ParkinSong Online: Feasibility of Telehealth Delivery and Remote Data Collection for a Therapeutic Group Singing Study in Parkinson’s

Abstract

Background

Parkinson’s disease can negatively affect vocal functioning and social wellbeing, particularly in the latter stages of disease progression. Face-to-face group singing interventions can improve communication and wellbeing outcomes, yet not all people can access in-person sessions. To help overcome barriers to participation, exploration of the feasibility and utility of online therapeutic singing programs is needed.

Objectives

To evaluate the feasibility, acceptability, and preliminary efficacy of a 12-week ParkinSong Online intervention on speech and wellbeing for people with Parkinson’s disease.

Methods

A total of 28 participants with idiopathic Parkinson’s disease were recruited to a single-arm feasibility study. Weekly 90-minute online sessions were co-facilitated by a music therapist and speech pathologist. Speech and wellbeing assessments were conducted pre and post intervention. Participant and facilitator surveys were administered after each session, with focus group interviews at the end of the program.

Results

The recruitment rate was high (90%) with no attrition, adverse events, or safety issues. There was good intervention fidelity, attendance (average 89%), and positive participant experience. Feasibility was good, with technology reported as the main challenge (connecting and navigating Zoom). No improvements were seen in voice measures or wellbeing outcomes in this small trial. The online format used in this study did not provide the same benefits as in-person ParkinSong sessions.

Conclusions

ParkinSong Online is feasible for recreational purposes and social engagement provided that people have adequate technological knowledge or support. The optimal online delivery format to achieve communication improvements in Parkinson’s awaits confirmation.

Keywords

Parkinson’s disease singing speech voice online telehealth feasibility dysarthria rehabilitation

Introduction

People living with Parkinson’s disease, a progressive neurological disorder affecting movement and cognition,¹ frequently experience motor symptoms affecting their voice and breathing that impact their psychosocial wellbeing.^2-7 To ameliorate these symptoms, in-person group singing interventions have been advocated, with research demonstrating improvements in vocal loudness and voice-related quality of life.^8-12 Such interventions are based on neuroscientific evidence showing that singing shares neural circuitry with speaking,⁷ and demands greater respiratory support and vocal effort than speaking.¹³ Considering restrictions to movement during the COVID-19 pandemic, and in response to geographical or economical barriers to accessing in-person singing interventions, there is a pressing need to investigate alternative delivery formats, such as using online platforms, to maintain and improve accessibility of therapeutic group singing programs for people with Parkinson’s and related disorders.¹⁴

Telehealth and virtual clinic consultations for people with Parkinson’s have been found to have similar benefits to in-person visits for a range of activities, such as exercise,¹⁵ dance therapy,^16,17 and cognitive-behavioral therapy,¹⁸ with the added value of reducing time and financial costs.^19,20 Increasing use of telehealth has also meant that Parkinson’s measures typically administered in-person needed validation for online administration.^21,22

There has been little research on online or remote delivery platforms for group singing interventions for people with Parkinson’s. A pilot study used pre-recorded vocal and singing exercises for 2 Parkinson’s groups in rural areas.²³ In this study 10 participants met together in local community venues where recordings were projected onto a large screen for participants to follow. A similar pattern of results was achieved to a parallel in-person group singing intervention.²⁴ Another small randomized trial reported a reduction in dysphagia symptoms following a combined singing and speech therapy program for 35 individuals with Parkinson’s delivered via telehealth.²⁵

Recent surveys have shown that many different formats for online choirs were used during the COVID-19 pandemic and that similar themes emerged for therapeutic and non-therapeutic choirs alike, particularly surrounding technology issues.^26,27 The primary limitation for online group singing is latency, or lag time, necessitating the use of the mute function while participants sing along with a facilitator. This inherent latency in videoconferencing platforms makes synchronous group music-making difficult to achieve online. Solutions for reducing online latency are emerging, but these currently require additional equipment and technological set up that is too complex for many novice users. While there is evidence that in-person singing groups have therapeutic value for diverse populations²⁸ including those with Parkinson’s,^29,30 stroke,^31,32 mental health conditions such as anxiety and depression,^33,34 dementia,^29,30 and chronic obstructive pulmonary disease,³⁵ there is little evidence for functional, social, or health-related benefits of online singing groups. The feasibility and potential benefits of an interactive online group singing approach, where participants join from individually convenient locations (eg, home), with pre and post assessments also completed online, have not been explored. It is important to explore the feasibility of remote data collection, both from an equity perspective (ie, allowing recruitment of geographically remote participants and reducing participant burden to travel for assessments), as well as to reduce risks related to infection control with vulnerable populations (particularly in the context of COVID-19). Remote data collection also provides an opportunity to minimize missing data. The growing need for alternative and accessible delivery models, in combination with the lack of research evidence, justify the need to explore online models for therapeutic singing groups. The aim of the current study, therefore, was to investigate the feasibility, acceptability, and preliminary efficacy of a 12-week online singing intervention (ParkinSong Online) and the feasibility of completely online data collection.

Methods

The ParkinSong Online feasibility study was designed to investigate the viability of adapting the usual in-person ParkinSong group intervention into an online format during the COVID-19 pandemic. The target sample size was 30 participants (2 groups of 15). Both groups received an identical experimental protocol, with separation into 2 groups due to maximum size limits for therapeutic and practical purposes only. Eligibility criteria included: a diagnosis of idiopathic Parkinson’s; no previous neurological, head and neck, or respiratory disorders; no visual or auditory impairment not remedied by aids; English speaking, a computer with high-speed internet (≥10 mbps download; 1 mbps upload) and web camera; available for the dates of the study; and Montreal Cognitive Assessment³⁶ (MoCA) score ≥ 18 (or if between 10 and 17, a carer was required to assist the participant for each session). Initial phone screening for eligibility, safety, and environmental assessment was conducted prior to obtaining informed consent from all participants via REDCap,³⁷ an electronic data capture tool hosted at the University of Melbourne. Ethics approval was provided by the University of Melbourne Human Research Ethics Committee (2021-14465-16053-3); and a peer reviewed protocol was published prior to commencement of data collection.¹⁴ Due to recruitment challenges, 1 change from the published protocol paper was the relaxing of the required condition of “reported changes in communication” (self-reported or reported by their caregiver or health professional); nevertheless 81% (n = 25) of participants still met this criterion. Recruitment was conducted through Fight Parkinson’s with the study promoted via online advertisements, email distribution, and referrals from local speech pathologists. Following initial telephone screening and consent, a time was scheduled to complete stage 2 of the screening (online administration of the MoCA, and the baseline assessment).

