Skip to main content
NCBI home page
European Journal of Physical and Rehabilitation Medicine logoLink to European Journal of Physical and Rehabilitation Medicine
. 2022 Sep 1;58(5):715–722. doi: 10.23736/S1973-9087.22.07313-0

Action observation and motor imagery have no effect on balance and freezing of gait in Parkinson’s disease: a randomized controlled trial

Paula T BEZERRA 1, Lorenna M SANTIAGO 1, 2, Isaíra A SILVA 1, Aline A SOUZA 1, Camila L PEGADO 1, Clécia M DAMASCENA 3, Tatiana S RIBEIRO 1, Ana R LINDQUIST 1,*
PMCID: PMC10019482  PMID: 36052889

Abstract

BACKGROUND

Combining action observation (AO) and motor imagery (MI) training may induce greater brain activity in areas usually involved in Parkinson’s disease (PD) and lead to greater behavioral and neurophysiological effects than when used separately.

AIM

To determine the effects of combining AO, MI, and gait training on balance and freezing of gait in individuals with PD.

DESIGN

This is a single-blinded, randomized controlled clinical trial.

SETTING

Laboratory of Intervention and Analysis of Movement (LIAM) from the Department of Physical Therapy of a Brazilian University.

POPULATION

Study sample consisted of individuals diagnosed with idiopathic PD by a neurologist specialized in movement disorders. METHODS: 39 individuals with PD were divided into experimental (EG=21) and control groups (CG=18). EG performed 12 sessions of AO, MI, and gait training, whereas CG watched PD-related educational videos and performed 12 sessions of gait training. Balance (measured using the Mini Balance Evaluation Systems Test [MiniBESTest]) and freezing of gait (measured using the Freezing of Gait Questionnaire) were reassessed one day after the end of the intervention.

RESULTS

We did not observe significant intra- and intergroup differences in freezing of gait. For the EG, we observed a significant intragroup difference in the total score of MiniBESTest (F=5.2; P=0.02), and sensory orientation (F=4.5; P=0.04) and dynamic gait (F=3.6; P=0.03) domains. MiniBESTest domains were not different between groups.

CONCLUSIONS

Combining AO, MI, and gait training was not more effective than isolated gait training for balance and freezing of gait in individuals with PD.

CLINICAL REHABILITATION IMPACT

MI training can moderate AO effects and enhance motor learning when both therapies are combined. Therefore, this approach may still have the potential to be included in the treatment of PD. New studies should investigate whether the factors that influence these results are related to the protocol’s sensitivity in changing the evaluated parameters or to the time and intensity of AO and MI training.

Key words: Parkinson disease, Postural balance, Freezing, Gait

Parkinson’s disease (PD) is one of the most frequent neurological disorders worldwide. Balance and gait dysfunctions are major therapeutic challenges in individuals with PD unresponsive to pharmacological replacement of dopamine after early stages of this health condition. These symptoms may aggravate with disease progression and impact independence and quality of life.1-3

Gait and balance disorders are observed with distinct patterns and at different stages of PD2. Individuals with PD may show inefficient anticipatory postural adjustments and automatic postural responses.4 Furthermore, some deficits may be evident after an episode of imbalance, including reduced trunk reaction (the trunk moves as a block towards postural disturbance) and slow compensatory stepping and arm reactions.5-7

Gait pattern in PD is characterized by reduced gait speed, step length, rhythmicity, and movement (bradykinesia), increased axial stiffness, inability to control stride frequency (festination), and freezing. Freezing of gait is an important target of investigation since it is highly distressing to individuals with PD.8, 9 The association of these symptoms results in several functional limitations and predisposes individuals with PD to fall, affecting 70% of these individuals at least once a year.10

Since motor dysfunctions affect gait and balance, cognitive processes must compensate and play an important role in gait regulation.11 Action observation (AO) and motor imagery (MI) are cognitive tools used to promote motor skill relearning.12, 13 AO is an effective resource to learn or improve a specific motor skill since performance observation activates the same neural structures responsible for actual performance of movements.14 Therefore, the observer may learn the observed action by internally performing it.15 On the other hand, MI is based on mental development of movements (visual or kinesthetic) without peripheral (muscular) activity.16 MI benefits neurological rehabilitation by increasing the repetitions of an exercise safely and without excessive fatigue, training complex or impossible motor tasks, and allowing patients to practice at the desired time and place.17

