Obstacle avoidance to elicit freezing of gait during treadmill walking


  • Anke H. Snijders MD,

    1. Department of Neurology and Parkinson Centre Nijmegen, Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands
    Search for more papers by this author
  • Vivian Weerdesteyn PT, PhD,

    1. Department of Rehabilitation, Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands
    Search for more papers by this author
  • Yolien J. Hagen MSc,

    1. Department of Neurology and Parkinson Centre Nijmegen, Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands
    Search for more papers by this author
  • Jacques Duysens MD, PhD,

    1. Department of Rehabilitation, Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands
    Search for more papers by this author
  • Nir Giladi MD, PhD,

    1. Movement Disorders Unit, Department of Neurology, Tel-Aviv Sourasky Medical Center, Sackler School of Medicine, Tel-Aviv University, Tel-Aviv, Israel
    Search for more papers by this author
  • Bastiaan R. Bloem MD, PhD

    Corresponding author
    1. Department of Neurology and Parkinson Centre Nijmegen, Donders Institute for Brain, Cognition and Behaviour, Radboud University Nijmegen Medical Centre, Nijmegen, The Netherlands
    • Parkinson Centre Nijmegen (ParC), Department of Neurology, 935, Radboud University Nijmegen Medical Centre, PO Box 9101, 6500 HB Nijmegen, The Netherlands
    Search for more papers by this author

  • Potential conflicts of interest: Nothing to report.


Freezing of gait (FOG) is a common and disabling feature of Parkinson's disease (PD). Detailed pathophysiological studies are hampered by the fact that FOG episodes are difficult to elicit in a gait laboratory. We evaluated whether the need to avoid sudden obstacles on a treadmill can provoke FOG. We included 21 PD patients (15 with self-reported off-period FOG). Patients were tested in the off-state. FOG during overground walking was assessed using a standardized gait trajectory and axial 360° turns. Subsequently, patients walked on a motorized treadmill with suddenly appearing obstacles that necessitated compensatory stepping. Performance was videotaped, and presence of FOG was scored visually by two independent raters. Thirteen patients showed FOG during overground walking. During treadmill walking, obstacle avoidance was associated with 13 unequivocal FOG episodes in eight patients, whereas only one patient froze during undisturbed treadmill walking (Wilcoxon z = −2.0, P = 0.046). FOG episodes elicited by obstacle avoidance were brief (typically <1 s). Almost all episodes were provoked when subjects had a longer available response time. In conclusion, suddenly appearing obstacles on a treadmill can elicit FOG in a controlled laboratory setting. However, the moving treadmill and the obstacle both act as cues, which apparently help to immediately overcome the provoked FOG episode. This may limit the ecological validity of this new approach. © 2009 Movement Disorder Society


Freezing of gait (FOG) is a common and disabling feature of Parkinson's disease (PD).1 A FOG episode is defined as an episodic inability to generate effective forward stepping movements, in the absence of any known cause other than Parkinsonism or higher cortical deficits.2 It is most commonly experienced during turning and step initiation, also when faced with spatial constraint, stress, or distraction. Focused attention and external stimuli (cues) can help to overcome the episode.2 Because of its sudden and unpredictable nature, FOG often leads to falls and injuries.3

Several therapies are used to treat FOG, including pharmacotherapy, stereotactic deep brain surgery and the use of visual, auditory, or mental cueing techniques.4–6 Unfortunately, these therapies generally fail to provide a gratifying and lasting improvement in FOG. Improved treatment strategies are needed, tailored to the specific underlying pathophysiology. However, the mechanisms leading to FOG remain insufficiently understood.

To further clarify the pathophysiology of FOG, there is a need for detailed gait analyses under carefully controlled experimental conditions, for example while patients walk on a treadmill in a gait laboratory. However, evaluating FOG proves extremely difficult in such a research environment, because even patients with clear FOG typically fail to demonstrate the phenomenon during formal experimental testing.7, 8

In this study, we aimed to evaluate a new procedure that would evoke FOG in an experimental gait laboratory. Inspired by pilot observations,9 we investigated whether suddenly appearing obstacles could be used to provoke FOG in patients with PD walking on a motorized treadmill. In addition, we investigated the effect of the available time to negotiate the obstacle. More preparation time may facilitate obstacle avoidance, but may also challenge a more complex neural circuitry10 and thereby promote FOG episodes.



