± Stride length (m) 1.23 ± ± Stride time (s)* 1.13 ± ± Normalized speed (% height s) 65.4 ± ± Normalized stride length (% height) 73.2 ± ± Subphases of gait cycle (%) Double support time 25.6 ± ± Swing time LST 38.1 ± ± Swing time SST 36.2 ± ± Left right asymmetry (%) Swing time asymmetry 4.8 ± ± *Stride time reported throughout based on stride time in SST; similar results obtained using the other foot. Lower numbers reflect less asymmetry (e.g. on this scale, perfect symmetry would have a score of 0.0%); see the description in the Methods section. LST and SST refer to the foot with the longer and shorter average swing time, respectively. Average ground reaction forces (units of body-weight) at heel-strike and toe-off Heel-strike force LST 1.38 ± ± 0.21 ns Heel-strike force SST 1.41 ± ± Toe-off force LST 1.26 ± ± Toe-off force SST 1.33 ± ± Left right asymmetry (%) Heel-strike force asymmetry 6.3 ± ± 10.1 ns Toe-off force asymmetry 10.5 ± ± 7.1 ns Timing of forces (%) Time heel-strike LST 14.4 ± ± 1.8 ns Time heel-strike SST 15.2 ± ± Time toe-off LST 20.9 ± ± Time toe-off SST 21.7 ± ± profile Number of peaks LST 1.84 ± ± 0.23 ns Number of peaks SST 1.82 ± ± ns, non-significant, P > Ground reaction forces have been normalized to body-weight and are therefore dimensionless. there was no significant difference between the heel-strike and toe-off forces, in the de novo PD group the forces at heel-strike were higher, compared with toe-off force, for both feet (P ¼ and for LST and SST, respectively). The relative time from initial contact to peak force at heel-strike and the relative time between peak force at toe-off to end contact time tended to be longer in the PD patients (Table 3). Group differences were significant for the timing of the SST foot (longer for PD patients) at heel-strike and for both feet for toe-off. Within the PD group, changes in temporal variables tended to be more pronounced on the SST foot, as was the case for percentage swing time (see Table 2; contrast the results for swing time SST and swing time LST ). Consistency: univariate analyses All measures of the consistency of the temporal measures were significantly different between the two groups (Table 4). Although PD patients tended to have reduced variability in the forces at heel-strike Table 4. Consistency of forces and timing (n ¼ 22) P-value Consistency of temporal measures (%) Stride time CV 2.8 ± ± Swing time CV LST 3.0 ± ± Swing time CV SST 3.0 ± ± Double support CV 5.9 ± ± Consistency of forces (%) Heel-strike force CV LST 2.9 ± ± 1.5 ns Heel-strike force CV SST 3.0 ± ± 1.1 ns Toe-off force CV LST 3.0 ± ± 0.8 ns Toe-off forcecv SST 2.8 ± ± 0.9 ns Consistency of timing of forces (%) Time heel-strike CV LST 11.3 ± ± 5.1 ns Time heel-strike CV SST 10.0 ± ± 6.9 ns Time toe-off CV LST 10.1 ± ± Time toe-off CV SST 11.5 ± ± ns, non-significant, P > 0.15.

