A systematic review of behavioural therapies for improving swallow and cough function in Parkinson’s disease

Abstract

Purpose: This systematic review evaluated the efficacy of therapeutic interventions on improving swallow, respiratory, and cough functions in Parkinson's disease (PD).

Method: A PRISMA systematic search was implemented across six databases. We selected studies reporting pre- and post-assessment data on the efficacy of behavioural therapies with a swallow or respiratory/cough outcome, and excluded studies on medical/surgical treatments or single-session design. Cross-system outcomes across swallow, respiratory, and cough functions were explored. Cochrane’s risk of bias tools were utilised to evaluate study quality.

Result: Thirty-six articles were identified and further clustered into four treatment types: swallow related (n = 5), electromagnetic stimulation (n = 4), respiratory loading (n = 20), and voice loading (n = 7) therapies. The effects of some behavioural therapies were supported with high-quality evidence in improving specific swallow efficiency, respiratory pressure/volume, and cough measures. Only eleven studies were rated with a low risk of bias and the remaining studies failed to adequately describe blinding of assessors, missing data, treatment adherence, and imbalance assignment to groups.

Conclusion: Behavioural therapies were diverse in nature and many treatments demonstrated broad cross-system outcome benefits across swallow, respiratory, and cough functions. Given the progressive nature of the condition, the focus of future trials should be evaluating follow-up therapy effects and larger patient populations, including those with more severe disease.

Keywords:

• Parkinson’s disease
• swallowing
• cough
• respiration
• behavioural therapies
• treatment effects

Introduction

Parkinson’s disease (PD) is a progressive neuromotor degenerative disorder affecting six million people globally (GBD 2016 Neurology Collaborators, 2018). Four out of five people with PD develop swallowing difficulties based on instrumental assessment findings (Kalf et al., 2012). Impaired cough function is also common including dysfunction in motor components (cough intensity) and cough reflex sensitivity (Ebihara et al., 2003). PD related swallow and cough dysfunctions are associated with many clinical complications including requirements for increased medication, malnutrition, dehydration, and aspiration with or without pneumonia. These complications negatively affect the quality of life of individuals with PD and increase morbidity and mortality rates (Beyer et al., 2001; Ebihara et al, 2003) with aspiration pneumonia reported to be one of the leading causes of death in PD (Beyer et al., 2001).

Pharmacological and surgical PD interventions have been extensively covered in previous reviews, and demonstrate only limited effects on laryngopharyngeal functions (Allen & Miles, 2020; Baijens & Speyer, 2009; Trail et al., 2008). In recent literature, nonpharmacological behavioural therapies have begun to demonstrate positive outcomes. Yet, previous systematic reviews (Baijens & Speyer, 2009; Gandhi & Steele, 2022; Kim & Kim, 2023; López-Liria et al., 2020; Park et al., 2019; van Hooren et al., 2014) have focused on swallowing specific behavioural therapies in treating dysphagia, excluding the nonswallow therapies that also demonstrate cross-system effects on swallow and cough function. Recent literature on behavioural techniques emphasise the spread effects from one system to another. For example, respiratory focused activities such as expiratory muscle strengthening training (EMST) demonstrate not only respiratory system improvements in cough function but also improved swallow function (Pitts et al., 2009; Troche et al., 2010). Furthermore, a number of clinical trials have been published in recent years (Cocks et al., 2022; Di Benedetto et al., 2009; Plaza & Ruviaro Busanello-Stella, 2022; Stegemöller, Hibbing, et al., 2017; Stegemöller, Radig, et al., 2017; Tamplin et al., 2019; Tamplin et al., 2020; Wang et al., 2018) which were not included in previous systematic reviews (Baijens & Speyer, 2009; López-Liria et al., 2020; van Hooren et al., 2014). To date, no systematic reviews have covered efficacy of interventions for both swallow and cough despite their critical interrelationship in determining airway protection. Therefore, in addition to traditional swallow interventions, this review targeted all behavioural therapies that report swallow and respiratory/cough outcome measures. The study aimed to systematically identify randomised controlled trials (RCT) and non-RCTs that report efficacy of therapeutic interventions in (a) improving swallow and respiratory/cough physiological function, (b) improving functional swallowing outcomes, (c) reducing events of aspiration, and (d) change in pulmonary health in individuals with PD. We also aimed to extract key intervention components such as delivery, dose, intensity, and timing to provide an overview to aid in clinical decision-making.

Method

This review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA; Moher et al., 2010) and was registered on the International Prospective Register of Systematic Reviews (PROSPERO: CRD42020185622).

Selection criteria for studies

We included RCTs and non-RCTs that investigated the efficacy of therapies that reported swallow, respiratory, or cough outcomes in individuals with idiopathic PD. Studies were excluded if the intervention was solely based on pharmacological or surgical therapies. Several studies were identified where intervention was provided on only a single occasion (Supplementary Appendix 1). Studies with single case/case series or single-session designs were also excluded, as they did not demonstrate long-term treatment effects nor follow-up measures. Two studies were excluded as they described only compensatory strategies (Supplementary Appendix 1).

Studies recruiting individuals with all stages and severity levels of PD and all severity levels of swallowing difficulties, including no self-reported swallowing problems, were included. Objective and/or subjective measures of the selected studies were categorised as: (a) instrumental swallow measures (videofluoroscopy, endoscopy, electromyography); (b) instrumental cough/respiratory measures (spirometer, manometer, respiratory function test); (c) functional swallow outcomes based on functional change in diet and/or change in quality of life (QOL) by validated, subjective, self-administered questionnaires; (d) events of aspiration and change in pulmonary health chest x-ray reports, temperature charts, or records of aspiration pneumonia; change in the incidence of aspiration rated by the Penetration-Aspiration Scale score; change in residue severity measured by videofluoroscopy or endoscopy); and (e) adverse effects associated with intervention such as deterioration in function and discomfort. We included respiratory training studies with no cough parameters as recent evidence suggests that functional respiratory measures were correlated with cough strength measures (Curtis et al., 2020; Curtis et al., 2022; Jo & Kim, 2016; Kaneko et al., 2019). Inclusion and exclusion criteria are presented in Table I.

Table I. Selection criteria of the studies included in this review.

Selection criteria	Inclusion criteria	Exclusion criteria
Participants	Adults with idiopathic PD All stages and levels of PD severity All severity levels of swallowing difficulties or without swallowing problems	Other Parkinsonian syndromes and neurological disorders
Study design	Randomised controlled trials, quasi-randomised controlled trials, case-control studies, prospective cohort studies, and comparative observational studies	Single case report, single-session study, repeated published studies, secondary data (reviews)
Study setting	Any setting (hospital, rehabilitation unit, and residence)
Intervention	Any behavioural therapy trials that report a swallow or respiratory/cough outcome in individuals with PD	Pharmacological and surgical treatments Compensatory strategies and single-session treatment design
Comparator	None or sham/placebo treatment, drug/surgical treatment, or other swallow rehabilitation
Outcomes	Quantitative or qualitative Outcomes in objective and/or subjective measures of swallow, respiration, and cough function	Studies without pre- and post-intervention data
Publication details	Peer-reviewed articles Articles in English	Book chapters, thesis publications, editorials, commentaries

Search strategy

A specialist university librarian was consulted for advice on databases and search terms. A comprehensive search strategy was designed using the Medical Subject Headings (MeSH) terms as relevant to the research questions and piloted on MEDLINE (Supplementary Appendix 2). The search strategy was adapted for searches in other databases (Embase, Scopus, BIOSIS, Web of Science, and the Cochrane Controlled Trial Register [CCTR]) and we completed the search in March 2021. We updated our search again in March 2022. Broad search terms used included “Parkinson’s disease,” “swallowing disorders,” “cough,” “respiration,” “exercise,” “training,” “therapies,” and “treatment.”

Data collection and synthesis

Selected studies were saved in Endnote (Clarivate Analytics, Philadelphia) and duplicates were removed. Titles of studies retrieved were screened against exclusion criteria (Table I) then abstracts were screened for potentially eligible studies. A hand search of the reference list of related review articles and all articles that met inclusion criteria was performed to identify any additional eligible studies. After the initial screening, full texts of eligible studies were reviewed by all three authors for inclusion or exclusion in the final study. Any disagreements during the eligibility review were resolved by discussion.

