2.1. Participants
Thirty-five participants diagnosed with Parkinson's were recruited through Parkinson’s UK advertisements and local Parkinson’s UK branches. All participants experienced regular freezing of gait (responding with a score of 3 or greater to the third item in the Freezing of Gait Questionnaire; (45)). Participants were excluded from the study if they had cognitive deficits (MiniCog score of < 3; (46)), or reported any musculoskeletal or neurological issue (other than Parkinson's) that significantly affected their walking. Prior to testing, all participants self-reported that they were able to stand unsupported for at least 60 seconds. While all participants typically experienced FoG at least once a day, they were divided into two groups based on whether they exhibited at least three recorded freezes within each condition described below (FoG-F, n = 17). For those participants exhibiting fewer (n = 3) or no (n = 15) freezes (FoG-NF, n = 18), analyses were carried out on successful steps from a voluntary stop. For the FoG-F group, data from the Baseline condition were entered into an initial comparison between successful and unsuccessful steps from a freeze. When comparing outcomes between experimental conditions, only data concerning successfully initiated steps were included. This approach was determined a priori, to afford direct comparisons to the FoG-NF group. Furthermore, such statistical comparisons were not feasible due to an insufficient number of unsuccessful attempted steps occurring during Verbal and Analogy trials.
The Unified Parkinson’s Disease Rating Scale motor section (UPDRS-III) was administered by a certified researcher (see Table 1).
Table 1
Participant characteristics
| Non Freeze (n = 18) | Freeze (n = 17) | P |
Age | 66.5 ± 9.24 | 69.9 ± 9.9 | .313 |
Sex (M/F) | 10/8 | 10/7 | .542 |
Years since diagnosis | 9.6 ± 7.5 | 8.6 ± 6.1 | .685 |
MoCA (0–30) | 26.8 ± 4.8 | 27.1 ± 1.8 | .266 |
FES (0-100) | 65.8 ± 19 | 62.1 ± 19 | .654 |
FOGQ (0–24) | 15.1 ± 2.9 | 16.3 ± 4.7 | .337 |
UPDRS III (0-108) | 19.13 ± 3.7 | 22.9 ± 5.3 | .062 |
H&Y Stage (1–5) | 2.3 ± 0.7 | 2.8 ± 0.9 | .123 |
2.2 Experimental Design And Task
The testing session was carried out after participants received their usual dopaminergic medication (i.e., the ‘ON’ state). All testing was completed within a single session. Participants were instructed to step in place (SIP) (i.e., alternately lift each foot from the ground without progressing in any direction) on a single force plate (dimensions = 600 × 400 mm; Kistler Group, Switzerland) and to keep stepping for 60 seconds or until a freeze occurred that resulted in the cessation of stepping. Participants were given the prior instruction that, if a freeze occurred, they should attempt to initiate forward walking using either foot. If no freeze occurred that led to a cessation of stepping within the 60 second trial, after 60 s of stepping, participants were asked to “stop” and “start walking forwards”. Participants were given the latter instruction prior to the start of each trial in an attempt to avoid the use of verbal cues related to gait initiation during each trial. At the start of the session, participants were instructed to practice stopping the SIP task and initiating forward walking at least three times in order to familiarise themselves with the protocol. Stepping trials were repeated up to six times (or until four attempted steps from FoG events were recorded) for each of three different conditions; Baseline, Verbal and Analogy.
Participants were securely fitted into a safety harness for all stepping trials and were given the opportunity to stop, sit down and have refreshments at any point between trials. Each stepping trial was completed once the participant had made an attempted step forward (successful or otherwise). At the end of each trial, participants were asked to return to their original position on the force plate (if necessary) with the help of a researcher to begin the next trial.
2.3 Inducing Freezing Of Gait
Inducing FoG in laboratory settings represents a significant challenge. Therefore, we integrated two procedures that have previously been shown to expedite FoG. First, the SIP task has been repeatedly shown to successfully induce FoG (44, 47). This task also carries the additional benefit that participants’ movements can be constrained so that, if FoG occurs, participants will be positioned on a force plate, thus permitting the necessary recordings. If the SIP task did not successfully induce FoG, participants were invited to suggest alternative strategies. Consequently, three participants in the FoG-F group experienced—and attempted to initiate forward gait from—freezes induced by a combination of SIP and turning from on the spot. Here, only trials were used when the freeze resulted in participants facing in the same direction as the intended walking path (to maintain task consistency between participants).
