The study selection process resulted in the inclusion of 56 studies (reported in 67 separate publications) that met the inclusion criteria for the systematic review, of which 10 were able to provide data for the meta-analysis (see Figure 1 for details of each stage in the process and reasons for exclusion). Studies came from 12 countries (Australia, Canada, Germany, Italy, Ireland, Japan, New Zealand, Norway, South Korea, Sweden, the UK, and the USA) and used a variety of quantitative designs. They were published from 1989 to 2020 and involved a total of 1531 participants with PD (sample size from one study not available); the number of participants ranged from 5 to 95 per study (median sample size 22). Studies covered four broad performing arts modalities: dance, music therapy, singing, and theatre. Music therapy was conceptualised as active interventions of a musical nature that did not solely involve singing. A full list of included studies is provided in Appendix 8.
Theatre was the performing arts modality that was least studied with only two included studies.[43-44] Both studies compared theatrical interventions led by professional performers to physiotherapy. Both studies were Italian, while one [44] randomised group allocation. In the other study [43], allocation to groups was determined by logistics rather than randomisation.
There were four studies assessing music therapy. Within this modality, each study was quite different in the music therapy intervention offered. Pohl et al [45] assessed the Ronnie Gardiner Rhythm and Music Method (RGMM) comprising musical exercises to challenge cognition and sensorimotor control. Spina et al [46] used an intervention that comprised musical exercises, singing and dancing, showing that interventions can draw on components of multiple performing arts. Pacchetti et al [47] assessed the benefit of instrumental musical improvisation. Pantelyat et al [48] considered a West African drum circle intervention and was the only music therapy study to not be randomised. Three studies compared music therapy to usual care, while the control in Pacchetti et al [47] was a physiotherapy intervention.
There were 12 studies assessing singing interventions. Studies differed in the details of the intervention, but were all choral-based singing interventions. It was notable that the study by Tamplin et al [49-50] also included morning or afternoon tea for social interaction and conversation practice, which was offered to both intervention and control participants. Notably, in this study, the intervention was offered in weekly and monthly versions and the weekly and monthly singing groups differed in terms of having professional and amateur leaders respectively and whether the control group was a weekly activity such as painting, dancing or tai chi, or a monthly peer support group. Only two singing studies included a control group and the only randomised controlled trial was by Matthews et al, [51] in which the control group undertook a passive music appreciation activity.
As expected, given the greater focus on dance rather other performing art forms in previous reviews, dance was the performing arts therapeutic medium for which there was the largest body of evidence with a total of 38 studies (see Appendix 1 for details of each study). Twenty-two of these dance studies included a control group: variously physiotherapy, [52] exercise, [53-59] education, [60-62] support groups, [63-64] usual care, [56, 65-78] and a waiting list control. [79] Of these, there were 14 separate randomised controlled trials. [52, 54-56, 58, 60, 62-71, 73, 76, 79-80] Across the 39 dance studies, a number of different dance styles were used. These could be broadly classified into PD-specific dance forms, such as the Dance for Parkinson’s Disease method as designed by the Mark Morris Dance Group and the Brooklyn Parkinson Group, [81] modern dance including improvisational dance forms, mixed-genre dance, Turo dance (based on Qi meridians), ballet, Irish set dancing, Ballu Sardu (a Sardinian folk dance) and tango (see Appendix 2 for details on dance styles used in each study). Two separate forms of tango were used – traditional Argentine tango and adapted tango, the latter adapting steps for people with PD. Traditionally, in tango, the lead role is danced by the male. In adapted tango, typically all participants danced both lead and follow roles, while some studies of Argentine tango also adopted this practice. It was noted that in the single group repeated measures study by Koch et al [82] on Argentine tango, there were three separate group workshops (with each participant attending one) and in the first workshop, the class was taught in English and translated into German, whereas the other two workshops had a different leader and were taught directly in German.