Data extraction

The data extraction template published by the Cochrane group was adapted to include the selected study data. The adapted data extraction sheet included general study information, study characteristics, patient characteristics, a detailed description of the intervention based on the TIDieR checklist (Hoffmann et al., 2014), and outcome measures were recorded as per the CONSORT 2010 statement (Moher et al., 2010). Data extraction was performed by the first author, and further amendments and expansions were undertaken by the second author. Any disagreements were resolved by discussion.

Assessment of risk of bias and reporting of study quality

Cochrane Collaboration’s tool for assessing the risk of bias (RoB2; Sterne et al., 2019) was used to assess the methodological quality of the clinical trials for randomised trials under five main domains and graded as low, high, or some concerns according to the designed algorithm for suggested judgement. Risk of Bias in Non-Randomised Studies - of Interventions (ROBINS-I; Sterne et al., 2016) was used to rate the non-randomised trials included in this review under five criteria and graded as low, moderate, serious, critical risk, or no information. All data were reviewed by the second author at both study selection and quality appraisal stages. Discrepancies were resolved through consensus.

Data synthesis and analysis

Types of therapies were broadly grouped based on the focus of treatment: swallow-related, electromagnetic stimulation, respiratory loading, and voice loading therapies. Each study’s intervention protocol, study design, and participants’ demographics were tabulated in a summary table. Mean values for participants’ age, disease duration, and disease severity stage were calculated from the raw data where it was not otherwise specified. After reviewing the results of the included studies, the studies did not meet the requirements for meta-analysis as there were significant variation between assessment protocols, heterogeneity of study designs, and a lack of poolable data for the outcome measures of interest. The results of the studies were summarised descriptively for all swallow measures, cough/respiratory measures, and intervention type, reporting effect sizes and statistical significance where available. An evidence hierarchy classification model was used to summarise the overall findings (where Level 1a = meta-analysis of RCTs and Level IV = expert option/clinical experience only; Supplementary Appendix 3).

Result

The initial search yielded a total of 709 citations (MEDLINE = 184, Embase = 182, BIOSIS = 106, PubMed = 45, Scopus = 99, CCTR = 93). A PRISMA flow chart describing the study selection is presented in Figure 1. Thirty-six studies met all review criteria and were included in the final synthesis: 15 randomised controlled trials, 7 non-randomised clinical trials, and 14 pre/post non-controlled studies. Table II summarises types of intervention and their outcomes, and Supplementary Appendix 4 provides a comprehensive description of each study’s methodological design, study population, intervention, results, risk of bias, and level of evidence. In total, studies included 1151 individuals with PD and 816 received treatment. Mean age was 69 years (range 55–75 years), mean disease duration was 7 years (range 5–11 years), and Hoehn and Yahr Scale disease severity stage ranged from 2 to 3 in most of the studies. Overall male:female ratio was 2:1. Sample size varied widely, from eight to 109 participants (Table III).

Figure 1. Adapted PRISMA flow diagram, presenting different phases of study selection for the systematic review.

Table II. Types of interventions and outcomes.

Intervention	Objective/ instrumental swallow measures	Objective cough/respiratory measures	Change in quality of life	Functional swallow outcome (or change in events of aspiration, secretion, residue severity)	Adverse effects (%)
Swallow-related therapies
Effortful swallow + tongue exercises (Wang et al., 2018)	√
Tongue isometric pressure exercises + traditional tongue training therapy (Plaza & Ruviaro Busanello-Stella, 2022)	√		√
Motor swallow exercise (vocal exercises + tongue hold + supraglottic swallow + swallow muscle range of movements; Argolo et al., 2013)	√		√	√
Video assisted swallow training (Manor et al., 2013)	√		√	√
Biofeedback in strength and skill training (Athukorala et al., 2014)	√^d		√^d
Electrical/magnetic stimulation therapies
rTMS (Khedr et al., 2019)	√		√		4
SES + logopaedic dysphagia therapy (Baijens et al., 2013)	√			√	0
NMES + logopaedic dysphagia therapy (Park et al., 2018)			√^d	↔
NMES + effortful swallow + conventional dysphagia therapy (Heijnen et al., 2012)	√			√	0
Respiratory loading therapies
Aerobic + resistance training (Mavrommati et al., 2017)		√
Upper and lower limb muscle strength training (Alves et al., 2019)		√^d	√^d		0
Treadmill training (Pelosin et al., 2009)		√	√
Pursed lip + diaphragmatic pulmonary exercise (Güngen et al., 2017)		√	√
Breath-stacking or incentive spirometry (Ribeiro et al., 2018)		√
Pulmonary rehabilitation + upper extremity exercises (Köseoğlu et al., 1997)		√
Aerobic + qigong^a (Burini et al., 2006)		√	↔		4
Inspiratory muscle training (Inzelberg et al., 2005)		√	↔
EMST (n = 8^b studies)	√	√^d	√	√^d	0
Respiratory muscle training (inspiratory muscle training or EMST; Huang et al., 2020; Reyes et al., 2018)		√	√
EMST + swallow postural techniques (Byeon, 2016)	√
EMST^c + air stacking (Reyes et al., 2020)		√
Voice loading therapies
Voice and choral singing treatment^e (Di Benedetto et al., 2009)	↔	√			20^#
Therapeutic singing^f (Stegemöller, Hibbing, et al., 2017; Stegemöller, Radig, et al., 2017)	√	√	√
ParkinSong^g (Tamplin et al, 2019; Tamplin et al., 2020)		√
Lee-Silverman voice treatment^h (Miles et al., 2017; El Sharkawi et al., 2002)	√	√	√	√

Note. rTMS = repetitive transcranial magnetic stimulation; SES = surface electrical stimulation; NMES = neuromuscular electrical stimulation; EMST = expiratory muscle strength training.

√ = significantly improved after intervention (p = .05); ↔ = no significant change after intervention; ^# = quality of voice worse after treatment.

^aQigong: breathing exercises with stretching and balancing.

^b(Claus et al., 2021; Cocks et al., 2022; Darling-White & Huber, 2017; Kuo et al., 2017; Pitts et al., 2009; Sapienza et al., 2011; Troche et al., 2010; Troche et al., 2014).

^cEMST program adapted for 6 d a week for 8 weeks.

^dDemonstrated the improvement in function with moderate to large effects (Cohen’s d).

^eVCST: oro-facial, neck shoulder exercises, and singing with cues.

^fTherapeutic singing: vocal exercises followed by group singing.

^gParkinSong: music-based vocal exercises, singing, social interaction, and conversational practice.

^hLee-Silverman Voice Treatment (LSVT): vocal exercises, reading, and conversation with loudness cue.

Table III. Participant characteristics.