Second, participants wore a virtual reality head-mounted display (HTC Vive, Sony Ltd) displaying one of two environments designed in Unity3D (Unity Technologies) to induce freezing by presenting scenes commonly observed to exacerbate FoG (12) and, similarly, through elevating anxiety (6). These environments depicted participants standing either at the top of a set of descending stairs with no handrail or in front of an open narrow doorway with tables and chairs on the other side. Participants were asked to select an environment to use throughout the study, based on which, they felt, was most likely to induce FoG. Prior to stepping trials, participants were given time to familiarise themselves with the VR environments in both seated and standing positions.
2.4 Baseline Trials
In the Baseline condition, prior to the start of each trial participants were given the instruction that when they stopped stepping (be it following a freeze or voluntary stop), that they should start walking forwards in their preferred manner (the use of physical devices to assist gait initiation was prohibited).
2.5 Verbal Condition
Participants were trained to use a weight shifting strategy that was based on a simplified description of an APA. During training, participants were invited to practice “Moving [their] weight sideways slightly towards the stepping leg, then swaying back and shifting all [their] weight on to the non-stepping leg before stepping forward”. Demonstrations were given by the researcher during training and participants were shown a printed image exemplifying an APA for reference (Fig. 1). Participants practiced the movement up to ten times, and in doing so, were asked to concurrently verbalise the instructions to move “right, left and forwards” (for those initiating gait with their right foot) or as a number sequence “One, two and three”. This allowed the researcher to confirm that participants understood the instructions and to give participants the opportunity to ask any questions about the Verbal strategy. During the Verbal condition, when participants stopped stepping (due to a freeze or voluntary stop) they were given the prior instruction to use the newly-learned verbal strategy to initiate forward walking.
2.6 Analogy Condition
Participants were trained to use a weight shifting strategy equivalent to the Verbal strategy, but in the form of a movement analogy. Participants were shown a printed imagine exemplifying an APA (same as Verbal condition) and the researcher gave both a description and demonstrations of the required weight-shifting movement, being careful not to give simplified verbal cues. The researcher then explained the concept of using movement analogies as a strategy to produce the required weight shift and initial forward step. Participants were given some examples of analogies to practice moving to. For example, “imagine you are stood on a set of traditional balancing scales, when one foot presses down, the other foot lifts up and makes a step”. Other examples related to swaying like a tree in the breeze or like a slalom skier. Participants were also encouraged to create their own personalised analogies in an attempt to make the process more meaningful and memorable to the individual (e.g., (48)). Thirteen participants opted to create and use their own analogies (FoG-F: n = 7/17, FoG-NF: n = 6/18). These included shifting weight like: i) a rugby player performing a ‘dummy’ turn, ii) a tennis player waiting to receive a serve, iii) a boxer moving towards an opponent, and iv) a person standing on a moving boat. During the analogy trials, when a freezing episode occurred or the researcher instructed the participants to stop, participants thought of their chosen analogy to perform the weight shifting movement to make a step forward.
Participants practiced shifting weight/initiating a step using their chosen analogy up to ten times. During the Analogy condition, when participants stopped stepping (due to a freeze or voluntary stop) they were given the prior instruction to use the newly-learned Analogy strategy to initiate forward walking.
The order of Verbal and Analogy conditions was counterbalanced. However, participants always completed the Baseline condition first in order to avoid carry-over effects (i.e., participants using any strategies learned in the other conditions during Baseline trials). Immediately following the final trial in each condition, participants were asked to retrospectively describe their thoughts when attempting to initiate forward walking in preceding trials. Responses were audio recorded, transcribed verbatim and allocated into one of four categories: i) Focus on environment, future planning or global aims (e.g., “thinking about doorway/stairs”, “counting how many steps to reach the doorway”, “just move”); ii) Instructions specific to foot movement (e.g., “pick your foot up”, “step forward and [put] heel down first”); iii) Instructions specific to weight-shifting (e.g., “counting 1, 2, 3 [to shift weight as instructed], “move right, left, then forward”); and iv) Analogies relevant to weight-shifting. We counted the number of participants reporting thought processes in each category. These data were not intended for any statistical analysis, but rather as a broad evaluation of the fidelity of each condition/manipulation.
2.7 Data Collection
A nine camera motion analysis system (VICON, Oxford Metrics, London, England) was used to reconstruct the position of reflective markers on the lateral malleolus, calcaneus and the head of the fifth metatarsal of each foot at a sampling rate of 100 Hz, and filtered with a fourth order Butterworth low pass filter with a cut-off frequency of 5 Hz. A single force plate was used to record medio-lateral (ML) centre of pressure (COP) motion with a sampling frequency of 1000 Hz. All analyses involving signal processing were carried out using bespoke scripts in Matlab (Mathworks Inc.), where the researcher was blinded to the condition pertaining to each trial. In all cases, data were plotted and visually checked to avoid artefacts.