Isolated AO may facilitate spontaneous movements in individuals with PD.18, 19 Pelosin et al. (2010) observed reduced freezing of gait episodes after submitting individuals with PD to AO training.20 In contrast, MI effects are ambiguous. Braun et al. (2011) did not identify superior effects of MI on gait performance and mobility compared to physical practice.21 In contrast, other studies demonstrated positive results and recommended using MI for improving dynamic balance and functional aspects of gait.22, 23

MI training may moderate AO effects and intensify motor learning when both therapies are associated. Although neurophysiological aspects of this interaction are not fully understood, different studies indicated that neural circuits involving AO, MI, and movement execution are extensively overlapped.13, 24, 25 To the best of our knowledge, only one study investigated the effects of AO and MI combined in a single experimental session, showing that AO and MI induced motor facilitation in individuals with PD and healthy, respectively.26 Similarly, only one study investigated the long-term effects of combining AO and MI in individuals with PD. After six weeks of training, physical practice preceded by MI and AO reduced postural instability and gait disturbances.27

Based on neural overlap, combining AO with MI training has been recommended for neurological rehabilitation since it can increase the activation of brain areas and generate greater behavioral and neurophysiological effects than their use in isolation.28-30 Thus, the present study aimed to investigate the effects of AO and MI on balance and freezing of gait in individuals with PD. We hypothesized that gait training associated with AO and MI would have superior effects than isolated gait training.

Materials and methods

Study design and location

This is a single-blinded, randomized controlled clinical trial developed by the Laboratory of Intervention and Analysis of Movement (LIAM) from the Department of Physical Therapy of the Universidade Federal do Rio Grande do Norte (UFRN).

Ethical aspects

This study was approved by the research ethics committee of UFRN (no. 2.057.658), registered on clinicaltrials.gov (NCT03439800), and performed according to the Declaration of Helsinki. All patients were informed about the objectives and procedures of this research and signed the informed consent form.

Participants

Study sample consisted of individuals diagnosed with idiopathic PD by a neurologist specialized in movement disorders, following the UK Parkinson’s Disease Society Brain Bank criteria.31 Recruitment was undertaken by convenience, selecting 153 individuals from a list of participants that volunteered in previous research projects of LIAM/UFRN and active search in reference centers in neurological care. Inclusion criteria were individuals with PD at modified Hoehn and Yahr stages 1.5 to 3;32 aged between 45 and 75 years; regular use of antiparkinsonian medication; walk independently for at least 10 meters without any orthosis or gait aid; no cognitive deficit according to the Mini-Mental State Examination (cutoff of 18 points for illiterate and 24 for those with school education);33 ability to imagine motor acts in kinesthetic modality, according to the Movement Imagery Questionnaire-Revised (MIQ-R), indicating at least “neither easy nor difficult to feel” the movement imagined kinesthetically;34 not submitted to stereotaxic surgery; without musculoskeletal or cardiorespiratory impairments affecting gait; and absence of other associated neurological diseases. Those who presented blood pressure >140/90 mmHg after three measurements (five-minute interval between measurements) before or during training,35 uncorrected visual and/or hearing alterations, severe pain and/or discomfort hindering performance of proposed activities, or did not understand the training protocol, were excluded. Forty-four out of 51 individuals assessed for eligibility met full inclusion criteria and were randomized into two groups: control group (CG) and experimental group (EG). After participants allocation, 39 completed the study protocol (CG=18 and EG=21).

Sample size calculation

Sample size was calculated using the open-source program OpenEpi, version 3.01,36 and it was based on hip range of motion of individuals with PD submitted to MI associated with gait training.23 A power of 80% and a 95% confidence interval were considered. Mean and standard deviation of hip range of motion were 54.7° and 7.2°, respectively. Sample size calculation indicated a minimum total sample of 34 individuals. Considering 10% of possible dropouts, the minimum final sample was estimated as 38 individuals.