Intervention	Sample size: IG/total	Age (years)^a		Gender (M/F)	Diagnosis (years)^a	Disease severity H&Y^a (range)	Dropout % IG/CG
Swallow-related therapies
Effortful swallow + tongue exercises (Wang et al., 2018)	26/26		62	19/7	9	2 (1–3)	0/NA
Tongue isometric pressure exercises +  traditional tongue training therapy (Plaza & Ruviaro Busanello-Stella, 2022)	30/60		69	34/26	7	2.8	0/0
Motor swallow exercises (Argolo et al., 2013)	17/17		59	10/5	7	2 (1–4)	12/NA
Video assisted swallow training (Manor et al., 2013)	21/42		69	24/18	8	2	0/0
Biofeedback in strength & skill training (Athukorala et al., 2014)	10/10		67	7/3	7	3 (2–3)	0/NA
Electrical/magnetic stimulation therapies
rTMS (Khedr et al., 2019)	22/33		59	NR	6	3^b	14/0
SES + logopaedic dysphagia therapy (Baijens et al., 2013)	67/109		69	66/24	5	2^b (1–5)	16/23
NMES + logopaedic dysphagia therapy (Park et al., 2018)	9/18		59	8/10	NR	NR	0/0
NMES + effortful swallow + conventional dysphagia therapy (Heijnen et al., 2012)	60/90		68^b	65/25	NR	2^b	3/0
Respiratory loading therapies
Aerobic + resistance training (Mavrommati et al., 2017)	37/37		65	21/16	NR	UPDRS III 16 (0–43)	31/NA
Upper and lower limb muscle strength training (Alves et al., 2019)	16/32		66	16/16	5	1 (1–2), UPDRS: 74	25/0
Treadmill training (Pelosin et al., 2009)	10/10		69	4/6	8	2 (1–2.5)	0/NA
Pursed lip + diaphragmatic pulmonary exercises (Güngen et al., 2017)	34/34		68	21/13	6	1.6	24/NA
Breath-stacking or incentive spirometry (Ribeiro et al., 2018)	14/14		66	9/5	NR	2 (1–3)	6/NA
Pulmonary rehabilitation + upper extremity exercises (Köseoğlu et al., 1997)	9/9		55	4/5	6	1 (1–2)	22/NA
Aerobic + qigong^a (Burini et al., 2006)	26/26		65	9/17	11	3 (2–3)	15/15
Inspiratory muscle training (Inzelberg et al., 2005)	10/20		62	18/2	8	3 (2–3)	50/40
EMST (Torche et al., 2010)	32/65		67	47/13	NR	3 (2–4)	6/9
EMST (Sapienza et al., 2011) continuation of study by Troche et al. (2010)	32/65		67	47/13	NR	3	6/9
EMST (Claus et al., 2021)	25/50		67	37/8	7	2.5 (2–4)	4/16
EMST (Cocks et al., 2022)	16/16		75	15/1	7	NR	24/NA
EMST (Kuo et al., 2017)	9/13		(51-70)	7/6	(3-10)	(1–3)	23
EMST (Pitts et al., 2009)	10/10		73	10/0	NR	3 (2.5–3)	0/NA
EMST (Troche et al., 2014)	10/10		70	8/2	NR	3 (2–3)	0/NA
EMST (Darling-White et al., 2017)	12/12		75	9/3	8	(1–3)	0/NA
RMT (inspiratory muscle training or EMST; Reyes et al., 2018)	21/31		70	17/14	NR	1 (1–3)	19/28
RMT (inspiratory muscle training or EMST; Huang et al., 2020)	38/75		64	42/33	6	2 (1–3)	0/0
EMST + swallow postural techniques (Byeon, 2016)	18/33		64	31/2	NR	H & Y ≤ 3: 26, H & Y > 4: 7	0/0
EMST + air stacking (Reyes et al., 2020)	22/33		69	18/15	NR	1 (1–3)	0/0
Voice loading therapies
Voice + choral singing treatment (Di Benedetto et al., 2009)	20/20		66	13/17	7	2^b IQR (1.5–2.5)	0/NA
Therapeutic singing (Stegemöller, Radig, et al., 2017)	18/24		68	8/16	7	UPDRS:55	0/0
Therapeutic singing (Stegemöller, Hibbing, et al., 2017)	18/27		67	10/17	NR	UPDRS:56	0/0
ParkinSong (Tamplin et al., 2019)	47/75		74	46/29	9	2 (2–3)	6/3
ParkinSong (Tamplin et al., 2020)	47/75		74	46/29	9	2 (2–3)	6/3
Lee-Silverman voice treatment (El Sharkawi et al., 2002)	8/8		70	6/2	NR	3 (1.5–2.5)	0/NA
Lee-Silverman voice treatment (Miles et al., 2017)	20/20		68	14/6	6	UPDRS : 25 (9–48)	0/NA

Note. IG = intervention group; CG = control group; M/F = male/female (in numbers); H&Y = Hoehn and Yahr Scale; UPDRS = Unified Parkinson's Disease Rating Scale;   NR = not reported; NA = not applicable; EMST = expiratory muscle strength training; NMES = neuromuscular electrical stimulation; RMT = respiratory muscle strength training; rTMS = repetitive transcranial magnetic stimulation; SES = surface electrical stimulation.

^amean.

^bmedian.

Dosage and intensity

The length and frequency of therapy varied across studies (Table IV). All the studies relied on interventions that were two or more weeks in duration (range 2–16 weeks) with the exception of ParkinSong, which lasted for 52 weeks. Fifty-eight percent of the studies had four or more sessions in a week and training session duration varied from 30 to 120 min. Interestingly, 71% of studies (n = 25/36) reported interventions that required specialist skills for delivery. Treatment adherence was reported in 67% of the studies (n = 24/36). Three-quarters of studies applied interventions during the "on" state of medication dosing. Maintenance of therapeutic effects was reported in 10 of the 36 studies (28%).

Table IV. Dosage and intensity of the interventions.

Intervention and description	Number of sessions	Session duration (minutes)	Intervention time period (weeks)
Swallow-related therapies
Effortful swallow + tongue exercises (Wang et al., 2018)	120 (2 × 5 × 12)	NR^a	12
Tongue isometric pressure exercises + traditional tongue training therapy (Plaza & Ruviaro Busanello-Stella, 2022)	√		√
Motor swallow exercise (vocal exercises + tongue hold + supraglottic swallow + swallow muscle range of movements; Argolo et al., 2013)	50 (2 × 5 × 5)	NR^a	5
Video assisted swallow training (Manor et al., 2013)	6	30	6
Biofeedback in strength and skill training (Athukorala et al., 2014)	10 (5 × 2)	60	2
Electromagnetic stimulation therapies
rTMS (Khedr et al., 2019)	25^b	NR^a	14
SES + logopaedic dysphagia therapy (Baijens et al., 2013)	15	30	3
NMES + logopaedic dysphagia therapy (Park et al., 2018)	13-15	30	3-5
NMES + effortful swallow + CDT (Heijnen et al., 2012)^c	20 (5 × 4)	60	4
Respiratory-loading therapies
Aerobic + resistance training (Mavrommati et al., 2017)	48 (2 × 24)	60	24
Upper and lower limb muscle strength training (Alves et al., 2019)	32 (2 × 16)	30-40	16
Treadmill training (Pelosin et al., 2009)	12 (3 × 4)	30	4
Pursed lip + diaphragmatic pulmonary exercises (Güngen et al., 2017)	NR	NR	12
Breath-stacking or incentive spirometry (Ribeiro et al., 2018)	2f	NR^a	3
Pulmonary rehabilitation + upper extremity exercises (Köseoğlu et al., 1997)	15 (3 × 5)	60	5
Aerobic + qigong (Burini et al., 2006)	40c	45-50	14
Inspiratory muscle training (Inzelberg et al., 2005)	72 (6 × 12)	30	12
EMST (Byeon et al., 2016; Claus et al., 2021; Kuo et al., 2017; Pitts et al., 2009; Sapienza et al., 2011; Troche et al., 2010; Troche et al., 2014)	20 (5 × 4)	10-20	4
EMST (Darling-White et al., 2017)	24d	30	8
Respiratory muscle training (inspiratory muscle training or EMST; Reyes et al., 2018)	48 (6 × 8)	NR^a	8
Respiratory muscle training (inspiratory muscle training or EMST; Huang et al., 2020)	20 (5 × 4)	50	4
EMST + swallow postural techniques (Byeon, 2016)	20 (5 × 4)	50	4
EMST + air stacking (Reyes et al., 2020)	48 (6 × 8)	NR^a	8
Voice loading therapies
Voice + choral singing treatment (Di Benedetto et al., 2009)	33e	180	10-13
Therapeutic singing (Stegemöller, Hibbing, et al., 2017; Stegemöller, Radig, et al., 2017)	16 (2 × 8)	60	8
ParkinSong (Tamplin et al., 2019)	12(1 × 12)	120	12
ParkinSong (Tamplin et al., 2020)	52(1 × 52)	120	52
Lee-Silverman voice treatment (El Sharkawi et al., 2002; Miles et al., 2017)	16 (4 × 4)	50-60	4

Note. NR = not reported; rTMS = repetitive transcranial magnetic stimulation; SES = surface electrical stimulation; NMES = neuromuscular electrical stimulation; CDT = conventional dysphagia therapy; EMST = expiratory muscle strength training.