2.8 Identifying A Freeze
A freezing of gait episode was defined both subjectively and objectively. Stepping trials were visually identified through playback of steps recorded on a video camera and outputs from the motion capture system. Audio recordings from this video were also used to document participants’ verbal reports of any freeze that they perceived. A FoG episode was defined as an episode of involuntary cessation of gait which was often accompanied by trembling of the legs of festination and when the vertical and anterior displacement of the toe marker dropped between zero and one standard deviation of the initial value of that trial (49). If FoG was subsequently confirmed through this initial process, motion capture and force plate data relating to each step during the trial were exported and analysed.
2.9 Calculating Medio-lateral Weight-shifting Amplitude, Duration And Stance Width
The onset of an APA (first measurable change in ML COP from freezing episode or stopped position) was detected by an automated threshold-based algorithm, with the threshold set at three standard deviations of the ML COP displacement of the freezing episode, or stopped position prior to making a step of each trial. ML-WS amplitude was assessed by analysing the peak ML displacement of the COP from the midpoint between the heel markers on each foot to the maximum lateral shift towards the stance foot (Fig. 2). ML-WS duration was the time of ML-WS whilst stance width was the distance between heel markers on each foot whilst in the freeze or stopped position before making a step.
2.10 Assessment of failed attempts to step from a freeze
For trials where participants verbally reported having experienced a freeze, we evaluated whether subsequent attempts to initiate forward walking from that freeze was successful or not (FoG-F group only). Video recordings of each trial were viewed and rated by three independent evaluators who were blinded to each condition. Each evaluator made categorical judgements about whether the participant, in their initial attempt, appeared to successfully or unsuccessfully initiate a forward step (50). Any discrepancy between evaluators resulted in a communal viewing of the relevant trial, a discussion regarding the rationale for the decisions made and a consensus reached. Throughout the entire dataset, this process was only necessary for 1 trial. Again, evaluators remained blinded to the trial condition throughout this process. We calculated the mean ratio of successful to unsuccessful attempts.
2.11 Analysing ML weight-shifts during unsuccessful attempts to step from a freeze
Failed attempts to initiate forward walking from a freeze (as identified by the independent evaluators) were timestamped on the video recordings and corresponding motion capture and force plate data for that time period were exported and analysed. To determine ML-WS for these unsuccessful attempts to step, we calculated the ML displacement of the COP from the midpoint of the feet to the maximum lateral shift towards the stance foot during the period immediately preceding the observed start hesitation. For two participants in the FoG-F, it was not possible to reliably determine the point of attempted step initiation in any trials. Therefore, comparisons of ML-WS between successful and unsuccessful attempted steps were carried out on data from the remaining 15 FoG-F participants.
2.12 Statistical Analysis
The first aim of the study was to evaluate associations between ML-WS amplitude between: i) successful and unsuccessful attempts to initiate a step from a freeze (within-subject comparison in FoG-F group), and ii) FoG-NF and FoG-F groups at Baseline (i.e., successful attempts to initiate a step from a voluntary stop and freeze, respectively. We conducted a paired-samples t- test to compare ML-WS between successful and unsuccessful in FOG-F during Baseline trials. Effects sizes are expressed as Cohen’s d. An independent samples t-test was used to compare participant characteristics between FoG-F and FoG-NF groups. We also carried out Spearman’s Rank Order correlations to evaluate possible associations between ML-WS amplitude when stepping at Baseline (where participants were not attempting to directly manipulate APA characteristics), disease severity (UPDRS III and H&Y Stage) and self-reports of freezing severity (FOGQ).
The second aim of the study was to compare: i) ML-WS amplitude and ii) the proportion of successful attempts to step from a freeze between a Baseline condition, and two experimental conditions where participants attempted to use either a Verbal or Analogy strategy to produce an APA and subsequent step initiation. A mixed design ANOVA (2 × 3 design) was performed for ML-WS, ML-WS duration and stance width, with group (FoG-F and FoG-NF) as a between-subject factor and stepping conditions (Baseline, Verbal and Analogy) as a within-subject factor. Significant (p < 0.05) effects were followed-up with Holm-Bonferroni-corrected post-hoc tests and effects sizes are expressed as partial eta2 (ANOVA) or Cohen’s d (t-test). Wilcoxon signed-rank tests were used to compare the ratio of successful to unsuccessful steps between conditions in the FoG-F group and effect sizes are expressed as