Evaluation instruments

Sociodemographic, anthropometric, and clinical information were collected using an identification form, and cognitive level was assessed using the Mini-Mental State Examination33 and the Montreal Cognitive Assessment.37 In addition, distinctness of mental image was assessed using MIQ-R;34 motor function and activities of daily living were assessed using the Unified Parkinson’s Disease Rating Scale (UPDRS);38 and level of physical disability was assessed using the Modified Hoehn and Yahr scale.32 Freezing of gait and balance were assessed using the Freezing of Gait Questionnaire (FOG-Q)39, 40 and Mini Balance Evaluation Systems Test (MiniBESTest),41 respectively.

Randomization and confidentiality of allocation

After initial assessment, participants were randomized (http://www.randomization.com) into EG and CG. Allocation was concealed, and a person not involved in the study arranged 44 opaque and sealed envelopes containing the randomization sequence. A researcher opened envelopes individually during the training period for each patient. A second researcher, blinded to treatment allocation, was responsible for pre- and post-intervention assessment and statistical analyses.

Interventions

Training started one day after the pre-intervention assessment, based on protocols used by El-Whishy & Fayez (2013), Santiago et al. (2015), and Nascimento et al. (2019).23, 42, 43 EG was submitted to AO, MI, and gait training. AO and MI training was based on analysis of gait videos and kinesthetic modality, respectively. CG watched PD-related educational videos and performed gait training. Both groups performed three 1-hour sessions per week for four weeks, with a total of 12 sessions. Initially, we recorded gait of participants using a standard camera positioned in two planes (coronal and sagittal), which was also used for gait analysis during AO. Participants were recorded in three different activities: free gait, gait with obstacles, and gait with obstacles associated with cognitive tasks (all in a six-meter course).

EG training consisted of three stages: AO, MI, and gait training. The six recorded videos were used in all sessions of stage one.

Stage one: therapist showed videos of typical gait of a healthy adult in the coronal and sagittal planes using a 21.5-inch monitor. Videos were shown at least twice and could be repeated until participants identified aspects of their own gait that needed improvement. Next, videos of participants’ own gait were shown, and participants were instructed to analyze the gait cycle, so they could understand it, facilitating motor planning and problem identification. The therapist guided important aspects that should be observed. Participants were able to compare their gait with a typical gait and use this comparison as feedback.

During the 12 sessions, complexity of stage one increased. From the first to the fourth session, participants analyzed videos of free gait on a flat ground; from the fifth to the eighth session, participants analyzed videos of gait with obstacles; and from the ninth to the twelfth session, participants analyzed videos of gait with obstacles associated with cognitive tasks.

Stage two: initially, participants performed relaxation. They sat on a chair with backrest, hands on their lap, eyes closed, and breathed slowly and deeply ten times. Subsequently, participants were instructed to perform a lap in the walking circuit described by Nascimento et al. (2019),43 focusing on movement sensations. Execution was timed and used to define the time of each repetition of MI. Next, participants were instructed to sit, and with eyes closed, imagine their gait in kinesthetic mode, attempting to “feel” the movement. During MI, patients should direct their attention to gait components identified in phase one that needed improvement. Three sets of ten repetitions were performed, and each repetition was guided with verbal instructions, with its execution having the same duration as gait training.

During the 12 sessions, complexity of stage two was also increased. From the first to the fourth session, participants imagined free gait on a flat ground. From the fifth to the eighth session, they imagined gait with obstacles (imagining themselves on a busy street with physical barriers). From ninth to twelfth session, they imagined gait with obstacles associated with cognitive tasks (imagining themselves shopping in a supermarket).

Stage three: participants performed three sets of ten repetitions of gait training. Repetitions were alternated with MI (described in stage two) and always preceded by repetition of MI. Gait was trained on flat and firm ground (six-meter long) with participants wearing shoes. Participants performed the training adjusting their gait according to movement aspects that needed improvement identified in previous stages.

