Two weeks of twice-daily prism adaptation treatment does not improve posture or gait in Parkinson’s disease: a double-blind randomized controlled trial

The study had a double-blind randomized controlled design with two parallel groups. The intended allocation ratio was 1:1 (real:sham); however, the final allocation ratio was 1.38:1 due to a deviation from the intended allocation procedure (see the “Randomization and allocation” section).

A summary of the study sessions and procedures is provided in Table 1. The participants attended two research visits lasting 3–4 h at the Dartmouth-Hitchcock Medical Center (the “pre” test and “one day post” test; 1d post). In the 14 days between these sessions (the “treatment period”), participants completed twice daily prism/sham adaptation at home. In the first research visit, the participants underwent a neurological and cognitive assessment to confirm eligibility for the study, and provided informed written and verbal consent. In both research visits, participants completed self-report questionnaires, and their posture and gait were assessed using standard clinical tests and mechanized measurements. Symptoms were also assessed 1 week (“one week post”; 1w post), 1 month (“one month post”; 1m post), and 3 months (“three months post”; 3m post) after the end of treatment using the self-report questionnaires, sent by post. The participants, and the researchers responsible for participant enrolment and assessing outcomes (PRLeB, CAJ, DP, and SLL) were all blinded as to which type of treatment each participant underwent. The trial ended when all enrolled participants had either returned the final self-report questionnaires, withdrew, or were deemed lost to follow-up.

Participants with idiopathic Parkinson’s disease were recruited between April 2015 and February 2017 from among patients referred to the Balance and Vestibular Centre of the Dartmouth-Hitchcock Medical Center for gait training (Fig. 1). The target sample size was 32, based on pragmatic considerations of the number that could be recruited and tested given the available time and resources, and sample sizes for two-branch studies about prism adaptation treatment in stroke patients [11‐13]. We recruited 31 patients into the study in the time available. Inclusion criteria were being aged 40–85 inclusive and meeting any one of three clinical criteria: (1) Hoehn and Yahr Stage II.5–IV; (2) scored > 0 on any of the gait, freezing of gait, and postural stability items of the Movement Disorder Society Unified Parkinson’s Disease Rating Scale [14] (MDS-UPDRS Items III.10–III.12); or (3) Timed Up and Go test duration > 12 s. Exclusion criteria were (1) known psychiatric or neurological condition other than Parkinson’s disease that in the opinion of the assessing Neurologist (SLL) would compromise participation in the study; (2) known neurological diagnosis, other than Parkinson’s disease, that can cause imbalance and gait impairment (e.g., multiple sclerosis, stroke); (3) normal score on the MDS-UPDRS part III; (4) injury or impairment to the right arm that would substantially affect pointing movements (other than that due to Parkinson’s disease); (5) lacking sufficient visual acuity to view the target and hand during prism adaptation; and (6) lacking sufficient understanding of verbal and written English to give informed consent.

Participants who were taking CNS-acting medications (e.g., benzodiazepines, hypnotics, antidepressants) were on a stable regimen for at least thirty days. One participant withdrew because of illness unconnected to the intervention, and another because they were unable to attend visit 2.

A neurologist with over 15 years’ experience in assessing and managing Parkinson’s disease (SLL) conducted a neurological assessment at the pre-test. The assessment included (1) a standard neurological examination, (2) the Montreal Cognitive Assessment (MoCA) [15], and (3) evaluation of Parkinson’s disease severity using the motor component of the MDS-UPDRS (Part III) [14]. Participants also completed the Patient-Centered Outcomes Questionnaire for Parkinson’s Disease [16] at baseline so that the perspectives and expectations about possible treatment success could be compared between groups, since these can impact on trial outcomes.

The effect of prism adaptation on postural control (primary outcomes) and gait (secondary outcomes) was measured by a physical therapist with over 20 years’ experience in vestibular rehabilitation and falls prevention (DMP), including daily use of the described outcome measures as part of her standard clinical practice.

The intervention involved making pointing movements to visual targets while wearing goggles fitted with Fresnel lenses that induced a 25-dioptre (~ 17°) upward shift in the visual image (“real” treatment), or that induced no shift (“sham” treatment). At the end of the first research session, a researcher who was not involved in assessing baseline symptoms or outcomes (JHB or FPA) trained the participant in the treatment and guided them through one treatment session. Regardless of treatment, the researcher stated that the goggles distorted vision, “for example, they restrict your field of vision.” Participants were seated in a chair within arms’ reach of a wall while wearing the goggles. A sheet of US letter-sized paper (the “target sheet”) was secured to the wall in portrait orientation, upon which were two 2-cm-diameter targets spaced 25 cm apart, one above the other. Participants were instructed to use the index finger of their right hand to touch each target alternately, bringing their finger to their chest between each pointing movement. They were instructed to perform the pointing movements as fast as possible with minimal head movements for 50 total movements (25 per target), and then to remove the goggles.