Twenty-one patients were randomly recruited from the outpatient clinic of Parkinson Centre Nijmegen. Inclusion criteria were idiopathic PD (diagnosed according to the UK Brain Bank criteria),11 Hoehn and Yahr stage 2-3 and ability to walk on a treadmill during an off period. Exclusion criteria were other neurological disorders, severe comorbidity, other clinically relevant causes of gait impairment, or severe cognitive impairment (mini mental state examination (MMSE) < 25 or frontal assessment battery (FAB) < 12).12 Fifteen patients were included with subjective FOG (answer “yes” to item no 3 of the FOG Questionnaire: Do you sometimes feel as if your feet get “glued to the floor”),13 mainly occurring during off periods. We did not include patients with mainly “on period” freezing. In addition, six patients without a history of FOG were included. Sixteen patients were examined in the morning during a practically defined off period at least 12 hours after intake of the last dose of antiparkinson medication. Because this was impossible for the five remaining patients (all freezers), they were measured during an end-of-dose off state. There were no differences in freezing outcomes (overground or treadmill FOG, described below) between these five patients and the other patients. All subjects gave informed written consent and the study was approved by the local ethics committee.

Clinical Assessment During Overground Walking

Before the treadmill experiment, Unified Parkinson's Disease Rating Scale (UPDRS) motor score and Hoehn and Yahr stage (H&Y) were obtained. In addition, cognitive status was assessed using the MMSE and the FAB.12

To grade patient-rated FOG severity, the FOG Questionnaire and the new FOG Questionnaire (FOGQ-II) were used.13, 14 To examine observer-rated FOG, a standardized gait trajectory was used containing specific elements to provoke FOG: starting walking, walking through a narrow passage (created by two chairs placed 50 cm apart), temporarily stop walking and start walking again. Patients also had to turn in a tight square of 1 by 1 m. First, they turned 360° in a self-selected direction, followed by a 540° turn in the opposite direction. Finally, patients walked 7.5 m back to the chair. This gait trajectory was repeated three times: at normal speed, as rapidly as possible, and while performing a dual task (mental arithmetic). In addition, patients were asked to make eight 360° axial turns from standstill in both directions, both at normal and at high speed.8, 15

Treadmill Experiment

Subjects walked on a motorized treadmill at a speed of 2 km/h. In two subjects 1.5 km/h was used, as 2 km/h was not feasible. Patients wore a safety harness attached to the ceiling to prevent falling, without interfering with natural walking. In addition, one of the examiners constantly watched the patient with an emergency brake at hand. Patients did not have a handrail.

First, patients were familiarized with treadmill walking for 5 minutes, and practiced obstacle avoidance with five to seven obstacles. Patients were instructed to avoid the obstacle and not to step alongside or onto the obstacle. The obstacles were sized 40 cm by 30 cm with a height of 1.5 cm and were attached by a magnet to a bridge placed over the treadmill, as described earlier (Fig. 1).16 Using remote control, the obstacles were unexpectedly released in front of one of the legs at random intervals (30–60 seconds apart). The patients could see the obstacle being attached to the magnet, but had no possibility to predict when the obstacle would fall. We manipulated the available time to negotiate the obstacle. For this purpose, subjects were positioned at either a short distance from the appearing obstacle (10 cm) or a long distance (1 m). This was repeated for both the left and right leg. During the short distance conditions, 15 obstacles fell in front of each leg at three different moments in the step cycle (mid stance, early swing and mid swing). For the long distance conditions, 10 obstacles fell in front of each leg. Hence, the whole experiment consisted of 50 obstacles. The order of the test conditions (short distance—most affected leg, short distance—less affected leg, long distance—most affected leg, long distance—less affected leg), was randomly alternated for each subject.

Figure 1.

Experimental set-up. The obstacle is attached by a magnet to a bridge placed over the treadmill.

The period between obstacle release until the obstacle left the treadmill was defined as “obstacle avoidance,” while the period from the moment the obstacle left the treadmill until eight strides later was defined as “baseline treadmill walking.”

Outcome Measures

The primary outcome measure was the presence of FOG according to the current gold standard: visual assessment of digitally videotaped gait performance.8 Performance during baseline assessment (overground walking) and the treadmill experiment was digitally videotaped for later objective evaluation. The presence of FOG during baseline overground assessment was scored by a single rater with considerable clinical experience with FOG. To guarantee valid rating of the “new” treadmill FOG, the treadmill performance was independently rated by two other raters with considerable clinical experience with FOG. Only unequivocal FOG in the eyes of both raters was classified as a FOG episode to ensure high specificity. An additional independent rater with extensive knowledge of obstacle avoidance ascertained that no variants of normal obstacle avoidance were inadvertently rated as FOG episodes. All four raters were not present during the experiments and were blinded to the clinical status (prior FOG yes or no) and to the other rater's score.