^aQuantified the exercises and repetition but not mentioned the exact duration of a session.

^b2 weeks (5 days/week for 10 days) and five booster sessions every month for 3 months.

^cAerobic training for 7 weeks (45 min for each session), 8 weeks no treatment, and qigong for 7 weeks (50 min for each session; 22 weeks in total).

^d4 Pre-training sessions (1.5 hr, once a week for 4 weeks) and four training sessions (30 min, once a week for 4 weeks only; for two participants it took 5 weeks).

^e20 hr of collective speech therapy (two sessions of 1 hr every week over 10 weeks for 20 sessions in total), plus 26 hr of choral singing (13 sessions of 2 hr every week for 13 weeks).

^fCross-over trial each technique was conducted once with an interval of 7 days between each of the intervention.

Study characteristics

A diverse range of physical therapies were identified: swallow-related (n = 5), electromagnetic stimulation (n = 4), respiratory loading (n = 20), and voice loading (n = 7) therapies. Flexible endoscopic evaluation of swallowing (FEES; n = 5), videofluoroscopic swallow study (VFSS; n = 11), and surface electromyography (sEMG; n = 4) were widely used to measure swallow function. The most commonly reported objective respiratory/cough outcome measures were pressure manometer, spirometer, and pulmonary function tests measures. Validated questionnaires were utilised to provide patient reported swallow-related quality of life measures (Supplementary Appendix 4).

Swallow-related therapies

Of the five studies grouped under the swallow-related therapies, three studies targeted treatment combinations of a range of movement exercises, tongue pressure exercises, and swallow manoeuvres (Argolo et al., 2013; Plaza & Ruviaro Busanello-Stella, 2022; Wang et al., 2018). The other two studies focused on biofeedback based swallow training and skill training approaches (Athukorala et al., 2014; Manor et al., 2013). Motor swallow exercises (Argolo et al., 2013) demonstrated significant reduction in pharyngeal residue, piecemeal deglutition, and improvement in bolus control but did not significantly change the swallow safety scores. Plaza and Ruviaro Busanello-Stella (2022) reported tongue isometric pressure exercises resulted in significant improvement in Functional Oral Intake Scale (FOIS) score, and large effects on tongue muscle strength (d = 2.13) and suprahyoid muscle activity (d = 6.33). Also, there was a strong significant correlation between suprahyoid muscle activity and tongue strength (r = 0.994, p < .001). Combination of effortful swallow manoeuvre and tongue range of movement exercises (Wang et al., 2018) demonstrated improvements in laryngeal excursion amplitude. Athukorala et al. (2014) reported large effects (d > 1) in swallow timing measures (pre-motor time, pre-swallow time, duration of submental muscle contraction) and time per swallow (d = 0.72) after biofeedback in strength and skill training (BiSSkiT). Video assisted skill training (VAST) study trial (Manor et al., 2013) showed significant reduction in pharyngeal residue, yet significant change in the swallow safety score was not evident. Significant improvement was reported in Swallowing Quality of Life (SWAL-QOL) and Eating Assessment Tool (EAT-10) scores (Argolo et al., 2013; Athukorala et al., 2014; Manor et al., 2013; Plaza & Ruviaro Busanello-Stella, 2022). VAST study showed follow-up improvement in QOL at 6 months (Manor et al., 2013). Maintenance effects of the swallow functions were evident for some treatments (tongue isometric pressure exercise for 4 weeks and BiSSkiT for 2 weeks).

Electromagnetic stimulation therapies

Of the four studies addressing the effects of electrical/magnetic stimulation, one study evaluated the effects of high-frequency repetitive transcranial magnetic stimulation (rTMS) on advancing PD with dysphagia (Khedr et al., 2019). The other three studies focused on combined treatments with the inclusion of electrical stimulation (surface electrical stimulation [SES]; Baijens et al., 2013) and neuromuscular electrical stimulation (NMES; Heijnen et al., 2012; Park et al., 2018) alongside behavioural therapies in the intervention group and dysphagia therapies alone in the control group (Baijens et al., 2013; Heijnen et al., 2012; Park et al., 2018). Significant effects (p < .05) were reported in swallow duration measures of hyoid movement and pharyngeal transit time following rTMS, but no effects demonstrated for swallow safety scores (Khedr et al., 2019). Significant reduction in piecemeal deglutition and initiation of pharyngeal reflex were evident following SES (Baijens et al., 2013). Large effects were evident in displacement of hyoid (d = 0.9) and swallow safety score was significantly reduced following NMES in combination with effortful swallow and conventional dysphagia therapy (Park et al., 2018). No effects on FOIS score (Heijnen et al., 2012) were reported after NMES. There was a significant reduction in Arabic Dysphagia Handicap Index (A-DHI) and Unified Parkinson's Disease Rating Scale (UPDRS III) scores after rTMS therapy (Khedr et al., 2019). No change was demonstrated between groups for Dysphagia Severity Scale (DSS) and SWAL-QOL after NMES (Heijnen et al., 2012). Treatment effects were maintained at 3 months for rTMS (Khedr et al., 2019). Self-reported outcomes were maintained for NMES (for 3 months; Heijnen et al., 2012) and rTMS (for a month; Khedr et al., 2019). None of the studies reported maintenance effects on objective outcomes.

Respiratory loading therapies

The 20 studies categorised under respiratory loading therapies targeted respiratory muscle strength training (inspiratory and/or expiratory; Claus et al., 2021; Cocks et al., 2022; Darling-White & Huber, 2017; Huang et al., 2020; Inzelberg et al., 2005; Kuo et al., 2017; Pitts et al., 2009; Reyes et al., 2018; Sapienza et al., 2011; Troche et al., 2010), general strength training (Alves et al., 2019; Mavrommati et al., 2017), functional training exercises (Pelosin et al., 2009), pulmonary exercises (Güngen et al., 2017), and breath-stacking or intensive spirometry (Ribeiro et al., 2018). Some studies used combination of approaches (Burini et al., 2006; Byeon, 2016; Reyes et al., 2020). Studies with EMST training had significant effects (p < .05) on voluntary cough measures (Pitts et al., 2009; compression phase duration, expiratory phase rise time, and cough volume acceleration; p < .05), large (d = 0.89; Reyes et al., 2018) to moderate effects (d = 0.77; Cocks et al., 2022; Reyes et al., 2020) on voluntary peak cough flow (PCF), and small effects on reflexive PCF (d = 0.27; Reyes et al., 2018; Reyes et al., 2020). A combination of EMST plus air stacking training had large effects on both voluntary (d = 1.00) and reflexive (d = 1.34) PCF measures (Reyes et al., 2020). A number of EMST studies demonstrated positive effects on swallow outcomes, which are explained in further detail below, under cross-system effects (Byeon, 2016; Claus et al., 2021; Cocks et al., 2022; Pitts et al., 2009; Troche et al., 2010). EMST training studies had significant effects on maximum expiratory pressure (MEP; Claus et al., 2021; Cocks et al., 2022; Darling-White & Huber, 2017; Huang et al., 2020; Inzelberg et al., 2005; Kuo et al., 2017; Pitts et al., 2009; Reyes et al., 2018; Reyes et al., 2020; Sapienza et al., 2011; Troche et al., 2010) and one follow-up study reported improved MEP was maintained after 3 months as the change was not significant when compared to immediate post-treatment values (Troche et al., 2014). Maximum inspiratory pressure (MIP) measure showed only small effect size (d = 0.42) following EMST training (Huang et al., 2020; Reyes et al., 2018; Reyes et al., 2020). Inspiratory muscle training (Inzelberg et al., 2005) also had significant effects (p < .05) on MIP and peak pressure in one study. No change in respiratory measures was reported in between group comparison of a dual arm respiratory muscle training study (with inspiratory and expiratory muscle training; Huang et al., 2020). Full-body exercises had significant positive effects on respiratory function (Mavrommati et al., 2017; Pelosin et al., 2009) and inspiratory pressure measures (Alves et al., 2019), while no change in expiratory pressures was noted (Alves et al., 2019). Pulmonary techniques (Güngen et al., 2017; Köseoğlu et al., 1997) and aerobic exercises (Burini et al., 2006) had significant positive effects on some of the pulmonary function test measures. Qigong technique alone did not affect any of the tested measures (Burini et al., 2006). Significant improvement was shown in tidal volume and minute ventilation after using breath-stacking and incentive spirometry techniques (Ribeiro et al., 2018). Treatment effects were maintained at 1 month for treadmill (Pelosin et al., 2009), 2 months for aerobic training (Burini et al., 2006), and 3 months for EMST (Claus et al., 2021; Troche et al., 2014). Pulmonary physio (Güngen et al., 2017) and full-body exercises (Pelosin et al., 2009) study trials had significant effects on disease severity measures, but this was not reported in other studies (Alves et al., 2019; Inzelberg et al., 2005; Kuo et al., 2017). Furthermore, two studies, each with two treatment arms, did not report any significant effects between groups on disease severity scores (Burini et al., 2006; Inzelberg et al., 2005).