Measurements

The main measure of intervention effect was speech loudness during conversational speech (ie, monolog). The hardware configuration was piloted and calibrated using a sound pressure level meter. This was then piloted in a patient to ensure compliance. Reliability and consistency of metrics of interest were examined and confirmed prior to data collection. We also assessed speech loudness in terms of sound pressure levels during reading (Rainbow passage), reading over background noise (crowd noise played through headphones), and an instructed loud phrase. The background sound file intensity was set at 85 dB during the recording process. All participants were provided with headset microphones (Microsoft LifeChat LX-3000 USB microphone) containing in-built headphones to control the proximity of sound source to receiver (ear). Other voice outcomes included: maximum phonation time, syllable repetition (diadochokinetic rate), and composite scores of intelligibility, naturalness, and disease severity.³⁸ Composite scores were calculated using a combination of acoustic features representing vocal control, voice quality, and timing derived from tasks within the speech battery. The composites are proprietary of Redenlab and have been validated against perceptual ratings conducted by 2 expert raters blinded to disease and severity (N > 1000). Speech and voice data were captured at baseline and follow up time points using Redenlab Online™ remote data collection software (via participant computers) with a headset microphone (at a mouth-to-microphone distance of 10 cm) connected to the participant’s computer. Data were sampled at 44.1 KHz, 16-bit quantization and saved via lossless .wav format.

Adherence to study requirements was optimized through several approaches, including standardised hardware, software, training, and stimuli. Hardware was standardized via a USB-connected head-mounted microphone. This was selected to mitigate potential challenges of bring-your-own-device protocols and variable mouth to microphone distances (important for loudness outcomes). USB-based microphones bypass inbuilt sound cards, converting analogue signals to digital signals before they reach the device. This plug and play approach is easy to use. Participants received training on the study protocol and a mock run was elicited prior to formal testing. Speech tasks are brief, can be easy to produce, stable in the absence of true change, and are used across multiple neurological populations.^39-41 Recording protocols were designed and implemented based on recommendations from multiple method studies and expert reviews.^42-46

Quality of life and wellbeing measures administered at baseline and follow up assessments included the Dysarthria Impact Scale,⁴⁷ Depression Anxiety and Stress Scale (DASS-21),⁴⁸ and Lille Apathy Rating Scale—Short Form (LARS-SF).⁴⁹ To measure Parkinson’s symptoms at baseline and follow up, the following measures were used: the Parkinson’s Disease Questionnaire (PDQ-39)⁵⁰ and a modified Movement Disorder Society – Unified Parkinson’s Disease Rating Scale (MDS-UPDRS) for online assessment (tests of rigidity and postural instability omitted for Part III motor assessment).^21,22,51

Participants completed a brief survey before and after each ParkinSong Online session to confirm they were in a hazard-free environment and to capture self-reported mood and energy levels using the Affect Grid.⁵² Participants were asked to report any practice between sessions using ParkinSong Online materials or other singing resources. A post-session survey captured participants’ overall experience of each session using a 5-point Likert scale. Survey questions included rating engagement in vocal and singing exercises, enjoyment of the session, and any difficulties related to the vocal exercises, songs, or technology. At the end of the intervention period, the Acceptability, Appropriateness, and Feasibility of Intervention Measure (AAFIM)⁵³ was completed by facilitators, and participants were invited to take part in focus group interviews about their experiences.

Intervention

The ParkinSong Online protocol consisted of 12 weekly 90-minute sessions based on our previous face-to-face intervention,⁹ but delivered via Zoom. Group size was limited to no more than 16 participants to maximize potential for individualized therapeutic intervention and group connection. Each group was co-facilitated by a music therapist and a speech pathologist using a transdisciplinary approach and optimizing Zoom audio settings for live music sharing (ie, “Original sound for musicians” and “High-fidelity music mode”). The format of each session was developed by the facilitators, within general guidelines provided by a facilitator manual and informed by participant preferences for genres of music and specific songs. A typical session consisted of breathing exercises and vocal singing warmups (approximately 15 minutes), speech exercises (approximately 10 minutes), intensive high effort singing, that is, with a strong focus on respiratory support, vocal technique, and loud vocal projection, as opposed to singing for purely recreational purposes (approximately 35 minutes), and finally 30 minutes of social communication practice in small groups (using the breakout group function of Zoom). Within this format, facilitators educated participants on optimal posture and vocal care for voice projection, muted participants as required during the voice exercises and singing, and facilitated social interactions in the breakout groups to practice using techniques taught during the first part of the session. Due to latency issues, participants were required to stay muted whilst singing. They could therefore only hear their own voice and the facilitator’s voice. The singing and voice exercises part of the sessions was recorded and made available to participants via a secure online platform (Canvas) hosted by the University of Melbourne for practice during the week and to catch up on missed sessions. These recordings were also used for fidelity checking of adherence to the intervention protocol.

Outcomes Analysis

Feasibility results for recruitment, retention, fidelity, safety, attendance, and acceptability, are reported descriptively. Baseline and follow up data for preliminary efficacy testing were analyzed using descriptive and inferential statistics (pairwise comparisons) with IBM SPSS Statistics version 27.⁵⁴ Qualitative interview data were analyzed using thematic analysis⁵⁵ with an inductive approach. To ensure a rigorous analysis, 1 author became familiar with the interview data, generated initial codes, and used these to develop themes. From here, a second author reviewed and verified the codes and themes generated, and suggested new themes.

Costs Analysis

Intervention costs were estimated, including facilitator and technical support costs. Computer and internet costs were not included as these are already available for most people and the Zoom video conferencing software was free. Although each group was designed to have a maximum of 16 participants, we conservatively estimated facilitator costs for 14 participants, which would be closer to an implementation scenario (15% vacancy in each group). For travel costs, we estimated the cost of car travel,⁵⁶ plus a conservative 1 hour per session for a carer to drive a participant to a face-to-face session. Travel costs were calculated by multiplying the distance by AUD$0.72/km, using cents-per-kilometer methods to claim business deductions per the 2021 to 2022 Australian Tax Office Guidelines with conservative estimates of 10 to 30 km distance traveled. Transportation time costs were calculated using the Fair Work Ombudsman Passenger Vehicle Transportation Award (MA000063) grade 3 ordinary hour rate in December 2022 of AU$ 31.05 per hour.