4 1818 R. Baltadjieva et al. and increased variability in the forces at toe-off, there was no significant group difference in the stride-to-stride variability of any of the forces. The variability of toe-off times in the PD group was significantly increased compared with controls for both legs, whereas the consistency of the timing of heel-strikes was not different in the two groups. Example data sets illustrating many of the changes observed in the de novo PD patients are shown in Fig. 1. Associations between different gait features Many of the gait characteristics were highly correlated with one another. This was true whether we examined all subjects or only the subjects with PD. To identify the independent characteristics of gait in de novo PD, variables from each aspect of gait (e.g. temporal parameters, ground reaction forces, inconsistency of timing, inconsistency of forces) that were found to be significant in univariate analyses and were not significantly correlated with each other were entered into a multiple regression model. Thus, three measures were tested in a final multivariate model: gait speed, force at toe-off LST and percentage swing time CV SST. This multivariate logistic model showed that percentage swing time CV SST (P ¼ 0.023) and force toe-off LST (P ¼ 0.019) were independent predictors of de novo PD. In this multivariate model, gait speed was a borderline significant predictor of de novo PD (P ¼ 0.061). This is consistent with the univariate results. Whereas gait speed was highly correlated with stride length (r ¼ 0.92, P < 0.01), it was only moderately correlated with swing time CV SST (r ¼ 0.48, P < 0.01) and was not significantly associated with stride time CV or the forces at toe-off. CONTROL De Novo PD A. Ground Reaction Ground Reaction Time (sec) Time (sec) B. Peak at Heel-Strike Peak at Heel-Strike Stride # Stride # C. Swing Time Series Swing Time (%) LST SST Swing Time (%) Swing Time Series LST SST Stride # Stride # Fig. 1. Example data sets from a patient and a control subject. (A) Vertical ground reaction forces as a function of time relative to heel-strike for multiple strides. is given in units of body-weight. Data from short swing time (SST) foot. The overlaid curves reflect the multiple strides during the walking test. Shown is the time from heel-strike to toe-off, i.e. the time when the foot is in contact with the ground. (B) Peak values of force at heel-strike as a function of stride number. is given in units of body-weight and is dimensionless. Data from short swing time (SST) foot. (C) Swing time as a function of stride number in the SST and LST feet. Features to note include: the double peak waveform is relatively preserved in this PD subject (A); peak forces are similar in the PD and control subjects (A and B); the time from heel-strike to toe-off is greater in the patient (0.8 vs. 0.7 s) (A); the time from toe-off to the peak at toe-off is relatively greater in the PD subject (A), and swing time variability and swing time asymmetry, as reflected by the gap between the SST and LST swing times, are larger in the patient (C).

5 Gait changes in de novo PD 1819 Discussion This study has a number of key findings. (1) patients walk more slowly, with shorter strides, increased stride time duration and double support time, and reduced swing times, compared with agematched controls. (2) The peak force at heel-strike is delayed and the peak force at toe-off is reached earlier (delayed heel-strike and earlier forefoot loading) in de novo PD, as compared with controls (see Table 3). (3) Increased gait asymmetry is already observed in the temporal parameters (e.g. swing times) of gait, consistent with reports in advanced PD (Koozekanani et al., 1987; Pedersen et al., 1997; Yogev et al., in press), but not in the ground reaction forces of de novo patients. (4) Compared with aged-matched controls, patients with de novo PD walk with increased timing variability in temporal measures, but no significant difference in the variability of the VGRFs are observed, as compared with controls. (5) Multivariate analysis indicates that the observed changes in gait rhythmicity are not simply a by-product of bradykinesia and a reduced gait speed. These findings indicate that significant gait changes including altered rhythmicity are already present early in the disease stage. Similar gait characteristics have been reported in advanced PD, both while off and on anti-parkinsonian medications, but with much more pronounced changes (Murray et al., 1978; Blin et al., 1991; Morris et al., 1994, 1996; Pedersen et al., 1997; Stolze et al., 2001). For example, in contrast to the values reported in Table 2, Stolze et al. (2001) observed that gait speed was reduced to 0.76 ± 0.29 m s and Blin et al. (1991) reported swing percentage times of 31 and 33% in patients off and on levodopa, respectively, in