Voice loading therapies

While vocal therapies were not the focus of this review, seven studies used intervention approaches specifically focused on improving voicing in PD, included voice and choral singing treatment (VCST; Di Benedetto et al., 2009), Lee Silverman voice treatment (LSVT; El Sharkawi et al., 2002; Miles et al., 2017), therapeutic singing (Stegemöller, Hibbing, et al., 2017; Stegemöller, Radig, et al., 2017), and ParkinSong (Tamplin et al., 2019; Tamplin et al., 2020). These intensive therapy programs incorporated vocal exercises with speech activities and/or singing. These studies met criteria for inclusion in this review as they demonstrated positive effects on swallow and respiratory/cough outcomes, which are explained in further detail below, under cross-system effects. It is outside the scope of this review to provide a detailed analysis of voice outcomes for these studies and we refer the reader to the scientific papers for these details (summary of the findings were listed in Table V).

Table V. Studies with cross-system effects.

Intervention (Author, year)	Study design	Immediate effects and maintenance effects	Risk of bias rating and level of evidence
Electrical/magnetic stimulation therapies
IG: rTMS CG: Sham rTMS (Khedr et al., 2019)	Randomised controlled trial	Swallow: VFSS (5 ml fluid, semisolid, solid): Solid swallows. ↓ Duration of hyoid movement*, ↓ pharyngeal transit time*, ↔ residue score. Fluid swallows. ↔ Duration of hyoid movement*, ↔ pharyngeal transit time*. ↔ PAS score. QOL: ↓ A-DHI^a, ↓ UPDRS III Follow-up: 3 months: quality of life measure improvements maintained at follow-up (no VFSS done at follow-up).	High I b
Respiratory loading therapies
IG: EMST CG: Sham EMST (Troche et al., 2010)	Randomised controlled trial	Swallow: VFSS (ten 5 ml trials of thin liquid and a trail of 3 oz sequential cup drinking): ↓ Duration of hyoid elevation, ↑ onset of bolus transit, ↑ UES opening*, ↑ UES maximum opening*, ↑ UES closure*, ↑ laryngeal closure, ↑ maximum laryngeal closure, ↑ laryngeal opening. QOL: ↑ SWAL-QOL. Functional swallow outcome: VFSS. ↔PAS score*, moderate effect (d = 0.55).	Low I b
IG: EMST CG: Sham EMST (Claus et al., 2021)	Randomised controlled trial	Swallow: FEES (3 × 8 ml puree-IDDSI 4; 3 × 5 ml liquid-IDDSI 0; 3 × 3 × 0.5 cm^3 white bread-IDDSI 7): ↓ Total score*, ↓ residue score*, ↔ premature spillage, ↔ PAS score (baseline measure indicating no evidence of penetration-aspiration). Magnetencephalography: ↔ cortical activation (localised activation in bilateral pericentral cortex). QOL: ↓SDQ*, ↔ SWAL-QOL. Follow-up at 3 months: Improved measures were maintained at follow-up.	Low I b
IG: EMST + swallow postural techniques CG: EMST only (Byeon, 2016)	Randomised controlled trial	Functional swallow: VFSS: ↓ Videofluoroscopy Studies Scale Score.^a^,^b	Low I b
IG: EMST (Pitts et al., 2009)	Non-controlled trial (Before and after trial)	Cough/ respiratory: Pressure manometer: ↑ Maximum expiratory pressure*. Pneumotachograph & spirometer: Voluntary cough measures. ↔ Inspiratory phase duration, ↓ compression phase duration*, ↓ expiratory phase rise time*, ↑ cough volume acceleration*, ↔ inspiratory phase peak flow, ↔ expiratory phase peak flow, ↔ inspiratory phase duration, expiratory phase rise time. Functional swallow: VFSS (30 ml thin sequential swallow task): ↓ PAS score*.	Moderate III
IG: EMST (50% peak pressure threshold for each week; Cocks et al., 2022)	Non-controlled trial (Before and after trial)	Swallow: ↑ MASA scores*. Other: ↓ Sialorrhea* (SCS-PD), ↑ lip strength* (IOPI). Cough/respiratory: Spirometer. ↑ Voluntary peak cough flow*. QOL: ↑ SWAL-QOL*.	Serious III
Voice loading therapies
IG 1: Therapeutic singing high dosage IG2: Therapeutic singing low dosage (once a week session for 8 weeks; Stegemöller et al., 2017)	Non-randomised controlled trial (Cohort study)	Electromyography (3 × 10 ml thin condition and 3 × 10 ml pudding thick condition): LM = laryngeal muscles; SM = submental muscles; AUC = area under curve; EMG = electromyography. Thin liquid: Amplitude: ↔ LM peak amplitude, ↔ LM AUC, ↔ SM peak amplitude, ↔ SM AUC. Timing: ↔ LM rise time*, ↔ LM fall time*, ↔ LM EMG duration*, ↔ SM rise time, ↔ SM fall time*, ↔ SM EMG duration*. Thick liquid: Amplitude: ↔ LM AUC, ↑ SM peak amplitude*, ↔ SM AUC*, ↔ LM peak amplitude, ↔ LM AUC. Timing: ↑ SM fall time*, ↔ SM EMG duration*, ↔ SM rise time, ↔ LM rise time*, ↔ LM fall time*, ↔ LM EMG duration*. Group and condition interaction effects: Thick-SM AUC*, Thin-SM fall time*, ↔ all other measures. QOL: ↔ SWAL-QOL, ↓ UPDRS total score*.	Serious II a
IG 1: Therapeutic singing high dosage IG2: Therapeutic singing low dosage (once a week session for 8 weeks; Stegemöller et al., 2017)	Non-randomised controlled trial (before and after trial)	Cough/respiratory: Pressure metre: ↔ maximum expiratory pressure^b, ↔ maximum inspiratory pressure^b Other measures: Voice and speech (decibel metre, chromatic tuner on iPad): ↔ phonation duration^b, ↔ phonation range, ↔ vocal intensity. QOL: ↔ voice related quality of life*.	Serious II a
IG1: ParkinSong (weekly for 12 weeks) IG2: ParkinSong (monthly for 3 months) CG1: Support group (weekly for 12 weeks) CG2: Support group (monthly for 3 months; Tamplin et al., 2019)	Non-randomised controlled trial (4-arms study)	Cough/respiratory: Group comparison: IG1 vs other groups Pressure metre: ↑ maximum expiratory pressure^a, ↔ maximum inspiratory pressure, ↔ sniff nasal inspiratory pressureOther measures: Group comparison: IG1 vs other groups Voice and speech: (PRAAT software for speech analysis and sound level metre): ↑ vocal intensity monologue^a, ↑ vocal intensity loud sentence task^a, ↑ intelligibility (speech intelligibility test), ↑ intelligibility (perceptual), ↔ maximum phonation length. ↓ Voice-related QOL*.	Serious II a
IG1: ParkinSong (weekly for 52 weeks) IG2: ParkinSong (monthly for 12 months) CG1: Support group (weekly for 52 weeks) CG2: Support group (monthly for 12 months; Tamplin et al., 2020)	Non-randomised controlled trial (continuation of Tamplin et al., 2019)	Cough/ respiratory: Group comparison: IG1 vs other groups Pressure metre: ↔ maximum expiratory pressure, ↔ maximum inspiratory pressure, ↔ sniff nasal inspiratory pressure. Other measures: Group comparison: IG1 vs other groups Voice and speech: (PRAAT software for speech analysis and sound level metre): ↑ vocal intensity (monologue^a),↑ intelligibility (perceptual), ↔ vocal intensity loud sentence task, ↔ intelligibility (SIT), ↔ maximum phonation time. ↓ Voice-related QOL^a	Serious II a
IG: VCST (Di Benedetto et al., 2009)	Non-controlled trial (pilot test-retest non-controlled study)	Swallow: Videostroboscopy: Vocal folds (VF) ↔ VF movement symmetry, ↔ VF movement amplitude, ↔ glottic closure configuration, ↔ VF rest tremor, ↔ VF tremor during phonation. Cough/respiratory: Spirometry: ↓ functional residual capacity*, ↔ forced vital capacity percentage, ↔ forced expiratory volume in one second percentage. Pressure metre: ↑ maximum expiratory*, ↑ maximum inspiratory pressure*. Other measures: Speech and voice (CSL- MDVP): ↑ maximum phonation time*, ↔ frequency measures, ↔ perturbation measures, ↔ frequency and amplitude tremor index. Quality of voice analysis: ↑ prosodia reading visual analogue scale*, ↔ prosodia monologue visual analogue scale, ↓ fatigue during reading*, ↔ fatigue during monologue.	Serious III
IG: LSVT (Miles et al., 2017)	Non-controlled trial (Prospective non-randomised single-blinded cohort intervention study)	Swallow: VFSS (20 ml thin liquid, 100 ml thin liquid sequential swallow, 5 ml paste): Timing measures: ↑ PES opening duration*, ↔ bolus transit time, ↔ airway closure time during swallow, ↔ duration of hyoid maximally displaced during swallow. Displacement measures: ↑ PES maximum opening*, ↔ maximum hyoid displacement, ↔ maximum hyolaryngeal displacement, ↔ pharyngeal constriction ratio. Anatomical measure: ↓ pharyngeal area at rest*. ↓ paste residue*. Straw drinking (100 ml): ↔ duration, ↔ number of swallows, ↔ number of breaths. Cough/respiratory: Spirometry reflexive cough: ↔ compression phase duration, ↑ peak expiratory flow rate*, ↑ peak expiratory flow rise time*, ↔ cough volume acceleration. Other measures: Voice and speech (sound level metre): ↑ maximum phonation time*, ↑ sound pressure level reading*, ↑ sound pressure level conversation*. QOL/disease severity: ↓ Parkinson’s Disease Questionnaire-8*, ↓ Eating Assessment Tool-10*. Follow-up (6 months): improvements were maintained at follow-up.	Low III
IG: LSVT (El Sharkawi et al., 2002)	Non-controlled trial (before and after trial)	Swallow: VFSS (Two trials of 1 ml, 3 ml, 5 ml, and 10 ml cup drinking, 2 ml pudding paste, ¼ cookie): Swallow motility pre-post treatment comparison: Improved: tongue lateralisation, tongue coordination, antero-posterior tongue movement (100% cup drinking), antero-lateral tongue stabilisation, tongue strength, pharyngeal swallow trigger (25% liquids, 66% cookie). Reduced: rock-like tongue motion, oral residue (50% liquids, 12.5% paste, 25% cookie), tongue base and valleculae residue (50% in total for all bolus sizes and consistency). Not improved: tongue lateralisation on cookie bolus, delayed laryngeal vestibule closure. Penetration and aspiration: not evident before or after intervention. Temporal measure: ↓ oral transit time for paste*, ↓ oral residue percentage for 3 ml and 5 ml*, ↑ oropharyngeal efficiency scorefor cup drinking*. (↔ for other volume and consistencies trialled) Other measures: Voice and speech (sound level metre and CSpeech): ↑ sound pressure level for vowel phonation, reading, happy day conversation*, ↔ fundamental frequency (reading and happy day), ↔ clarity of speech, ↔ voice handicap index.	Moderate III