During the 12 sessions, complexity of stage three increased. From the first to fourth session, participants walked on the ground free of obstacles and without additional motor or cognitive demands, except for postural and movement pattern corrections. From fifth to eighth session, participants walked on the ground with obstacles (round trip in a different circuit in each session, including zigzag between two cones, cross a 50-cm wide portal, overtake three barriers, ascend and descend steppes and two stair steps, and spin on a pillow). From ninth to twelfth session, participants walked on the ground with obstacles while performing cognitive tasks (speaking out loud names of objects, fruits, and animals with different initial letters and subtracting by three), motor tasks (handling a ball), or both.

CG training consisted of two stages: analysis of PD-related educational videos and gait training.

Stage one: participants watched PD-related educational videos that did not address physical therapy treatments related to gait. Time for watching videos was equal for both groups. AMPARO Network produced the videos (available on https://amparo.numec.prp.usp.br/), which were trimmed to ten minutes and played during the 12 sessions of training.

Stage two: This stage was similar to stage three of EG, except for the alternation with MI. Participants were not requested to direct their attention to their gait, and the therapist did not correct gait and posture patterns. Likewise, gait training was conducted on a flat and firm ground in three sets of ten repetitions each.

During the 12 sessions, complexity of stage two increased. From the first to fourth session, participants walked on the ground free of obstacles and without additional motor or cognitive demands. From fifth to eighth session, they walked on a ground with obstacles (round trip in a different circuit in each session, including zigzag between two cones, cross a 50-cm wide portal, overtake three barriers, ascend and descend steps and two stair steps, and spin on a pillow). From ninth to twelfth session, participants walked on a ground with obstacles associated with cognitive tasks (speaking out loud names of objects, fruits, and animals with different initial letters and subtracting by three), motor tasks (handling a ball), or both.

Outcome measures

Balance and freezing of gait were assessed using the MiniBESTest and FOG-Q at baseline and one day after interventions.

MiniBESTest evaluates postural control deficits in individuals with PD and predicts falls.44 The test contains 14 items divided into four domains: anticipatory postural adjustments, reactive postural control, sensory orientation, and dynamic gait. Total score ranges from 0 to 32 points, and higher scores correspond to increased postural balance. Cutoff point for risk of fall in individuals with PD is 20 points.45 The Brazilian version of MiniBESTest exhibits adequate reliability, response stability, and capacity to distinguish among various balance ability levels in individuals with PD.41 We analyzed the scores for each domain and the total score.

FOG-Q contains six items, and the third item directly evaluates the presence or absence of symptoms. Responses to each item range from 0 (absence of symptoms) to 4 (most severe stage). Total score ranges from 0 to 24 points, and higher scores correspond to more severe freezing of gait.39 The Brazilian version of FOG-Q is reliable and valid for identifying freezing of gait in individuals with PD.40

Statistical analysis

A researcher blinded to treatment allocation performed statistical analyses using SPSS 20.0 (IBM Corp., Chicago, IL, USA). Shapiro-Wilk test verified data normality. Considering dropouts (Figure 1), we performed intention-to-treat analysis. Therefore, data collected for each variable during pre-intervention were repeated in the post-intervention when participants dropped out. Outcome measures were submitted to a 2-way repeated-measures Analysis of Variance (2x2 ANOVA) with intragroup (pre- and post-intervention) and intergroup (experimental and control) comparisons. Significance level of all tests was set at P<0.05.

Figure 1.

Figure 1

Open in a new tab

—Flowchart of the process of sample selection.

Results

Thirty-nine individuals with PD participated in this study: 25 men (CG=11 and EG=14) and 14 women (CG=7 and EG=7). Figure 1 shows the flow diagram of recruitment and dropouts. Table I shows the characteristics of participants (as mean/median and standard deviation/interquartile range) for both groups.

Table I. —Demographic and clinical characteristics of participants at baseline.