Results

Recruitment commenced in March 2022 and ceased in April 2022 when a total of 28 participants with idiopathic Parkinson’s (16 male) were recruited, with a recruitment rate of 90% and no dropouts during the 12 weeks of online sessions. Recruitment rate is an upper estimate based on number of REDCap screening surveys opened; some initial inquirers may have been identified as not eligible before proceeding to screening in REDCap. Figure 1 shows the participant flow through the study. Group A ran from 31st March to 16th June 2022 and Group B ran from 2nd June to 18th August 2022.

Figure 1.

CONSORT flowchart.

Baseline demographics are presented in Table 1. Most participants were in the earlier stages of disease progression and had minimal cognitive impairment. Twenty-five participants were taking Levodopa medication over the course of the intervention and 4 participants had received deep brain stimulation more than 6-months prior to enrolment. The majority enjoyed singing and half had previously attended group singing activities.

Table 1.

Baseline Demographics.

	M (SD)
Age	68.0 (7.5)
Years since diagnosis	7.8 (6.2)
MoCA total score	25.3 (3.7)
	n (%)
Female gender	12 (43)
Education >13 years	23 (82)
Hoehn & Yahr
Stage 1	6 (22)
Stage 2	17 (63)
Stage 3	4 (15)
History of deep brain stimulation >6 months	4 (14)
Prefer assistance	3 (11)
Carer available	24 (86)
Current or previous ParkinSong or online singing attendance	14 (50)
I have attended a choir or singing group	14 (50)
I like to sing	24 (86)
I like to sing frequently^a	10 (36)

This was a branching question, that is, if a participant indicated that they never like to sing then the frequency question did not appear.

Session attendance was high, ranging from 67% to 100% or 8 to 12 sessions, with average compliance 89% or 10 to 11 sessions. Reasons given for absenteeism included accidents, injuries, operations, poor sleep the night before, previously planned trips or social events, and unforeseen family demands. There were no adverse events or safety issues reported. Intervention fidelity (determined from session recordings) and acceptability (determined from facilitator and participant feedback) were high. Online intervention costs included staffing costs totaling AUD$586 per participant for the 12-week program or AUD$41 per session. In comparison, in a hypothetical scenario where participants would spend 1 hour for a 30 km round trip to attend 12 in-person sessions, it would cost AUD$ 632 or AUD$53 per session.

There were no improvements in any secondary (preliminary efficacy) outcomes from pre- to post-intervention in this feasibility study. As predicted with a progressive condition, there were small overall declines on most outcomes, and these reached statistical significance for decibel levels on the loud sentence task and anxiety ratings on the DASS. Tables 2 and 3 show the preliminary efficacy testing results for speech and voice outcomes and wellbeing outcomes respectively. Overall, 33% of participants increased their speech loudness on the monolog task, and 67% declined.

Table 2.

Speech and Voice Results.

Outcome measure	n	Baseline M (SD)	Follow-up M (SD)	Mean difference (95% CI)	Paired t-test	Cohen’s d
Monolog (dB)	21	57.9 (7.5)	54.5 (5.8)	−3.4 (−7.1-0.3)	P = .07	0.51
Loud sentence (dB)	23	78.3 (6.3)	74.6 (5.6)	−3.7 (−6.2-−1.2)	P = .006	0.64
Reading (dB)	23	58.8 (7.6)	57.6 (5.7)	−1.2 (−3.4-1.0)	P = .27	0.24
Reading over noise (dB)	18	62.6 (7.6)	60.5 (5.7)	−2.1 (−5.1-0.8)	P = .15	0.35
CPPs^a	24	26.4 (3.1)	27.0 (3.8)	0.6 (−0.6-1.7)	P = .34	0.20
Diadochokinetic rate (syllables per second)	24	3.7 (0.8)	3.6 (0.9)	−0.2 (−0.7-0.4)	P = .59	0.17
Maximum Phonation Time (sec)	23	14.4 (8.7)	15.7 (10.3)	1.3 (−4.6-1.9)	P = .41	0.13
Disease Severity Composite score (log10)	23	3.4 (0.9)	3.1 (1.2)	−0.3 (−1.0-0.4)	P = .35	0.20
Intelligibility Composite score (log10)	21	23.4 (0.7)	23.3 (0.6)	−0.1 (−0.3-0.1)	P = .26	0.25
Naturalness Composite score (log10)	21	5.2 (1.5)	5.0 (1.4)	−0.3 (−1.0-0.4)	P = .44	0.17

CPPs, Cepstral peak prominence-smoothed, is a measure of dysphonia. Higher scores are better on all variables except Disease Severity Composite score. Participants whose dB recordings were more than 3 standard deviations below the mean of the sample were excluded from descriptive and inferential statistics, as were participants who only had a useable recording at either baseline or follow up.

Table 3.

Wellbeing Results.

Outcome measure	Score range	Baseline M (SD)	Follow-up M (SD)	Mean difference (95% CI)	Paired t-test	Cohen’s d
Dysarthria impact	0 to 110	36.9 (20.0)	39.5 (18.8)	2.7 (−8.3-3.0)	P = .34	0.13
DASS depression	0 to 21	3.2 (2.6)	3.9 (2.7)	0.7 (−1.6-0.2)	P = .14	0.26
DASS anxiety	0 to 21	3.6 (2.5)	4.5 (2.2)	0.9 (−1.7-−0.1)	P = .04	0.38
DASS stress	0 to 21	4.3 (3.7)	5.0 (3.8)	0.7 (−1.7-0.3)	P = .18	0.19
LARS-SF	−15 to 15	−9.2 (4.0)	−10.2 (3.5)	−1.0 (−2.3-0.3)	P = .14	0.27
PDQ-39	0 to 156	39.0 (21.9)	42.7 (21.9)	3.7 (−8.9-1.4)	P = .15	0.17
MDS-UPDRS I	0 to 35	2.8 (3.0)	2.9 (2.5)	0.1 (−1.1-1.0)	P = .88	0.04
MDS-UPDRS II	0 to 80	21.7 (9.6)	23.7 (9.6)	2.0 (−4.0-0.0)	P = .05	0.21
MDS-UPDRS III	0 to 126	17.0 (9.2)	16.2 (7.3)	−0.8 (−4.7-3.1)	P = .68	0.10
MDS-UPDRS IV	0 to 24	4.8 (3.2)	5.0 (3.8)	0.1 (−1.5-1.3)	P = .87	0.06
MDS-UPDRS total	0 to 265	45.7 (17.5)	45.6 (14.8)	−0.1 (−6.2-5.9)	P = .97	0.01
Hoehn & Yahr	1 to 4	2.0 (0.6)	2.04 (0.5)	−0.04 (−0.3–0.2)	P = .77	0.06