PD patients with more advanced disease. Significant changes in the spatio-temporal aspects of gait are apparently already present early in the disease process and these changes are exacerbated as disease progresses. Nonetheless, the observed changes in de novo PD were relatively mild. Gait speed, a good global marker of locomotor function, was not much different from control values. The reduced time between peak forces at heelstrike and toe-off is consistent with reports in advanced PD (Pedersen et al., 1997; Nieuwboer et al., 1999) and is considered to reflect the flat-footed gait pattern and reduced roll-off typical of the advanced disease state. Whereas it disappears in the advanced stages of the disease, in de novo PD, the double-peaked, saddle shape of the VGRF is still relatively preserved, despite the presence of mild features of a flat-footed gait pattern. Similarly, while it is present in the more advanced stages of the disease (Nieuwboer et al., 1999), asymmetry of weight bearing is not present early in the disease. Thus, while many of the classic parkinsonian gait alterations are already quantifiable early in the disease stage, the degree of severity is less than that seen in advanced PD, and not all features have yet become involved. An important finding of the present study concerns the measures of consistency and variability. Already early in the disease process, gait rhythmicity is impaired (e.g. see Table 4). Both temporal and spatial variability of gait are higher in advanced PD (Blin et al., 1990; Hausdorff et al., 1998; Stolze et al., 2001), which some authors have attributed to an increased inherent variability in muscle force production (Stelmach et al., 1989; Blin et al., 1991; Ebersbach et al., 1999b). The within-subject increase in the inconsistencies of stride time, swing time and toe-off timing in de novo PD, but not in peak forces at toe-off or heel-strike, likely reflects a deficiency in the time-keeping function (Ebersbach et al., 1999a) or impairment in the generation of the motor programming and internal cueing needed to walk in a rhythmic fashion (Sheridan & Flowers, 1990; Morris et al., 1994; Schaafsma et al., 2003). Thus, the increased temporal variability can be interpreted as a disturbed rhythmicity that cannot simply be attributed to inconsistency of force production. In de novo PD, gait rhythmicity is impaired, while force variability is relatively unchanged. This altered gait pattern is consistent with the idea that gait changes in PD reflect deficits in the neural circuits controlling locomotion and a return to an immature gait pattern (Forssberg et al., 1984). This disturbance in walking rhythmicity relatively early in the disease process might be the result of patho-physiology in dopaminesensitive pathways such as those that have been related to altered gait rhythmicity in more advanced PD (Schaafsma et al., 2003); however, pathologic changes in non-dopaminergic pathways that are already present early in the disease process (Braak et al., 2003) might also contribute to the observed findings. The present study has a number of limitations. For example, in the future, it would be helpful to measure muscle strength, power and EMG timing, and perhaps to compare neuroimaging findings in de novo patients with clinical measures such as scores on the Unified Parkinson s Disease Rating Scale and gait findings. Although all subjects were non-demented, as per DSM IV criteria, quantitative measures should be taken into account in the future to see if the early changes in gait are related to changes in cognitive function. Furthermore, it is possible that microangiopathic lesions not detected by CT may have influenced the gait of some of the patients (Ebersbach et al., 1999b). Prospective, follow-up study of de novo patients would also enable us to better understand the time course of the disease and its effect on gait and to determine when and why other changes observed in advanced disease become manifest. The ability to quantify and detect changes in gait in de novo PD also suggests the possibility that such measures may be used to augment the monitoring of disease progression and the response to therapy; however, longitudinal studies are needed to assess this question more fully. Many of the changes observed in advanced PD are already present early in the disease process. There is evidence supporting the idea that motor programming deficits in gait, as revealed by increased gait variability and asymmetry in timing, exist in de novo patients. PD apparently impinges on the regulation of a consistent gait rhythm even early in the course of the disease when observed alterations are not the result of pharmacologic treatment. Acknowledgements This work was supported in part by NIH grants AG RR-13622, HD-39838, AG-08812, the Parkinson s Disease Foundation and the National Parkinson