After the participant completed the treatment and the researcher was satisfied that they understood the instructions, the participant was provided with the target sheet, written instructions, a weblink to a video demonstration of the treatment session (as performed by an actor), contact information for the researcher, and a logbook. Participants were instructed to perform the treatment at home twice a day (morning and evening), starting with the morning of the following day. Participants could perform the treatment alone or with a caregiver, according to their preference. They were asked to record in the logbook the time and duration of each treatment, and their perceived levels of attention, fatigue, and effort during each treatment. Participants were asked not to alter any other treatments or rehabilitation techniques that they were engaged with during the course of the study period, however aside from medications (Supplementary Tables 1 and 2), we did not record any co-interventions participants were using (e.g. physiotherapy and strength training). Aside from the logbook entries, there were no other efforts to monitor adherence during the intervention period, although we emphasized that the participants could contact the researchers at any time with questions.

The participants were allocated to a treatment group during the first research visit by the researcher who trained them in performing the treatment. The participants were randomized to undergo real or sham prism adaptation through a sealed envelope-style blind draw. At the beginning of the study, sixteen of each type of goggles were placed by JHB in individual opaque zip-lock bags and mixed together in the same box. To allocate each patient, the researcher responsible for treatment training (JHB or FPA) extracted a set of goggles from the box. It was intended that this allocation procedure would be used until each set of goggles had been provided to a participant, resulting in two groups of 16 patients. However, some goggles were returned to the box after previous participants were finished with them. That is, instead of the goggle allocation occurring without replacement of previously used goggles into the box, the allocation occurred with replacement. This resulted in 18 patients undergoing real adaptation and 13 undergoing sham adaptation. Of these, two patients allocated to the real adaptation condition withdrew from the study (Fig. 1).

The groups had comparable baseline patient characteristics (Table 2; individual patient characteristics provided in Supplementary Tables 1 and 2 in Additional file 2). There were no baseline differences on the New Freezing of Gait questionnaire; Falls Efficacy Scale; Activities-specific Balance Confidence Scale; mobility, Activities of Daily Living, or total scores on the Parkinson’s Disease Questionnaire 39; or the Patient-Centered Outcomes Questionnaire for Parkinson’s Disease (ps >.094). The groups did not differ in the number of training sessions self-reported as completed according to their log books [sham group: M=27.0, SEM=0.40; real group: M=26.4, SEM=0.68; t(27)=0.7, p=.462)]. The groups also did not differ in self-rated attention, fatigue, and effort during treatment (ps ≥ 0.217). There was no difference in the percentage of people in each group who chose “sham,” “real,” or “no idea” when asked to guess the treatment that they had been assigned to in the post-treatment session [sham group: 54%/31%/15%; real group: 37.5%/37.5%/25%; Χ² (2) = 3.8, p=.149]. For those patients who guessed sham or real, the confidence (%) of their guesses were not different between the groups (guessed sham: M=68.8, SEM=4.5; guessed real: M=78.5, SEM=6.3; t(19)= 1.3 p=.216). No participants reported any harms or unintended effects.

Questionnaire return rate was higher than 87% across all five testing points, and higher than 97% for the pre and 1d post test. Of the questionnaires that were returned, missing data due to incomplete items were negligible (9 data points across five questionnaires). Missing data were replaced with the group mean for that item. One patient (VA32) arrived late for the 1d post test, leaving insufficient time to run assessments for the Berg Balance Scale, Sensory Organization Test, and Motor Control Test. Since partial post-treatment data were available for this patient, their missing values were replaced with the group mean. Additional missing data constituted less than 5% of the data and were replaced with group means.

Figures 2 and 3 illustrate the results of the mechanized measures, and other physiotherapist assessments, respectively. The results, reported in detail below, revealed almost no differences between the real and sham treatment groups on treatment outcomes. The exception is that the real treatment group had a significant increase in maximum velocity for forward stepping after treatment (based on the Limits of Stability test), whereas the sham treatment group had no such change. Overall, the weight of evidence supported no significant change in posture or gait in either group.