Abbreviations: LARS-SF, Lille Apathy Rating Scale Short Form; Dysarthria Impact Scale; DASS, Depression, Anxiety, and Stress Scale; PDQ-39, Parkinson’s Disease Questionnaire—39 items; MDS-UPDRS, Motor Disorders Society Universal Parkinson’s Disease Rating Scale. Higher scores on all measures indicate more pathology.

Affect Grid mood and energy scores (ranging from 1 to 9) were recorded by each participant before and after each ParkinSong Online session. Average scores were calculated for each participant and aggregated to compare difference between pre and post energy and mood levels for the overall cohort. No significant changes in aggregated scores were observed pre to post session for either mood or energy levels. Practice between sessions ranged from 0 to 6 practice sessions per week, with average overall practice for individual participants ranging from 1 to 4 times a week. Some participants used the session recordings, while other preferred to sing with YouTube clips or radio.

Overall experience reported at the end of each ParkinSong Online session was very positive with an average score of 3.3 (0.4) from a possible range of 0 (“Unsatisfactory”) to 4 (“Great”). The pace of the sessions had a possible range of 0 (“Very slow”) to 4 (“Very fast”), with 2 being “Just right.” On average participants perceived initial sessions as “a bit slow” (rating of 1), after which most sessions were rated close to or exactly “Just right.” No participants found the sessions “Too fast.” Feedback was similar from the 2 groups, suggesting consistency across the different facilitators.

To assess feasibility from the facilitator perspective we used the AAFIM.⁵³ Individual facilitator ratings (n = 4) were averaged for the subdomains of acceptability, appropriateness and feasibility. From a score range of 4 to 20, the mean scores of 17.0 (2.0) for acceptability, 16.8 (2.2) for appropriateness, and 17.8 (2.1) for feasibility corresponded to a high degree of facilitator endorsement for the ParkinSong Online intervention. Despite this, facilitators also identified challenges within sessions as they were required to manage multiple tasks simultaneously and were unable to hear individual voices to give targeted feedback. Tasks included muting participants who had difficulty with the technology, changing online slides with lyrics or exercise instructions, pinning the main facilitator for view layout, and monitoring participant engagement while modeling and leading exercises or songs.

Qualitative Results

The thematic analysis of qualitative data identified 2 overall themes directly related to participants’ experience of the ParkinSong Online intervention in terms of: (1) perceived benefits (Table 4), and (2) perceived challenges (Table 5). There were 7 subthemes under the overall theme of perceived benefits. Many reported that their participation in ParkinSong Online generated positive feelings (subtheme 1.1), and increased their awareness of, and use of their voice (subtheme 1.2). There was an expectation and perception of voice improvement related to ParkinSong participation (subtheme 1.3) and participants appreciated the program structure and the skills of the facilitators (subtheme 1.4). They experienced ParkinSong Online as “a time away from Parkinson’s” (subtheme 1.5), suggesting joyful moments of respite from the challenges of Parkinson’s and valuing it as a safe space where they felt understood (subtheme 1.6). They also appreciated the accessibility of the online delivery (subtheme 1.7), particularly those for whom a face-to-face ParkinSong program was not available in their area.

Table 4.

Theme 1: Perceived Benefits of ParkinSong Online.

Theme 1: Perceived benefits
Participating in ParkinSong Online generated positive feelings	“I found myself singing songs in my head so much of the time when I’m doing things now, which I’ve never had before, so that’s good. That’s a big part of stopping other things from getting to me. . . I always feel so much better.” (Participant 10)“I feel much happier now that I’ve done the [ParkinSong] course, it’s really good.” (Participant 30)
Participating in ParkinSong Online increased participants’ awareness of, and use, of their voice	“If I’d hear one of the songs that that we were doing, or a song that I particularly liked on the radio, I would start singing along with that, which I didn’t used to do really. Yes, that’s been a change. And to use my voice more I guess, because I live alone.” (Participant 2)
There was an expectation/perception of voice improvement related to ParkinSong participation	“I had the expectation that it would help my voice and I think it did. My sister’s noticed that my voice is stronger.” (Participant 2)“I didn’t know what to expect, I just went into and hoped it would help me, but it has helped me.” (Participant 30)
Participants appreciated the program structure and the skills of the facilitators	“I thought the structure was very good. It seemed therapeutic, well designed. . . It was comforting. . .I knew what was going to happen each week with the format, so it gave a little security and purpose.” (Participant 12)“You could see each week they would try and improve their sessions because they throw another little thing in that made a difference and I felt that the techniques they used improved over time as well. Learning for us, but learning for them on the spot, which is probably very difficult, but they did it really well.” (Participant 34)
ParkinSong Online was a time away from Parkinson’s	“When we did the ParkinSong group, it’s a time away from Parkinson’s really, you know.” (Participant 17)“Whilst you’re doing it, you’re not thinking about your Parkinson’s, because you got another focus and it’s enjoyable and requires a fair bit of concentration as well.” (Participant 22)
ParkinSong Online provided a safe space where participants felt understood	“It [online] was good, because it didn’t matter if you couldn’t sing, you had to sing by yourself.” (Participant 30)“One [benefit] was the presenters being able to mute my voice when I was singing. I thought. . . they haven’t got to put up and hear my croaky voice.” (Participant 7)“To see a ‘room’ full of other people with Parkinson’s was very reassuring. . . just to know that there’s other people all going through the same sorts of issues.” (Participant 14)
Participants appreciated the accessibility of online ParkinSong delivery	“The benefits were you could do it from any computer anywhere.” (Participant 25)“I think it serves a real purpose, because particularly for those of us. . . who may have difficulties in getting to a live session in your area.” (Participant 22)“This [online format] made it that you had to turn up because there’s no excuse. You’re in your own home.” (Participant 34)

Table 5.