For each video fragment with FOG, the circumstances of FOG were determined: which leg (most/least affected), which distance to the obstacle (10 or 100 cm), before or after crossing the obstacle. Furthermore, the duration of the FOG episode and the need to stop the treadmill were determined visually.

Data Analysis

The proportion of subjects that showed at least one unequivocal FOG episode during overground clinical assessment, baseline treadmill walking and during obstacle avoidance on the treadmill was compared using a Friedman test with a significance level of P = 0.05. If this test was significant, a Wilcoxon's signed rank test was used to assess differences between these conditions. Patient characteristic differences between groups were analyzed using a Kruskal-Wallis test. Agreement between raters was assessed using Cohen's Kappa. Data were analyzed using SPSS 14.0 for windows, using a significance level of P = 0.05.


FOG During Overground Walking

Twelve of the 15 freezers (80%) showed at least one FOG episode during overground assessment (mean number of FOG episodes 5.6, range 0–16). Only two freezers (13%) also showed FOG during straight walking (one freezer during gait initiation, while crossing a narrow passage and upon reaching the destination; the other freezer only during gait initiation and upon reaching the destination. All other FOG episodes were associated with turning). One subjective “nonfreezer” (17%) unexpectedly showed a FOG episode during the off-state overground assessment, although it was very brief (<1 s). This patient was therefore classified as a freezer for the remainder of the experiment.

FOG During Treadmill Walking

Characteristics of the patients and occurrence of FOG are summarized in Table 1. During treadmill walking, obstacle avoidance was associated with 13 unequivocal FOG episodes in eight patients (all of whom also showed overground FOG), whereas only one patient (patient 6 in Table 1) froze during undisturbed treadmill walking (Friedman Chi square = 16.9, P < 0.001, baseline treadmill walking versus obstacle avoidance Wilcoxon z = −2.0, P = 0.046). The patient that also froze during undisturbed treadmill walking, was the same patient that also showed the most FOG during straight overground walking. The patients with unequivocal FOG episodes on the treadmill experienced these with 6% of obstacles that had to be avoided.

Table 1. Clinical characteristics of all participants in relation to the occurrence of FOG
PatientsAgeGenderUPDRSH&YFOGQ scoreFOGQ-II scoreOver-ground FOG*FOG during OA on treadmill*
  1. UPDRS = Unified Parkinson's Disease Rating Scale; H&Y = Hoehn and Yahr stage; FOGQ-score is FOG Questionnaire score13; FOGQ II = new FOG Questionnaire score14; OA = obstacle avoidance; * = Number of episodes.

Overground freezers64 322.31016100%67%
Freezers only on history63 382.710120%0%
Non-freezers58 302.2200%0%

More patients froze during overground walking (n = 13) compared with obstacle avoidance on the treadmill (n = 8, Wilcoxon z = −2.6, P = 0.008) or baseline treadmill walking (n = 1, Wilcoxon z = −3.3, P = 0.001). However, only two patients showed FOG episodes during overground straight walking, as almost all FOG episodes occurred during turning. If only FOG episodes during straight walking were taken into account, obstacle avoidance on the treadmill provoked more FOG than overground walking (n = 2, Wilcoxon z = −2.4, P = 0.014). There was no difference in age, UPDRS score, Hoehn and Yahr stage, or cognitive status between patients with or without FOG on the treadmill. In addition, there was no difference in the number of overground FOG episodes between freezers with or without FOG on the treadmill.

Characteristics of FOG Episodes on the Treadmill

Table 2 states which conditions provoked FOG, and whether FOG occurred before or after obstacle crossing. Video 1 shows an example of FOG during the long distance condition, directly after the obstacle crossing. Video 2 shows an example of FOG during the short distance condition, between obstacle release and obstacle crossing. Video 3 shows one patient (patient 5 in Table 1 and 2) where FOG already occurred when the obstacle was placed but not yet released, so this was apparently provoked by the mere anticipation of the upcoming obstacle. Two patients only showed a clear FOG episode during the practice trials. In the other patients, the FOG episodes were distributed equally over the experiment.

Table 2. Occurrence of FOG during obstacle avoidance on the treadmill
PatientsShort distance conditionLong distance conditionObstacle before MA legObstacle before LA legFOG before obstacle crossFOG after obstacle cross
  1. 0 = No FOG; 1 = at least one unequivocal FOG episode occurred; MA leg = most affected leg; LA leg = less affected leg; FOG = Freezing of gait; NAP = Not able to perform due to festination.