Methodological limitations were frequent and SURE analysis (Appendices 5 and 6) highlights that common limitations included sampling, allocation methods and absence of control groups. The discussion reflects on these methodological issues and their implications.
Narrative synthesis of outcomes for people with Parkinson’s disease
Quality of life
Twenty-two studies assessed the impact of dance interventions on quality of life (Appendix 4), of which nine were randomised and eight had no control group. The most common dance interventions were PD-specific dance forms (9 studies) and tango or tango-based interventions (9 studies). Turo, Irish set dancing, American Ballroom and mixed-style partnered dance were all also studied. Across studies, the balance of the evidence supported a benefit of dance for quality of life, and this supported chiefly PD-specific and tango or tango-based dance forms, as these had been studied most. Only one study assessed American Ballroom [68-70] and found no evidence of significant benefit on quality of life. One study [60] found that following in tango offered greater quality of life benefit than leading. Four studies considered the impact of music therapy interventions on quality of life. Compared to usual care, Pantelyat et al [48] found a beneficial effect of the drum circle, while Pohl et al [45] found the same for RGRMM, as did Spina et al. [46] Compared to physiotherapy, Pacchetti et al [47] found a beneficial effect of instrumental musical improvisation on health-related quality of life. Four studies assessed the impact of a singing intervention on quality of life. [49-51, 83-86]. Of these, Matthews et al [51] used a RCT design compared to a passive music appreciation activity, while Tamplin et al [49-50] used a non-randomised controlled trial design with various comparators, the limitations of which were discussed above. Studies differed as to what aspects of quality of life they assessed and how these were measured. General quality of life was assessed by three studies. Both Irons et al [83-84] and Matthews et al [51] used the Parkinson’s Disease Questionnaire 39 Items (PDQ-39), [87], while Stegemöller et al [85-86] used the World Health Organization Quality of Life questionnaire (WHO-QOL).[88] Voice-related quality of life (VRQoL) [89] was assessed by two studies. [49-50, 85-86] Additionally, Stegemöller et al [85-86] assessed swallow-related quality of life (SWAL-QOL), [90] although no significant effect was found on this outcome. Two studies considered the impact of theatrical interventions on quality of life. Mirabella et al [43] and Modugno et al [44] both found group theatrical interventions led by professional performers to be more effective than physiotherapy in improving overall health-related quality of life. Mirabella et al [43] additionally found a greater benefit on emotional wellbeing.
Speech
Of the 12 studies considering singing interventions, 11 assessed speech outcomes. The body of evidence across studies (Appendix 4) supports a benefit of group singing on speech, with only one study [91] finding no evidence of benefit. However, statistical significance was not always reached likely as a result of small sample sizes. Speech outcomes for which there may be benefit of group singing interventions include phonation, intelligibility and vocal intensity, although the precise patterning of speech features for which evidence of benefit was found differed between studies. Only Tanner et al [92] reported clinical significance, and clinically significant improvements were found for intensity range in read speech and fundamental frequency variation, while the improvement in fundamental frequency in read speech was possibly clinically significant. Only two studies included a non-singing control group. In an RCT, Matthews et al [51] compared a singing intervention to a passive music appreciation activity and found evidence of a significant benefit on phonatory measures. In a non-randomised controlled trial, Tamplin et al [49-50] compared weekly and monthly singing interventions to a weekly session of painting, dancing or tai chi or a monthly per support group, and found that singing significantly improved speech intensity but not phonation, while a greater benefit was found in the weekly group. It should be noted that the weekly singing intervention was delivered by a professional music therapist and the monthly singing intervention was delivered by recreational local musicians, which further complicates interpretation of the findings, since it is unclear whether it is weekly delivery or a professional teacher that drives the benefit. Moreover, both the intervention and control groups also attended a morning or afternoon tea alongside each session for socialising and conversational practice. No studies considered the impact of any other performing arts modalities besides singing on speech.