Theme 2: Perceived Challenges of ParkinSong Online.

Theme 2: perceived challenges
Lack of feedback as ParkinSong Online facilitators couldn’t hear participants	“You can remain behind the screen as more of an observer instead of participating I suppose.” (Participant 25)“They don’t hear me singing anyway because it’s all muted. So it’s hard to pick up anything from that point of view.” (Participant 40) “You don’t know if you. . . you were never told if you were doing something right. And I don’t know how they would know if you were doing something right or not, because sometimes I wasn’t sure if I was actually breathing in the right place or whatever. . .” (Participant 14)
Singing online is not the same as singing together in person	“You can never really sing together. . . [so] it’s different to a normal choir; a face-to-face choir where the musical director would detect an aspect of the song that, you know, the choir hadn’t quite got right. So they would go back and go over that.” (Participant 21)“It is much easier when you’re all together [in person] and you can sing together.” (Participant 19)“I’ll be honest and say. . . I do prefer to sing [in person] with other people, but any excuse to sing is good.” (Participant 44)
The limitations of technology were frustrating	“One of the things I did find frustrating was the technology. The concept of people you know, anywhere joining in is sort of, you know, mind blowing. But it wasn’t. It’s not. . . And you can’t do anything about this. But I just found it a frustration because, you know, I really enjoy singing in a group. And when we had group singing [online], of course it doesn’t work because there’s. . . delays here and there depending on where people are.” (Participant 21)“Every week I seemed to have sound issues. . . We didn’t grow up with the technologies that young people do today.” (Participant 7)“Initially I had trouble using the link and joining. . . but it did become easier.” (Participant 40)
Desire for sustainability and ongoing access to ParkinSong programs	“I really enjoyed it and I miss it very much.” (Participant 30)“It’s a bit of a sad ending to think that. . . it’s not continuing. . . I’d like to see something like this become a more regular thing for everybody with Parkinson’s to do.” (Participant 44)

The main challenges cited by participants related to inherent limitations of the online platform (such as needing to mute during synchronous exercises or singing), learning to use, and needing assistance with the technology. Participants understood that the muting requirement during group singing meant that ParkinSong Online facilitators could not hear them and were thus unable to provide feedback (subtheme 2.1). The need for participants to stay muted during group singing differentiated the experience from in-person singing (subtheme 2.2) and caused frustrations with the technology (subtheme 2.3). However, participants all enjoyed the ParkinSong Online program, were disappointed when it ended, and talked about their desire for ongoing access and sustainable therapeutic singing programs (subtheme 2.4).

Discussion

Based on participant data and facilitator responses to feasibility measures, the ParkinSong Online intervention was feasible to administer despite some technological challenges for some participants. Recruitment rate was high, with no dropouts or adverse events reported during the intervention period. The online data collection protocol also worked well, although there was some missing or unusable speech data, so the remote collection of recorded speech data with this population may need further exploration. Acceptability of the intervention was good, with strong attendance, high facilitator ratings for intervention acceptability, and many positive comments in the qualitative data. However, despite these indications of acceptability, both participants and facilitators reported frustrations with the lack of capacity for participants to sing together off mute due to latency issues. The ramifications were that participants did not experience the Lombard effect (where they would have to project their voice over the group sound), nor were they able to benefit from facilitator feedback as no one could hear them. Thus, it is perhaps understandable that we found no improvements in speech outcomes, in contrast to our previous in person ParkinSong study findings.⁹ It is likely that the restrictions to group singing caused by online latency had 2 limiting factors: (1) facilitators could not hear participants’ voices or see their full upper body, and thus could not provide appropriate guidance and feedback, and (2) the intervention did not stimulate high enough vocal effort to reach the threshold for a therapeutic effect on speech loudness. As participants were aware that no one could hear them, it is also possible that they felt less motivated or accountable to deliver maximum vocal effort.

It should be noted that the study design was primarily to evaluate feasibility and as such was not powered for efficacy testing. Further, in a degenerative condition, a pre-post study with no control group is not an ideal design for evaluating treatment efficacy. With that caveat in mind, the wellbeing measures also did not change from baseline to follow-up, apart from anxiety, although a mean increase from 3.6 to 4.5 on this 21-point scale is not clinically important.⁵⁷ Furthermore, it should be noted that most wellbeing measure scores were already low at baseline, indicating a group of individuals who were already coping well. Although apathy, depression, and anxiety are commonly reported non-motor symptoms of Parkinson’s, it is possible that people experiencing difficulty with such symptoms are less likely to volunteer for an interventional research study. Consequently, our recruited participants may represent a skewed sample with higher levels of baseline wellbeing and motivation. This may also explain the aggregated minimal change in affect from pre to post session, as most participants were already feeling positive prior to the session. However, despite the lack of change on quantitative measures, participants reported positive impacts on mood in the qualitative results.

Our feasibility results are consistent with other recent studies^23,25 indicating that therapeutic singing interventions can be delivered remotely to increase accessibility for people with Parkinson’s. The differences in intervention effect can possibly be explained by the various delivery methods. The Mohseni study²⁵ used an individual intervention and was thus not limited by latency. Facilitators could hear participants and provide targeted feedback, however, a critique for this approach is that individual therapy is less cost effective than group therapy. The Stegemöller study²³ used a hybrid model, where groups met together to sing in remote locations following facilitator instructions from previously recorded group sessions. In this study, participants had a standard in-person group singing experience where the Lombard effect was likely activated.

Online therapy programs offer greater access to services for people in rural and regional areas where there is a higher incidence and prevalence of Parkinson’s and limited healthcare facilities.⁵⁸ While facilitator costs to deliver the intervention would likely be similar for online and in-person sessions (apart from the need to provide technical support for online sessions), the expense and burden for participants is lower without travel requirements. Our study shows that accessibility and cost savings may be increased even further through the remote administration of diverse baseline and follow up measures. This is broadly consistent with recent findings of the feasibility of telemedicine assessment for individuals with Parkinson’s,^19,20,59 including for motor symptoms.²¹ One caveat is that carers may sometimes be required to help assist individuals with Parkinson’s use the technology. Although the MoCA score was used as a measure to determine who would require additional support for the ParkinSong Online sessions, this turned out to be a very approximate indicator, and further refinement of the screening process for people with Parkinson’s would be required to ensure availability of adequate technology support resources during online sessions.