Note. IG = intervention group; CG = control group. ↑ = significantly increased; ↔ = no significant change; ↓ = significantly decreased (improved); level of significance = p < .05; * = significant within-group change from pre- to post-intervention in the intervention group. A-DHI = Arabic Dysphagia Handicap Index; EMST = expiratory muscle strength training; FEES = fiberoptic endoscopic evaluation of swallowing; IDDSI = International Dysphagia Diet Standardisation Initiative; IOP I = Iowa Oral Performance Instrument; LSVT = Lee Silverman Voice Treatment; MASA = Mann Assessment of Swallowing Ability; PAS = Penetration-Aspiration Scale; P ES = pharyngoesophageal sphincter; QOL = quality of life; rT MS = repetitive transcranial magnetic stimulation; SCS-P D = Sialorrhea Clinical Scale for Parkinson’s Disease SDQ = Swallowing Disturbances Questionnaire SWAL-QOL = Swallowing Quality of Life questionnaire; UES = upper esophageal sphincter; UP DRS = Unified Parkinson’s Disease Rating Scale; VCST = Voice and choral singing treatment; VFSS = videofluoroscopy swallow study.

^aChange in intervention group significantly larger than change in control group.

^bSignificant within-group change from pre- to post-intervention in both intervention and control group.

Cross-system outcomes

Sixty-seven percent of the studies (n = 24/36) used more than one treatment strategy and one-third of the studies (n = 11/36) investigated the spread effects of a treatment specifically focused on a voice, swallowing, or respiratory activity to another body function (swallow, cough, respiratory). Table V summarises the outcomes of the interventions that demonstrated cross-system effects.

Swallow outcomes

Only one electrical/magnetic stimulation study (using rTMS alone) showed significant improvement (p < .05) in swallowing measures, with improved duration of hyoid movement and pharyngeal transit time for solid swallows in the intervention group when compared to the sham group (Khedr et al., 2019). However, no significant change was reported for liquid bolus trials. Other electrical stimulation studies (SES, NMES) with combination of swallow therapies did not demonstrate significant improvement in swallow measures in between-groups comparison, although, significant changes were reported in pre- and post-session comparisons within groups.

Amongst respiratory loading therapies, five of the expiratory muscle strength training studies evaluated the effects in objective swallow measures (Byeon, 2016; Claus et al., 2021; Cocks et al., 2022; Pitts et al., 2009; Troche et al., 2010). Studies using EMST found significant (p < .05) positive effects on swallow timing (Pitts et al., 2009; Troche et al., 2010), residue score (Claus et al., 2021), and PAS score (Pitts et al., 2009; Troche et al., 2010). However, Claus et al. (2021) found no significant change in PAS scores after EMST, as all patients lacked penetration and aspiration at baseline. Byeon (2016) reported significant reduction in overall VFSS score following EMST. A 3 months follow-up study of EMST with 10 participants by Troche et al. (2014) reported that change of PAS score was not significant when compared to immediate post-treatment. In questionnaire metrics, Swallowing Disturbance Questionnaire (Claus et al., 2021) and SWAL-QOL (Cocks et al., 2022; Troche et al., 2010) scores were significantly improved after EMST therapy.

Of the voice loading studies, two small non-randomised LSVT studies demonstrated significant change (p < .05) in swallow temporal measures (El Sharkawi et al., 2002; Miles et al., 2017), bolus residue (El Sharkawi et al., 2002; Miles et al., 2017), and maximum displacement of pharyngoesophageal sphincter (PES; Miles et al., 2017) on the VFSS findings. Treatment effects were maintained at 6 months for LSVT (Miles et al., 2017). No significant change was evident in hyoid timing and hyolaryngeal displacement measures after pre/post comparison of LSVT (Miles et al., 2017). A study utilising therapeutic singing (Stegemöller, Hibbing, et al., 2017) reported significant changes (p < .05) in electromyography study for submental muscles amplitude and submental muscle fall time for thick liquid trials in between group comparison of low and high frequency treatment groups. In contrast, VCST showed no significant change in vocal fold movements, tremor, and glottic closure in videostroboscopy ratings (Di Benedetto et al., 2009). There was a significant reduction (improvement; p < .05) in Eating Assessment Tool-10 and Parkinson’s Disease Questionnaire-8 scores after LSVT (Miles et al., 2017). Other authors report no significant change in SWAL-QOL score, but significant improvement in UPDRS scores after therapeutic singing (Stegemöller, Radig, et al., 2017).