Functional communication
Two studies discussed the impact of singing interventions on functional communication. Shih et al [91] found no significant change in functional communication after a group singing intervention of one 90-minute session per week for 12 weeks. However, a study by Elefant et al [93-94] found that a group singing intervention of one 60-minute session per week for 20 weeks significantly improved communicative facial expression and physical communication, although improvements in overall communication, plus functional and emotional subscales did not reach statistical significance. Neither study included a control group, which is a substantial limitation in terms of interpreting any observed benefit. No studies considered the impact of any other performing arts modalities besides singing on functional communication.
Cognitive status
Ten studies considered the impact of dance interventions on cognitive status (Appendix 4), of which six were randomised and only one study [95] did not have a control group. The evidence sub-divides into PD-specific dance forms (5 studies), tango, either Argentine or adapted (4 studies) and Ballu Sardu (1 study). At least some evidence of benefit on cognition was found for all studies across both dance styles, except one study on PD-specific individually customised dance. [63-64] One study [60] compared leading and following tango, and found that participants assigned to follow (this was not based on gender) improved significantly more in cognition than participants assigned to lead. Three studies considered the impact of music therapy interventions on cognitive status. Pohl et al [45] found a benefit of the RGRMM on cognitive function, while Spina et al [46] found the same with an active music therapy intervention comprising music, singing, and dancing. However, Pantelyat et al [48] did not find evidence of a beneficial effect of a West African drum circle intervention on cognition. Among these three studies, all included a control group and in each case the control group was usual care. The studies by Pohl et al [45] and Spina et al [46] were both randomised. No studies considered the impact of singing interventions on cognitive status, although one study [83-84] considered a cognitive quality of life subscale, which was considered a quality of life measure. One study [43] considered the impact of theatrical interventions on cognitive status and found no evidence of improvement in either the intervention or the physiotherapy control group. It used a non-randomised controlled trial design.
Motor function
Thirty-one studies considered the impact of dance interventions on motor function, of which 16 were randomised and eight lacked a control group (single group designs). The most common dance interventions were tango or tango-based dance (13 studies) and PD-specific dance forms (6 studies). Modern dance, improvisational dance, American Ballroom, mixed-genre or various partnered dance, ballet, turo, Irish set dancing, Ballu Sardu and dance/movement therapy were all also studied. Across studies, the balance of evidence supported a benefit of dancing for improving motor function, with the greatest volume of evidence being for tango and tango-related dance as well as PD-specific dance forms. One study [60] compared leading and following in tango and found that generally following was significantly motor effective than leading for improving motor function, although the opposite finding was found specifically for medication-related motor fluctuations. Only one study [96] considered clinical significance and found that the statistically significant benefit in motor function associated with improvisational dance fell slightly short of clinical significance. Four studies considered the impact of music therapy interventions on motor function. One study compared instrumental music improvisation to physiotherapy [47] and found that music therapy was more effective for improving motor function. RGRMM was found to be more effective than usual care for motor function, [45] while the evidence for a benefit of the drum circle on motor function was not conclusive [48] and no evidence of a benefit of the Spina et al [46] music therapy intervention was found for motor function. One study considered the impact of a singing intervention on motor function. Using a design without a control group, but with high and low dosage intervention groups, which were allocated according to clinical and logistical factors rather than randomly, Stegemöller et al [85-86] found a benefit of a weekly group singing session for eight weeks on the motor subscale of the Unified Parkinson’s Disease Rating Scale (UPDRS). [97] Two studies considered the impact of theatrical interventions on motor function. Both compared to physiotherapy, one study [44] found evidence of a beneficial effect of theatre on motor function, while the other [43] did not.