The qualitative results revealed some unexpected findings on the unique virtues of the online delivery method, in particular the ease of access to attend from home and from remote locations (subtheme 1.7). However, as well as the affordances of accessibility, there were also frustrations with technology. In addition to connection and sound issues (subtheme 2.3) participants reported missing the experience of a group sound and hearing each other sing, as reported in other studies of online singing.^60,61 Singing off mute was explored and the cacophony caused by latency effects was tolerated for periods by participants to experience the sense of singing together. On these occasions song choice was important (eg, slower songs with fewer participants worked better). Interestingly the need to stay muted when singing was perceived both positively and negatively by participants. It created a sense of safety (subtheme 1.6) as no one could judge their singing, but also was seen as a limitation because they couldn’t receive feedback from the therapists (subtheme 2.1).

As part of the facilitator guidelines within the ParkinSong Online protocol, there was opportunity for experimentation and modification of sessions considering participant feedback during and after each session. This responsiveness to participant feedback created a sense of ownership and co-development among participants within the group and contributed to subtheme 1.4 about participants’ positive experience of the session facilitation. Revisions to the protocol would be required to determine whether therapeutic benefits to voice outcomes are achievable. These revisions should include consideration of intervention dosage and intensity (to create conditions where participants are adequately vocally stimulated to reach thresholds required for therapeutic benefit), as well as creating opportunities for individual feedback on vocal output from therapists to combat vocal deterioration caused by Parkinson’s.

Despite only meeting virtually online, participants appeared to reap significant social benefits. They felt safe and understood in a group of other people with Parkinson’s (subtheme 1.6) and found it provided respite from Parkinson’s (subtheme 1.5). The opportunity for social interaction and communication were seen as positive aspects of online delivery and were not perceived as challenges. Although these are obviously different online when compared with face-to-face interaction, they were still perceived as positive opportunities. Many participants reported they would miss the group and wished to continue to meet after the trial finished (subtheme 2.4).

Limitations

Most participants (86%) reported that they liked to sing (in private, public, or both) and this may have led to a skewed sample of active singers. The main challenge to implementing ParkinSong Online was the need for technology support. For many participants this was only required initially, however, for some, ongoing support was needed. Ongoing technology support would be important to factor into future group-based telehealth singing interventions with this population in order for the sessions not to be disrupted. Therapists also need support for implementing online therapy as capability and experience with online platforms can be limited for therapists,⁶² as well as the need to manage low digital literacy of some participants.

Conclusions

ParkinSong Online is a feasible and acceptable group singing intervention for people with Parkinson’s. It affords valuable social support and connection if participants are adequately screened for technology support where required. Further, online delivery offers potential for increasing cost effectiveness as well as accessibility. The social benefits of group participation can be considerable, given service access issues and social isolation experienced by many. Therefore, although further research is needed to determine therapeutic efficacy, online singing groups for Parkinson’s may be a welcome addition, supplement, or in some cases a necessary alternative to in-person singing groups.

Footnotes

Acknowledgements

We would like to acknowledge Fight Parkinson’s for their financial and in-kind support for the implementation of this research. We also thank the people living with Parkinson’s who participated in this trial and their partners/carers who supported them with the technology. Special thanks also to Helen Crouch, Bonnie Nilsson, Laura Murray, and Hannah Carter for their contribution as ParkinSong Online facilitators.

Correction (November 2024):

Table 1 and Table 3 have been amended with the correct Hoehn & Yahr data. These data table changes did not affect any of the results or conclusions.

Author Contributions

A/Prof. Tamplin was responsible for the conceptualization, design, and supervision of the study. She conducted some of the focus group interviews, led the analysis of qualitative data, contributed to the statistical analysis and interpretation of quantitative data, and drafting and revision of the manuscript.

Dr. Haines was the ParkinSong Online project manager. He scheduled and conducted assessments, provided technical support during the ParkinSong Online sessions, and conducted focus group interviews. He also led the statistical analysis and contributed to the qualitative analysis and drafting and revision of the manuscript. Prof. Baker contributed to design and conceptualization of the study, interpretation of data and manuscript preparation. Dr. Sousa contributed to design and conceptualization of the study, data collection, assessment of costs, interpretation of data, and manuscript preparation. Dr. Thompson contributed to data collection, interpretation of data, and manuscript preparation. Ms. Crouch contributed to data collection, interpretation of data, and manuscript preparation. Mr. Dunn contributed to design and conceptualization of the study, data collection, interpretation of data, and manuscript preparation. Mrs. Tull contributed to design and conceptualization of the study, data collection, interpretation of data, and manuscript preparation. Prof. Vogel contributed to design and conceptualization of the study, analysis and interpretation of data, and manuscript preparation. Prof. Morris contributed to design and conceptualization of the study, interpretation of data, and manuscript preparation.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: A/Prof. Tamplin, Prof Baker, Dr Sousa, and Dr Thompson receive research funding from the Medical Research Future Fund, Australia (GNT20007411) and salaried support from The University of Melbourne. Dr. Haines, Ms Crouch, Ms Tull, and Mr Dunn report no disclosures. Prof. Vogel is an employee of Redenlab Inc. and receives salaried support from the Australian Research Council, Australia (220100253) and institutional support from The University of Melbourne. Prof. Morris receives research funding from the National Health and Medical Research Council, Australia (GNT11523) and has a joint appointment with Healthscope and La Trobe University.

Ethical Approval

Ethical approval was obtained from The University of Melbourne Human Research Ethics Sub-Committee (Approval no.: 2021-14465-16053-3, approval date: July 28, 2021).

ORCID iDs

Jeanette Tamplin

Simon J. Haines

Adam P. Vogel

Meg E. Morris

Trial registration number

ACTRN12621000940875 (registered July 19, 2021).

References

Magee

Copland

Vogel

AP.

Motor speech and non-motor language endophenotypes of Parkinson’s disease. Expert Rev Neurother. 2019;19(12):1191-1200. doi:https://doi.org/10.1080/14737175.2019.1649142

Pont-Sunyer

Hotter

Gaig

, et al. The onset of nonmotor symptoms in Parkinson’s disease (the ONSET PD study). Mov Disord. 2015;30(2):229-237.