Total (%)40%75%63%38%63%38%

Duration of the FOG episodes during obstacle avoidance was always less than 1 second, and the treadmill never needed to be stopped because of these episodes. The FOG episodes that occurred “spontaneously” during undisturbed treadmill walking without the need to actually avoid an obstacle did sometimes force an emergency stop (Video 3).

Agreement Among Raters

The rating of separate trials as containing either “FOG” or “no FOG” had an acceptable agreement between both clinical raters (Kappa 0.70, P < 0.001). There was perfect inter-rater agreement for the classification of individual patients as “treadmill freezer” or “treadmill nonfreezer” (Kappa 1.0, P < 0.001). Also, all trials that were rated as “FOG” by the clinical raters, were also rated as “abnormal” by the obstacle avoidance rater.


Our findings show that suddenly appearing obstacles during treadmill walking can provoke unequivocal FOG episodes. This observation opens a new avenue to study FOG in a controlled laboratory setting. However, these FOG episodes were less common and much briefer compared with episodes during overground walking.

The ability of sudden obstacles to provoke FOG, even while walking in an experimental environment, may be interpreted in several ways. A first explanation is that upcoming obstacles force subjects to adjust their gait, necessitating a rapid planning and execution of adaptive movements. During normal walking, PD patients partially rely upon cortical compensation to bypass their defect basal ganglia.17 The need to combine consciously controlled walking with a deliberate obstacle avoidance may well outweigh the available resources.18, 19 In support of this notion, we observed that more FOG episodes were provoked when the obstacle fell in front of the most affected leg. Probably, movement adaptation with the most affected leg needs more resources than adaptation with the least affected leg. In addition, the long distance condition produced even more FOG episodes than the short distance condition, even though the task of obstacle avoidance is presumably easier when subjects have more time to prepare. We suspect that obstacle avoidance in the short distance condition engages a fast pathway that runs largely over subcortical structures, releasing motor programs stored in the reticular formation.20–23 Hence, this “cued” fast response could bypass the defective basal ganglia and mesencephalic pedunculopontine nucleus. In contrast, the long distance condition likely involves a longer and more complex circuitry, involving the parietal and motor cortex, the lateral cerebellum and pontine gray nuclei, as was demonstrated in the cat.10 This complex circuitry may be more susceptible to sudden interference, leading to relatively more FOG episodes during the long distance condition.

Second, sudden obstacle avoidance calls for a switch in motor tasks, requiring subjects to make corrective steps instead of walking straight. PD patients have great difficulties switching between motor tasks,18, 24, 25 and this also applies to gait.19, 26 These switching difficulties may be related to deficits in frontal executive functions, which are increasingly implicated in the pathophysiology of FOG.27, 28

Third, increased anxiety may have contributed to the occurrence of FOG. In everyday life, patients report that FOG episodes may be elicited by anxiety and panic.29, 30 The upcoming obstacle during treadmill walking may well increase anxiety. Indeed, one patient froze while merely anticipating the upcoming obstacle and two other patients only showed FOG during the practice trials. However, moderate stress, such as visiting a doctor, may also suppress FOG, which may partly explain the difficulty to elicit FOG in clinical practice.8, 31 In this study, fear of falling, being harnessed, and being forced to keep moving by the treadmill may all have contributed to the state of mind of our patients.

Fourth, obstacle avoidance may have caused FOG by interfering with the mechanically required motor output. Obstacles interfere with step amplitude, forcing the patient to take smaller steps, either to avoid the obstacle, or to regain stability after obstacle crossing (when an extra long step with a more forward movement was used to avoid the obstacle). A decrease in step length could provoke FOG via the presumed “sequence effect,” where a successive step-to-step amplitude reduction ultimately culminates in a motor block.32 In addition, gait rhythm is disturbed by the obstacle. Patients with FOG have a disturbed timing of gait, manifesting as an increased stride variability and phase fluctuations in between FOG episodes,33–35 as well as an uncontrolled timing of steps just before a FOG episode.7 Our finding that FOG can be provoked by interrupting step length and gait rhythm by a sudden obstacle suggests that these motor deficits might provoke FOG, and represent more than just “coincidental” features of advanced disease.

A final explanation is supported by interesting recent work, showing that FOG may be regarded as a consequence of defective integration between preparatory balance adjustments and stepping movements.36, 37 Obstacle avoidance likely requires extra balance adjustments (including lateral weight shifts) while the subject is forced to keep moving. This may particularly explain the FOG episodes occurring immediately after obstacle avoidance.