Main methodological concerns
The main methodological concerns that were applicable to the body of evidence as a whole included small sample sizes, the absence of control groups in over half of the included studies (this was not an issue for the music therapy and theatrical studies, but was common in dance studies, and very common in singing studies), considerable variation in the frequency and duration of intervention delivery, a wide range of disciplinary backgrounds and levels of experience among session leaders, substantial heterogeneity of outcome measures, especially for cognition but also for motor function, as well as a focus on statistical rather than clinical significance. These issues are discussed in detail in the discussion section.
Different performing arts modalities
No studies directly compared different performing arts modalities. Studies assessing each performing arts modality were conducted by different research teams, suggesting a tendency for scholars to work on a specific performing art modality rather than undertake multidisciplinary research across dance, music therapy, singing, and/or theatre. Not all of the outcomes of interest – speech, functional communication, cognitive status, motor function and quality of life – were assessed with regard to each performing art. In particular, speech and functional communication outcomes were only assessed in relation to singing interventions. Studies using singing interventions focused strongly on speech outcomes, with comparatively few considering a wider range of outcome domains. Furthermore, the volume of studies differed substantially between performing arts modalities. By far, the largest number of studies were conducted on dance interventions (39 studies) followed by singing interventions (12 studies). Comparably few studies assessed music therapy (not singing-only) interventions (4 studies) and theatrical interventions (2 studies). Among dance interventions, the greatest evidence was found for tango – either Argentine or adapted tango – as well as PD-specific tango forms, such as Dance for Parkinson’s Disease. [81] The body of evidence is currently insufficient to determine conclusively which performing arts modalities are most effective for which specific outcome domains due to the lack of comparative studies.
Meta-analysis results
Following assessment of feasibility (Appendix 7, Part A), six meta-analysis sets could be analysed, including a total of ten unique studies. Tabulated data for each comparison are shown in Appendix 7, Part B, and forest plots in Appendix 7, Part C.
The six feasible comparisons were: 1) UPDRS motor for tango-based dance vs exercise (3 studies), 2) UPDRS motor for tango-based dance vs usual care (2 studies), 3) UPDRS motor for theatre vs physiotherapy (2 studies), 4) TUG for PD-specific dance vs usual care (2 studies), 5) TUG for tango-based dance vs exercise (2 studies) and 6) PDQ-39 for PD-specific dance vs usual care (2 studies). Analyses were restricted to follow-up data except for comparison #6 for which the analysis could be conducted on change score data. The reason why change score based analyses could not be conducted for the other analysis sets was the unavailability of standard deviation data for the difference between baseline and follow-up scores for many studies, or a measure that could be converted into a standard deviation.
In analysis set one, meta-analysis did not show any evidence of a statistically significant difference in UPDRS motor scores at follow-up between participants undertaking tango-based dance and exercise (Z=0.05, p=0.96, Appendix 7, Part C), although there were only three studies able to contribute to the meta-analysis and heterogeneity was a concern (I2=57%). In analysis set two, participants undertaking tango-based dance were statistically significantly superior on UPDRS motor at follow-up than participants undertaking exercise (Z=2.87, p=0.004, Appendix 7, Part C), although there were only two studies and heterogeneity was a serious concern (I2=97%). In analysis set three, there was no statistically significant difference in UPDRS motor scores between participants undertaking theatrical interventions or physiotherapy (Z=0.37, p=0.71, Appendix 7, Part C), although there were only two studies. In analysis set four, there was no statistically significant difference in TUG between participants undertaking PD-specific dance and usual care (Z=0.98, p=0.33, Appendix 7, Part C), although there were only two studies and heterogeneity was a concern (I2 = 64%). In analysis set five, participants undertaking tango-based dance exhibited statistically significantly superior TUG scores at follow-up than participants undertaking exercise (Z=11.25, p<0.00001, Appendix 7, Part C, although there were only two studies. In analysis set six, participants undertaking PD-specific dance experienced statistically significantly superior improvement in PDQ-39 from baseline to final follow-up than participants undertaking usual care (Z=3.77, p=0.0002, Appendix 7, Part C).