Miller

Speech, voice and language in Parkinson’s disease: changes and interventions. Neurodegener Dis Manag. 2012;2(3):279-289. doi:10.2217/NMT.12.15

Landau

Harris

Burn

, et al. Anxiety and anxious-depression in Parkinson’s disease over a 4-year period: a latent transition analysis. Psychol Med. 2016;46(3):657-667.

Titova

Chaudhuri

KR.

Non-motor Parkinson disease: new concepts and personalised management. Med J Aust. 2018;208(9):404-409.

Schapira

Chaudhuri

Jenner

Non-motor features of Parkinson disease. Nat Rev Neurosci. 2017;18(7):435-450.

Ozdemir

Norton

Schlaug

Shared and distinct neural correlates of singing and speaking. NeuroImage. 2006;33:628-635.

Tamplin

Morris

Marigliani

Baker

Noffs

Vogel

AP.

ParkinSong: outcomes of a 12-month controlled trial of therapeutic singing groups in Parkinson’s disease. J Parkinsons Dis. 2020;10(3):1217-1230.

Tamplin

Morris

Marigliani

Baker

Vogel

. ParkinSong: a controlled trial of singing-based therapy for Parkinson’s disease. Neurorehabil Neural Repair. 2019:453-463. doi:10.1177/1545968319847948

10.

Di Benedetto

Cavazzon

Mondolo

Rugiu

Peratoner

Biasutti

. Voice and choral singing treatment: a new approach for speech and voice disorders in Parkinson’s disease. Eur J Phys Rehabil Med. 2009;45(1):13-19.

11.

Stegemöller

Radig

Hibbing

Wingate

Sapienza

Effects of singing on voice, respiratory control and quality of life in persons with Parkinson’s disease. Disabil Rehabil. 2017;39:594-600. doi:10.3109/09638288.2016.1152610

12.

Yinger

Lapointe

LL.

The effects of participation in a group music therapy voice protocol (G-MTVP) on the speech of individuals with Parkinson’s disease. Music Ther Perspect. 2012;30(1):25-31.

13.

Mendes

Brown

Sapienza

Rothman

HB.

Effects of vocal training on respiratory kinematics during singing tasks. Folia phoniatr logop. 2006;58:363-377.

14.

Tamplin

Morris

Baker

, et al. ParkinSong Online: protocol for a telehealth feasibility study of therapeutic group singing for people with Parkinson’s disease. BMJ open. 2021;11(12):e058953. doi:10.1136/bmjopen-2021-058953

15.

Flynn

Preston

Dennis

Canning

Allen

NE.

Utilising telehealth to support exercise and physical activity in people with Parkinson disease: a program evaluation using mixed methods. BMC Health Serv Res. 2023;23(1):224. doi:10.1186/s12913-023-09194-0

16.

Morris

McConvey

Wittwer

, et al. Dancing for Parkinson’s disease online: clinical trial process evaluation. Healthcare. 2023;11:604.

17.

Morris

Slade

Wittwer

, et al. Online dance therapy for people with Parkinson’s disease: feasibility and impact on consumer engagement. Neurorehabil Neural Repair. 2021;35(12):1076-1087.

18.

Dobkin

Interian

Durland

Gara

Menza

MA.

Personalized telemedicine for depression in Parkinson’s disease: a pilot trial. J Geriatr Psychiatry Neurol. 2018;31(4):171-176. doi:10.1177/0891988718783274

19.

Dorsey

Venkataraman

Grana

, et al. Randomized controlled clinical trial of “virtual house calls” for Parkinson disease. JAMA Neurol. 2013;70(5):565-570. doi:https://doi.org/10.1001/jamaneurol.2013.123

20.

Evans

Mohamed

Thomas

Using telemedicine and wearable technology to establish a virtual clinic for people with Parkinson’s. Mov Disord. 2020;35:S648-S649. doi:10.1136/bmjoq-2020-001000

21.

Abdolahi

Scoglio

Killoran

Dorsey

Biglan

KM.

Potential reliability and validity of a modified version of the Unified Parkinson’s Disease Rating Scale that could be administered remotely. Parkinsonism Relat Disord. 2013;19(2):218-221.

22.

Myers

Tarolli

Adams

, et al. Video-based Parkinson’s disease assessments in a nationwide cohort of Fox Insight participants. Clin Parkinsonism Relat Disord. 2021;4:100094. doi:10.1016/j.prdoa.2021.100094

23.

Stegemöller

Diaz

Craig

Brown

The feasibility of group therapeutic singing telehealth for persons with parkinson’s disease in rural Iowa. Telemed e-Health. 2020;26(1):64-68.

24.

Stegemöller

Hibbing

Radig

Wingate

Therapeutic singing as an early intervention for swallowing in persons with Parkinson’s disease. Complement Ther Med. 2017;31:127-133.

25.

Mohseni

Saffarian

Mohamadi

Abolghasemi

Habibi

SAH

. Effect of conventional speech therapy combined with music therapy on swallowing in patients with Parkinson’s disease (telerehabilitation): a randomized-controlled trial. Middle East J Rehabil Health Stud. 2023;10(1):e131572. doi:10.5812/mejrh-131572

26.

Daffern

Balmer

Brereton

Singing together, yet apart: the experience of UK choir members and facilitators during the Covid-19 pandemic. Front Psychol. 2021;12:303.

27.

Tamplin

Thompson

How health-focused choirs adapted to the virtual world during the COVID-19 pandemic–an international survey. Arts Psychother. 2023;82:101997.

28.

Price

Whitfield

Singing for health: do members of a generic singing for health group experience similar effects on health and well-being as those in condition-specific groups?

J Appl Arts Health. 2018;9(3):351-366.

29.

Bannan

Montgomery-Smith

‘Singing for the brain’: reflections on the human capacity for music arising from a pilot study of group singing with Alzheimer’s patients. Perspect Public Health. 2008;128(2):73-78. doi:10.1177/1466424007087807

30.

Sakano

Ryo

Nakayama

, et al. Possible benefits of singing to the mental and physical condition of the elderly. BioPsychoSoc Med. 2014;8(1):11. doi:10.1186/1751-0759-8-11

31.