We should note that while we used sudden obstacles to provoke FOG, a small obstacle on the floor (such as an inverted walking stick) is commonly used by patients as an external cue to overcome FOG.30, 38 One explanation for this apparent contradiction is the unpredictability and perceived consequence of the appearing obstacles. Unlike therapeutic cues, a sudden and unpredictable event needs more attentional resources, and may also lead to increased arousal or stress.

Several limitations of this study need to be mentioned. First, the current “gold standard” of FOG, visual assessment of digitally videotaped gait performance, does not accommodate the crucial but subjective feeling of the feet becoming glued to the floor. Only the more severe FOG episodes on the treadmill indeed generated the feeling of “getting glued.” Hence, we cannot be sure whether we really observed FOG, or merely episodes that looked to experienced eyes very similar to FOG.

Second, we used a fixed speed of the treadmill of 2 km/h. As PD patients display abnormalities in the stride length-cadence relationship depending on the velocity,39 a variable treadmill velocity might have yielded more reliable results. However, to vary treadmill velocity depending on the overground individual velocity is not always feasible. For example, we noticed that the preferred overground individual velocity is often much higher than the preferred treadmill individual velocity. In addition, the optimal “individual speed” in relation to preferred speed for eliciting FOG would still have to be determined. Introducing more experimental conditions might have caused more fatigue, and this could have affected the results. It would be interesting for future studies to evaluate which treadmill velocities are best able to provoke FOG in individual patients.

Third, the duration of FOG episodes was less on the treadmill compared with overground walking. This may be expected as the treadmill does not allow a major slowing since speed is constant. Obstacles and the treadmill itself could have acted as cues to overcome FOG and quickly “abort” the episode.40 This may limit the ecological validity of this approach, as FOG is often defined as lasting at least one second.25 However, there is probably a spectrum of “ineffective stepping” before this truly develops into a FOG episode that is recognizable for both patients and observers. Future detailed electrophysiological assessment of FOG may help to detect such subtle FOG episodes that currently escape the clinical eye.

We showed that the appearance of sudden obstacles can provoke brief FOG episodes during treadmill walking. This ability to provoke brief FOG episodes in a standardized environment, where it can be combined with e.g., kinematic measures, accelerometry, forceplates, or electromyography, offers new opportunities to study the pathophysiology of FOG. Eventually, this may allow for development of an electrophysiological definition of FOG that could support clinicians in diagnosing or evaluating FOG.


Video 1: A typical FOG episode evoked just after obstacle crossing

Video 2: A typical FOG episode evoked between obstacle release and obstacle cross

Video 3: A more severe FOG episode evoked by the threat of a coming obstacle, needing an emergency stop. Hereafter, the patient shows a very subtle FOG episode just before obstacle cross.


The authors thank Roland Loeffen, Cristel van Wezenbeek, and Charlotte Haaxma for their assistance.

Author Roles: A.H. Snijders involved in conception, organization and execution of the research project; design and execution of the statistical analysis; writing of the first draft of the manuscript. V.Weerdesteyn involved in conception, organization and execution of the research project; rating of the obstacle avoidance episodes; review and critique on the manuscript. Y. Hagen involved in organization and execution of the research project; review and critique on the manuscript. J. Duysens involved in conception and organization of the research project; review and critique on the statistical analysis and on the manuscript. N. Giladi involved in conception of the research project; rating of the FOG episodes; review and critique on the manuscript. B.R. Bloem involved in conception and organization of the manuscript; rating of FOG episodes; review and critique on the statistical analysis and on the manuscript.

Financial Disclosure:

Including Stock Ownership medically related, Consultancies, Advisory Boards, Partnerships, Honoraria, Grants, Intellectual Property Rights, Expert Testimony, Employment, Contracts, Royalties and others, A.H. Snijders: Prinses Beatrix Fonds (PBF), Netherlands Organisation for Health Research and Development (NWO); MD-medical research trainee (grant 92.003.490); employment Radboud University Medical Centre (RUNMC); V. Weerdesteyn: RUNMC, Sint Maartens Kliniek (SMK) Nijmegen. Y. Hagen: None. J. Duysens: KU-Leuven, SMK, grants from KU-Leuven (IDO and OT). N. Giladi: Tel Aviv Medical Centre, Consultancies for Teva-Lundbeck, Merck, UCB, Intec Pharma, Advisory Boards UCB and Teva-Lundbeck, Honoraria Teva, GSK, Novartis, Lundbeck, Allergan, expert testimony for GSK. B.R. Bloem: RUNMC, grants from NWO/ZonMw, PBF, MJ Fox Foundation, Consultancy for GlaxoSmithKline, Boehringer Ingelheim, TEVA, UCB, Novartis.