Respiratory/cough outcomes

Five of the voice loading therapies (Di Benedetto et al., 2009; Miles et al., 2017; Stegemöller, Hibbing, et al., 2017; Stegemöller, Radig, et al., 2017; Tamplin et al., 2019; Tamplin et al., 2020) reported objective respiratory/cough indices as one of the primary outcome measures (Table II). Significant improvement (p < .05) in reflexive cough measures (peak expiratory flow rate and peak expiratory flow rise time) were reported after LSVT and the effects were maintained at 6 months (Miles et al., 2017). VCST showed significant improvement in functional residual capacity and maximum inspiratory and expiratory pressure measures (Di Benedetto et al., 2009). Therapeutic singing treatment showed significant improvements in MIP and MEP measures despite a dosage difference between groups (Stegemöller, Radig, et al., 2017). ParkinSong treatment showed significant effects (p < .05) on MEP measure after 3 months, but the change was not significant at 12 months post-treatment (Tamplin et al., 2020; Tamplin et al., 2019).

Methodological quality and risk of bias

Table VI presents the best evidence statements of behavioural therapies from the 11 studies rated low risk of bias. Risk of bias ratings of the 15 randomised controlled trials are displayed in Figure 2 for each of the RoB2 criteria (Sterne et al., 2019). Twenty-three percent of studies were categorised as exhibiting high risk of bias. No information on blinding of the assessors was rated as an area of high risk in five of the RCTs as it is most likely that assessment was influenced by knowledge of intervention received. A high percentage of missing data reported in one study was categorised as a significant risk, resulting in half of the allocated participants being excluded from the final analysis (Inzelberg et al., 2005).

Figure 2. Risk of bias for randomised (RoB2) and non-randomised (ROBINS-I) trials.

Table VI. Best evidence statement of behavioural therapies in improving swallow and respiratory/cough measures in individuals with Parkinson’s disease.

Outcome measures	Intervention	Best evidence statement (summary findings of the significant changes of the outcome measures from the studies rated low risk of bias)
Swallow	VAST (Manor et al., 2013)	Level 1b evidence: reduction in pharyngeal residue (FEES).
	Surface electrical stimulation and logopaedic dysphagia therapy (Baijens et al., 2013)	Level 1b evidence: reduction in piecemeal deglutition (VFSS), reduction in delayed initiation pharyngeal reflex (FEES; entire population pre- and post-comparison, no group difference).
	EMST (Troche et al., 2010)	Level 1b evidence: VFSS durational measures. Reduction in hyoid elevation, increase in onset of bolus transit, increase in UES opening, increase in UES maximum opening, increase in UES closure, increase in laryngeal closure, increase in maximum laryngeal closure, increase in laryngeal opening. Penetration-Aspiration Scale scores: moderate effect on swallow functional outcome (d = 0.55).
	EMST (Claus et al., 2021)	Level 1b evidence: reduction in total score, reduction in residue score (FEES).
	EMST and swallow postural techniques (Byeon, 2016)	Level 1b evidence: reduction in Videofluoroscopy Studies Scale score.
	Motor swallowing exercise (Argolo et al., 2013)	Level III evidence: significant association between treatment failure vs lingual pumping and treatment failure vs dental absence. VFSS examination: reduction in loss of bolus control, reduction in piecemeal swallow, reduction in tongue residue, reduction in pyriform residue, reduction in vallecular residue.
	LSVT (Miles et al., 2017)	Level III evidence: increase in PES opening duration, increase in PES maximum opening, reduction in pharyngeal area at rest, reduction in paste residue (VFSS).
Respiratory/ cough	EMST (Sapienza et al., 2011)	Level 1b evidence: increase in maximum expiratory pressure (pressure manometer).
	Full-body strength training (Alves et al., 2019)	Level 1b evidence: increase in maximum inspiratory pressure with moderate effect size (manovacuometer).
	Aerobic training (Burini et al., 2006)	Level 1b evidence: reduction in peak oxygen uptake, reduction in double product peak, reduction in oxygen uptake rate (spirometer), reduction in perception of breathlessness (Borg scale). Increase is six minute walking test (exercise tolerance).
	Breath-stacking and incentive spirometer (Ribeiro et al., 2018)	Level 1b evidence: increase in total tidal volume (plethysmography).
	LSVT (Miles et al., 2017)	Level III evidence: spirometry reflexive cough study. Increase in peak expiratory flow rate, increase in peak expiratory flow rise time.
QOL/disease severity	VAST (Manor et al., 2013)	Level 1b evidence: decrease in Swallowing Disturbance Questionnaire score, increase in SWAL-QOL score.
	EMST (Troche et al., 2010)	Level 1b evidence: increase in SWAL-QOL score.
	EMST (Claus et al., 2021)	Level 1b evidence: decrease in Swallowing Disturbance Questionnaire score.
	Motor swallowing exercises (Argolo et al., 2013)	Level III evidence: SWAL-QOL. Increase in symptom frequency, increase in fear domains. Swallow complaints. Decrease in difficulty in moving food when chewing.
	LSVT (Miles et al., 2017)	Level III evidence: decrease in Parkinson’s Disease Questionnaire-8, decrease in Eating Assessment Tool-10 score.

See Supplementary Appendix 3 for full definitions of levels of evidence (Level 1a = meta-analysis of RCTs and Level IV = expert option/clinical experience only).

Note. Reported measures: level of significance p ≦ .05. EMST = expiratory muscle strength training; FEES = fiberoptic endoscopic evaluation of swallowing; LSVT = Lee Silverman voice treatment; PES = pharyngoesophageal sphincter; QOL = quality of life; SWAL-QOL = Swallowing Quality of Life questionnaire; UES = upper oesophageal sphincter; VAST = video assisted swallow training; VFSS = videofluoroscopy.

Figure 2 shows the calculations for each of the ROBINS-I criteria (Sterne et al., 2016) for the 21 non-randomised trials. Significant limitations were found in the majority of the studies leading to an overall risk rating of moderate to serious risk in 90% of the non-randomised studies. Treatment adherence and blinding of assessors were not reported in many studies. Participant recruitment processes were not clearly defined in three of the studies (Darling-White & Huber, 2017; Huber et al., 2003; Pitts et al., 2009; Reyes et al., 2020). Five studies reported on imbalanced assignment of participants to intervention and control groups (no power calculation was provided to gauge the sensitivity of the study to detect an effect) and this was categorised as selection bias (Mavrommati et al., 2017; Stegemöller, Hibbing, et al., 2017; Stegemöller, Radig, et al., 2017; Tamplin et al., 2019; Tamplin et al., 2020). Two studies were rated serious risk for missing data, as more than 20% of the participants dropped out before final analysis and no reasons were given (Güngen et al., 2017; Mavrommati et al., 2017).

Discussion

This review has evaluated a substantial number of studies that support the efficacy of behavioural therapies in improving swallow, respiratory, and cough functions in individuals with PD. A total of 36 studies met the selection criteria and interventions were clustered into four thematic treatment types: swallow-related therapies, electrical/magnetic stimulation therapies, respiratory-loading therapies, and voice loading therapies. Behavioural therapies are gaining a strong body of supporting evidence in PD rehabilitation and represent as an important intervention modality that is feasible, safe, and beneficial. Most of these studies evaluated treatment effects using objective, instrumental measures of swallowing and cough/respiratory functions and several therapies were rated Level 1b evidence for positive swallowing, cough, and QOL outcomes. Our aims to evaluate the effects of therapies on aspiration and pulmonary health were not the focus of published papers. Many therapies require high-quality RCT replication to add strength to their evidence-base, such as video assisted swallow training and strength and skill-based swallow programs. Evidence for electrical/magnetic stimulation therapies remains in its infancy. EMST and LSVT have the advantage of improving both swallowing and respiratory/cough function. Our critical review of study methodology and results was consistent with previous systematic reviews (Baijens & Speyer, 2009; Gandhi & Steele, 2022; Kim & Kim, 2023; López-Liria et al., 2020; Park et al., 2019; van Hooren et al., 2014). However, we provide an updated and novel critical appraisal of studies examining cross-system effects of other therapy programs, which have not previously been reported.