Tamplin

Baker

Jones

Way

Lee

‘Stroke a chord’: the effect of singing in a community choir on mood and social engagement for people living with aphasia following a stroke. Neurorehabilitation. 2013;4(32):929-941. doi:10.3233/NRE-130916

32.

Siponkoski

S-T

Pitkäniemi

Laitinen

, et al. Efficacy of a multicomponent singing intervention on communication and psychosocial functioning in chronic aphasia: a randomized controlled crossover trial. Brain Commun. 2022;5(1):fcac337.

33.

Clift

Hancox

Morrison

Hess

Kreutz

Stewart

The significance of choral singing for sustaining psychological wellbeing: findings from a survey of choristers in England, Australia and Germany. Music Perform Res. 2010;3(1):79-96.

34.

Clift

Morrison

Group singing fosters mental health and wellbeing: findings from the East Kent “singing for health” network project. Mental Health Soc Incl. 2011;15(2):88-97. doi:10.1108/20428301111140930

35.

Skingley

Page

Clift

, et al. “Singing for breathing”: participants’ perceptions of a group singing programme for people with COPD. Arts Health. 2014;6(2):59-74. doi:https://doi.org/10.1080/17533015.2013.840853

36.

Nasreddine

Phillips

Bédirian

, et al. The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc. 2005;53(4):695-699.

37.

Harris

Taylor

Thielke

Payne

Gonzalez

Conde

JG.

Research electronic data capture (REDCap)—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42(2):377-381.

38.

Vogel

Graf

Magee

Schöls

Rommel

Synofzik

Home-based biofeedback speech treatment improves dysarthria in repeat-expansion SCAs. Ann Clin Transl Neurol. 2022;9(8):1310-1315.

39.

Caverlé

Vogel

AP.

Stability, reliability, and sensitivity of acoustic measures of vowel space: a comparison of vowel space area, formant centralization ratio, and vowel articulation index. J Acoust Soc Am. 2020;148(3):1436-1444.

40.

Vogel

Maruff

Monitoring change requires a rethink of assessment practices in voice and speech. Logoped Phoniatr Vocol. 2014;39(2):56-61.

41.

Vogel

Fletcher

Snyder

Fredrickson

Maruff

Reliability, stability, and sensitivity to change and impairment in acoustic measures of timing and frequency. J Voice. 2011;25(2):137-149.

42.

Vogel

Rosen

Morgan

Reilly

Comparability of modern recording devices for speech analysis: smartphone, landline, laptop, and hard disc recorder. Folia phoniatr logop. 2015;66(6):244-250.

43.

Noffs

Cobler-Lichter

Perera

, et al. Plug-and-play microphones for recording speech and voice with smart devices. medRxiv. 2023:2023.09. 30.23296391.

44.

Vogel

Reece

. Recording speech: methods and formats. In: Ball

, ed. Manual of Clinical Phonetics. Routledge; 2021:217-227.

45.

Vogel

Morgan

AT.

Factors affecting the quality of sound recording for speech and voice analysis. Int J Speech-Lang Pathol. 2009;11(6):431-437.

46.

Vogel

Maruff

Comparison of voice acquisition methodologies in speech research. Behav Res Methods. 2008;40(4):982-987.

47.

Vogel

Graf

Bade

Hepworth

Synofzik

Development and validation of the dysarthria impact scale. Paper presented at: International Congress for Ataxia Research; November 2022; Dallas.

48.

Henry

Crawford

JR.

The short-form version of the Depression Anxiety Stress Scales (DASS-21): construct validity and normative data in a large non-clinical sample. Br J Clin Psychol. 2005;44(2):227-239.

49.

Dujardin

Sockeel

Carette

A-S

Delliaux

Defebvre

Assessing apathy in everyday clinical practice with the short-form Lille Apathy Rating Scale. Mov Disord. 2013;28(14):2014-2019. doi:10.1002/mds.25584

50.

Jenkinson

Fitzpatrick

Peto

Greenhall

Hyman

The Parkinson’s Disease Questionnaire (PDQ-39): development and validation of a Parkinson’s disease summary index score. Age Ageing. 1997;26:353+.

51.

Tarolli

Andrzejewski

Zimmerman

, et al. Feasibility, reliability, and value of remote video-based trial visits in Parkinson’s disease. J Parkinsons Dis. 2020;10:1779-1786. doi:10.3233/JPD-202163

52.

Russell

Weiss

Mendelsohn

GA.

Affect grid: a single-item scale of pleasure and arousal. J Pers Soc Psychol. 1989;57(3):493.

53.

Weiner

Lewis

Stanick

, et al. Psychometric assessment of three newly developed implementation outcome measures. Implement Sci. 2017;12:1-12. doi:10.1186/s13012-017-0635-3

54.

IBM Corp. IBM SPSS Statistics for Macintosh Version 24. IBM Corp.; 2016.

55.

Clarke

Braun

. Thematic Analysis: A Practical Guide. Sage; 2022:1-100.

56.

Saiyed

Hill

Russell

Theodoros

Scuffham

Cost analysis of home telerehabilitation for speech treatment in people with Parkinson’s disease. J Telemed Telecare. 2022;28(7):524-529. doi:10.1177/1357633x20948302

57.

Ronk

Korman

Hooke

Page

AC.

Assessing clinical significance of treatment outcomes using the DASS-21. Psychol Assess. 2013;25(4):1103.

58.

Duncan

Rositano

Parkinson’s disease in regional Australia. Rural Remote Health. 2011;11(4):28-40.

59.

Dobkin

Mann

Weintraub

, et al. Innovating Parkinson’s care: a randomized controlled trial of telemedicine depression treatment. Mov Disord. 2021;36(11):2549-2558. doi:https://doi.org/10.1002/mds.28548

60.

Thompson

Baker

Clark

, et al. ‘A light at the end of the tunnel’: the experiences of members of a therapeutic community choir for people living with dementia and their care-partners who ‘went online’ during the COVID19 pandemic: an arts-based phenomenological study. Voices: World Forum Music Ther. 2023;23(2). doi: 10.15845/voices.v23i2.3491

61.

Datta

‘Virtual choirs’ and the simulation of live performance under lockdown. Soc Anthropol. 2020;28(2):249-250.

62.

Morris

Brusco

Jones

, et al. The widening gap between the digital capability of the care workforce and technology-enabled healthcare delivery: a nursing and allied health analysis. Healthcare. 2023;11(7):994.