Variables Preintervention information
CG
(N.=18)		 EG
(N.=21)		 P
Age (years) 60.7±6.8 64.6±9.3 0.10
Education (years) 11.0 (11.0-16.0) 11.0 (4.5-16.0) 0.37a
Diagnosis period (months) 84.0 (54.0-120.0) 72.0 (60.0-84.0) 0.26 a
MoCA 23.0 (21.5-24.5) 21.0 (17.5-25.0) 0.39 a
H&Y 2.5 (2.0-3.0) 2.0 (2.0-3.0) 0.20 a
MIQ-R – Kinesthetic 22.5 (20.0-24.0) 17.0 (14.5-22.0) 0.01 a*
MIQ-R – Visual 21.5 (16.2-24.2) 16.0 (10.0-21.5) 0.06 a
UPDRS – ADL 14.0 (10.0-23.0) 13.0 (9.0-18.5) 0.39 a
UPDRS – Motor evaluation 27.5 (18.0-41.2) 23.0 (15.5-32.5) 0.23 a
Open in a new tab

Values are presented as mean±standard deviation and median (quartiles 25-75). *P˂0.05. aSignificance determined by the Mann-Whitney Test. CG: control group; EG: experimental group; MoCA: Montreal Cognitive Assessment; H&Y: Modified Hoehn & Yahr scale; MIQ-R: Motor Imagery Questionnaire-Revised; UPDRS: Unified Parkinson’s Disease Rating Scale; ADL: activities of daily living.

No significant differences were found between the groups before the intervention, except for the mental image sharpness score in the kinesthetic modality, assessed using the MIQ-R. In this case, the EG showed less sharpness when compared to the CG.

Eight patients from EG dropped out. Of these, 4 did not complete the protocol due to fatigue, and 4 did not want to continue in the study. Two patients from CG did not wish to continue in the study and dropped out.

Table II shows the mean and standard deviation values for outcome variables at pre- and post-intervention for each group, and the intragroup difference of means and standard deviation.

Table II. —Balance and freezing of gait variables of the experimental and control groups at pre- and postintervention.

Preintervention
mean (SD)		 Postintervention
mean (SD)		 Intragroup difference pre- and postintervention
mean (SD)		 Intergroup difference pre- and postintervention
mean (95% CI)
CG
(N.=18)		 EG
(N.=21)		 CG
(N.=18)		 EG
(N.=21)		 CG
(N.=18)		 EG
(N.=21)		 EG-CG
MiniBESTest
Domain – Anticipatory postural adjustments 5.38±0.5 5.77±0.5 5.23±0.4 6.00±0.5 -0.15±-0.1 0.23±0.0 0.77 (-0.84 to 2.00)
Domain – Reactive postural control 5.90±0.4 6.00±0.5 6.23±0.4 6.22±0.4 0.33±0.0 0.22±-0.1 -0.01 (-1.39 to 1.31)
Domain – Sensory orientation 4.81±0.2 5.00±0.2 5.09±0.2 5.50±0.2* 0.28±0.0 0.50±0.0* 0.41 (-0.40 to 0.99)
Domain – Dynamic gait 7.52±0.4 7.38±0.4 7.66±0.3 8.00±0.4* 0.14±-0.1 0.62±0.0* 0.34 (-1.00 to 1.20)
Total score 23.61±1.3 24.16±1.4 24.23±1.3 25.72±1.4* 0.62±0.0 1.56±0.0* 1.49 (-2.82 to 4.85)
FOG-Q 9.75±1.5 9.33±1.6 8.70±1.5 8.83±1.6 -1.05±0.0 -0.5±0.0 0.13 (-4.32 to 4.60)
Open in a new tab

*P˂0.05 in intragroup difference pre- and post-intervention. SD: standard deviation; 95% CI: 95% confidence interval; CG: control group; EG: experimental group; MiniBESTest: Mini Balance Evaluation Systems Test; FOG-Q: Freezing of Gait Questionnaire.

In addition, Table II shows the effect size described as post-intervention intergroup difference. Repeated-measures ANOVA did not demonstrate significant intra- and intergroup differences in freezing of gait symptoms. Regarding the analysis of MiniBESTest domains, ANOVA revealed a significant difference within the experimental group in the sensory orientation (F=4.5; P=0.04) and dynamic gait (F=3.6; P=0.03) domains, also reflected in the difference for total score for this group (F=5.2; P=0.02). However, no difference was observed between groups for any MiniBESTest domain.

Discussion

This study investigated the effects of combining AO, MI, and gait training on balance and freezing of gait in individuals with PD. Contrary to our hypothesis, our results revealed that combining these three interventions was not more effective than just performing gait training.