Foundation. The authors thank all of the subjects for their participation, time and effort. References Anderson, F.C. & Pandy, M.G. (2003) Individual muscle contributions to support in normal walking. Gait Posture, 17, Bazner, H., Oster, M., Daffertshofer, M. & Hennerici, M. (2000) Assessment of gait in subcortical vascular encephalopathy by computerized analysis: a cross-sectional and longitudinal study. J. Neurol., 247, Blin, O., Ferrandez, A.M., Pailhous, J. & Serratrice, G. (1991) Dopa-sensitive and dopa-resistant gait parameters in Parkinson s disease. J. Neurol. Sci., 103, Blin, O., Ferrandez, A.M. & Serratrice, G. (1990) Quantitative analysis of gait in Parkinson patients: increased variability of stride length. J. Neurol. Sci., 98, Braak, H., Rub, U., Gai, W.P. & Del Tredici, K. (2003) Idiopathic Parkinson s disease: possible routes by which vulnerable neuronal types may be subject to neuroinvasion by an unknown pathogen. J. Neural Transm., 110, Ebersbach, G., Heijmenberg, M., Kindermann, L., Trottenberg, T., Wissel, J. & Poewe, W. (1999a) Interference of rhythmic constraint on gait in healthy subjects and patients with early Parkinson s disease: evidence for impaired

6 1820 R. Baltadjieva et al. locomotor pattern generation in early Parkinson s disease. Mov. Disord., 14, Ebersbach, G., Sojer, M., Valldeoriola, F., Wissel, J., Muller, J., Tolosa, E. & Poewe, W. (1999b) Comparative analysis of gait in Parkinson s disease, cerebellar ataxia and subcortical arteriosclerotic encephalopathy. Brain, 122, Forssberg, H., Johnels, B. & Steg, G. (1984) Is parkinsonian gait caused by a regression to an immature walking pattern? Adv. Neurol., 40, Frenkel-Toledo, S., Giladi, N., Peretz, C., Herman, T., Gruendlinger, L. & Hausdorff, J.M. (2005) Effect of gait speed on gait rhythmicity in Parkinson s disease: variability of stride time and swing time respond differently. J. Neuroeng Rehabil., 2, / Gelb, D.J., Oliver, E. & Gilman, S. (1999) Diagnostic criteria for Parkinson disease. Arch. Neurol., 56, Hausdorff, J.M., Cudkowicz, M.E., Firtion, R., Wei, J.Y. & Goldberger, A.L. (1998) Gait variability and basal ganglia disorders: stride-to-stride variations of gait cycle timing in Parkinson s disease and Huntington s disease. Mov. Disord., 13, Hausdorff, J.M., Rios, D. & Edelberg, H.K. (2001) Gait variability and fall risk in community-living older adults: a 1-year prospective study. Arch. Phys. Med. Rehabil., 82, Hennerici, M.G., Oster, M., Cohen, S., Schwartz, A., Motsch, L. & Daffertshofer, M. (1994) Are gait disturbances and white matter degeneration early indicators of vascular dementia? Dementia, 5, Hoehn, M.M. & Yahr, M.D. (1967) Parkinsonism: onset, progression and mortality. Neurology, 17, Koozekanani, S.H., Balmaseda, M.T. Jr, Fatehi, M.T. & Lowney, E.D. (1987) Ground reaction forces during ambulation in parkinsonism: pilot study. Arch. Phys. Med. Rehabil., 68, Morris, M.E., Iansek, R., Matyas, T.A. & Summers, J.J. (1994) The pathogenesis of gait hypokinesia in Parkinson s disease. Brain, 117, Morris, M.E., Iansek, R., Matyas, T.A. & Summers, J.J. (1996) Stride length regulation in Parkinson s disease. Normalization strategies and underlying mechanisms. Brain, 119, Murray, M.P., Sepic, S.B., Gardner, G.M. & Downs, W.J. (1978) Walking patterns of men with parkinsonism. Am. J. Phys. Med., 57, Nieuwboer, A., De Weerdt, W., Dom, R., Peeraer, L., Lesaffre, E., Hilde, F. & Baunach, B. (1999) Plantar force distribution in Parkinsonian gait: a comparison between patients and age-matched control subjects. Scand. J. Rehabil. Med., 31, Pedersen, S.W., Oberg, B., Larsson, L.E. & Lindval, B. (1997) Gait analysis, isokinetic muscle strength measurement in patients with Parkinson s disease. Scand. J. Rehabil. Med., 29, Schaafsma, J.D., Giladi, N., Balash, Y., Bartels, A.L., Gurevich, T. & Hausdorff, J.M. (2003) Gait dynamics in Parkinson s disease: relationship to Parkinsonian features, falls and response to levodopa. J. Neurol. Sci., 212, Sheridan, M.R. & Flowers, K.A. (1990) Movement variability and bradykinesia in Parkinson s disease. Brain, 113, Stelmach, G.E., Teasdale, N., Phillips, J. & Worringham, C.J. (1989) production characteristics in Parkinson s disease. Exp. Brain Res., 76, Stolze, H., Kuhtz-Buschbeck, J.P., Drucke, H., Johnk, K., Illert, M. & Deuschl, G. (2001) Comparative analysis of the gait disorder of normal pressure hydrocephalus and Parkinson s disease. J. Neurol. Neurosurg. Psychiatry, 70, Yogev, G., Giladi, N., Peretz, C., Springer, S., Simon, E.S. & Hausdorff, J.M. (2005) Dual tasking, gait rhythmicity, and Parkinson s disease: which aspects of gait are attention demanding? Eur. J. Neurosci., 22, Yogev, G., Plotnik, M., Peretz, C., Giladi, N. & Hausdorff, J.M. (in press) Gait asymmetry: The effects of dual tasking in patients with Parkinson s disease and elderly fallers Exp Brain Res., in press.