Freezing of gait occurs when the intention to move, associated with adjusting the movement to external situations or internal motor commands, induces a system obstruction. However, the nature of freezing of gait and its mechanisms are still unclear. Literature suggests different theoretical models of freezing of gait are correlated with different observations in clinical environment.46

The threshold model is based on accumulation of motor deficits until reaching a threshold, leading to freezing. Therefore, increased cadence and complexity of coordination or decreased step amplitude may cause freezing of gait.47 The interference model explains freezing as a temporary collapse of simultaneous processing of cognitive and limbic information during motor tasks that can be induced by increasing the number of concurrent tasks and their difficulty.48 The cognitive model is a deficit of conflict resolution in situations requiring decision making. Imposition of fast decisions and incongruence level would also induce freezing.49 Last, the decoupling model characterizes freezing as a disconnection between pre-planned motor programs and movement execution, such as a step at the beginning of gait, inducing freezing.7

Freezing of gait in individuals with PD can occur in different contexts. However, valid instruments developed to assess freezing of gait do not account for different contexts. Therefore, combining AO and MI may have benefited one of the freezing models described above if the evaluation instrument had considered precisely the freezing nature. In addition, based on the mean score of each group, participants included in this study had mild to moderate freezing of gait. The protocol used was probably not effective in changing this outcome significantly.

Regarding balance, the proposed intervention (combining AO, MI, and gait training) did not improve postural balance more than the control intervention (gait training) in any of the four domains of the MiniBESTest. Significant difference was observed only within EG. The central nervous system of individuals with PD struggles to process vestibular, proprioceptive, and visual afferences, resulting in inadequate muscle responses and postural imbalance. In addition, individuals with PD have difficulty with movement automatization, increasing attention demand during daily activities and challenging the association of cognitive and motor tasks.50 For that reason, the protocol used in this study may have been insufficient to meet demands of postural balance. Future studies should adapt our protocol, targeting this outcome to investigate whether differences would be found post-intervention.

It should also be considered that, despite the randomization and secrecy of the allocation, the EG presented less sharpness than the CG to imagine motor acts. This lower antecedent capacity may have influenced the results of this study, since the greater image capacity of an individual is more effective in taking advantage of training with MI and that the training of AO and MI can improve the ease with which the mental image is generated.51

Although studies using MI in individuals with PD are scarce, some outcomes similar to ours can be found in the literature. Braun et al. (2011) did not identify superior effects of MI compared to relaxation techniques, both in association with gait training, after six weeks of intervention in individuals with PD.21 In addition, Santiago et al. (2015) did not find superior effects of MI on gait of individuals with PD compared with physical practice after a single training session.42

On the other hand, positive results were observed in patients with PD undergoing AO training, with a reduction in the number of gait freezing episodes.20 Favorable outcomes were also obtained in studies that investigated the effects of MI training on PD, characterized by reduced bradykinesia,22 improved mobility, gait speed and biomechanical gait parameters,23 in addition to improving postural instability and gait disturbances through of the combined use of MI and AO.27

MI is a relatively new approach to treat individuals with PD, and it needs to be adjusted according to specificities and skills of individuals and different contexts.46, 52 Since we did not find superior effects of our protocol on balance and freezing of gait, further studies should increase the period and intensity of training and develop a specific protocol for the evaluated outcomes. Moreover, instruments that allow the evaluation of freezing of gait in different contexts are needed since this was a limitation of this study. As a result, identifying beneficial interventions for a particular freezing model could be elucidated, allowing a more specific rehabilitation approach for individuals with PD.

Limitations of the study

This study also had other limitations, such as participant dropout, lack of control if treatment was applied under on or off drug condition, and sample calculation based on hip range of motion, which may not detect differences in freezing of gait and balance. The association of these factors may have interfered with results and should be considered in future studies.

Conclusions

Adding AO and MI to gait training did not modify parameters of balance and freezing of gait in individuals with PD. However, this approach might still have potential to be included in PD treatment. Further studies should investigate whether the factors influencing these results are related to protocol sensitivity in changing the evaluated parameters or to time and intensity of AO and MI training.

References

Articles from European Journal of Physical and Rehabilitation Medicine are provided here courtesy of Edizioni Minerva Medica S.p.A.

RESOURCES