Cross-system effects

Respiration, cough, and swallow functions are physiologically interrelated through shared anatomy and functional interdependence. There is a critical interrelationship through cortical control coordinating breathing and swallowing (Martin-Harris et al., 2022; Troche et al., 2010). Recent studies have investigated the correlational aspects of respiration and cough parameters, respiratory/cough-swallow coordination, and the interrelationship of cortical control of these functions. Study findings (Jo & Kim, 2016; Kaneko et al., 2019) demonstrated a significant correlation between respiratory parameters such as forced vital capacity (FVC), MIP, MEP, and cough peak flow. Curtis et al. (2020) demonstrated a significant interrelationship between tidal breathing and respiratory-swallow coordination in people with PD.

With this in mind, respiration-swallow cross-system approaches are beginning to be utilised to treat patients with swallow dysfunction (Martin-Harris et al., 2017). Though EMST primarily focuses on strengthening the expiratory muscles, studies with high-quality evidence show promising effects on swallow efficiency measures that are in line with previous reviews (Kim & Kim, 2023; López-Liria et al., 2020; Mancopes et al., 2020; van Hooren et al., 2014). Seven studies (LSVT, therapeutic singing, ParkingSong, and voice and choral singing) categorised under voice training also demonstrate statistically significant improvement in selected swallow and respiratory/cough measures. However, none of these studies as yet have proven the effects with high-level of evidence and six of these studies (Di Benedetto et al., 2009; El Sharkawi et al., 2002; Stegemöller, Hibbing, et al., 2017; Stegemöller, Radig, et al., 2017; Tamplin et al., 2019; Tamplin et al., 2020) have methodological limitations that prevent the drawing of strong conclusions. This review, however, provides promising evidence that many of the treatment options available for improving respiratory and cough functions demonstrate promising spread effects benefitting swallowing function. Clinically, swallow management in PD needs to be extended beyond traditional dysphagia therapy and utilise a combination of approaches to achieve best outcomes for airway protection, as well as swallowing efficiency, particularly given the lack of benefit from pharmacological treatments.

Methodological quality and best level of evidence

Although the selected 36 studies demonstrate positive gains following a variety of behavioural therapies, only a small number of studies were rated as low risk of bias with high-quality evidence. This reflects the difficulty in performing such interventional studies, particularly longitudinally, and in finding suitable comparison groups. In this review, nine randomised controlled trials with low-risk bias ratings have provided the best level of evidence regarding improved swallow and respiratory/cough functions in individuals with PD (Table VI). Among swallow-related therapies, video assisted swallow therapy (Manor et al., 2013) is supported with high-quality evidence showing reduction in pharyngeal residue. Surface electrical stimulation study (Baijens et al., 2013) also showed improvement in swallow efficiency parameters through combining traditional swallow and logopaedic therapy. The scarcity of other controlled trials in swallow therapies perhaps hinders generalisations of the findings.

In comparison, EMST studies (Byeon, 2016; Claus et al., 2021; Sapienza et al., 2011; Troche et al., 2010) demonstrate consistent improvements in instrumental swallowing measures (total score and durational measures for VFSS; total and residue scores for FEES) as well as significant improvements in maximum expiratory pressure measures, which supports its role in cough augmentation (Jo & Kim, 2016). Miles and colleagues’ pilot LSVT study (2017) demonstrated improvement in VFSS findings for pharyngoesophageal sphincter durational and displacement measures, reduction in paste residue, and improvement in reflexive cough peak expiratory flow measures. Yet these findings need validation with high-level evidence and larger cohorts with varying disease severity to generalise to the wider PD population. Full-body exercises (Alves et al., 2019), aerobic training (Burini et al., 2006), and breath-stacking or incentive spirometry (Ribeiro et al., 2018) techniques were all effective treatments to improve pulmonary measures in PD and were supported by high-quality evidence.

Combination of treatment approaches

Interestingly, use of a combination of treatment approaches is gaining support in recently published literature. Many studies included in this review used a combination of treatment approaches. In swallow behavioural therapies, specific swallow manoeuvres were combined with the range of movement exercises (Argolo et al., 2013; Wang et al., 2018) and resistance pressure exercises were combined with the range of movement exercises (Plaza & Ruviaro Busanello-Stella, 2022). Combining biofeedback therapy (Athukorala et al., 2014; Manor et al., 2013) along with individualised swallow skill/behavioural training also demonstrated promising effects on swallow efficiency scores. Similar approaches are now being explored for cough, with emerging studies exploring skill training and voluntary upregulation of reflex cough with biofeedback approaches (Brandimore et al., 2017; Curtis et al., 2020; Curtis & Troche, 2020; Hegland et al., 2012). Combination therapies, while common, make it difficult to draw conclusions on efficacy of individual therapies.

All the electromagnetic swallow therapies (except rTMS) were coupled with traditional logopaedic therapies (Baijens et al., 2013; Heijnen et al., 2012; Park et al., 2018). Respiratory training interventions targeted combining breathing and body exercises (Burini et al., 2006; Köseoğlu et al., 1997; Mavrommati et al., 2017) or resistance pressure training and air stacking (Reyes et al., 2020). EMST plus air stacking approaches demonstrated large effects on voluntary and reflexive cough parameters (Reyes et al., 2020). However, further RCTs with low risk of bias ratings are warranted. Another RCT targeted combining EMST and swallow postural techniques (Byeon, 2016), demonstrating improved VFSS parameters when compared to EMST alone. All the voice training interventions incorporated more than one voice training/singing or functional speech activity in their protocol, with minimum 50 min per session. Further studies with high-quality evidence, low bias, with detailed study protocols and objective outcomes are needed to further understand the effects on respiratory/cough and/or swallow function.

Study limitation and future directions

This review included only original articles published in English. It is possible that some eligible studies were missed due to this search strategy (Supplementary Appendix 2), although every effort was made to perform a comprehensive database search. Data extraction, analysis, and risk of bias assessments were conducted manually by one author and cross-checked with a second author. We included articles that had limitations to methodological quality. Common concerns found in most of the studies were a lack of information on blinding of assessors, treatment adherence, participant recruitment processes, and imbalanced assignment of the participants to groups. High drop-out rate was a risk factor for bias reported in one study. Lack of blinding of personnel and therapists was commonly found across the studies; however, we acknowledge it is difficult to blind clinicians in manual therapy experimental studies. Most of the studies included predominantly mild to moderate disease stage patients and excluded patients having comorbidities, which limit the generalisation of the results to the wider PD population. Heterogeneity of the quantitative (swallow) measures used across studies also prevented reliable comparison of treatment effects and ability to make stronger conclusions. Only a few studies reported measures of treatment effect size, therefore, by highlighting studies with p < .05, where power was not achieved, improvement reported may not be clinically significant. It is critical that researchers report beyond p values and provide effect sizes in order for us to evaluate treatment efficacy and potential impact on functional performance. Many of the studies’ baseline measures did not show impairments in the evaluated symptoms, which may mute the actual positive effects. Most studies do not demonstrate significant effects on swallow safety as many of the participants included in these studies had mild baseline impairments (low aspiration rates and high FOIS scores) leading to a ceiling effect. In this review, we excluded single case studies and single session studies, perhaps missing some newer therapies without a substantive evidence-base yet. These methodological limitations may affect the strength of recommendations, nevertheless, in health research, we understand well-designed RCTs are difficult to perform. Although PD is a progressive neurological condition, only a few studies reported maintenance effects or long-term follow-up. When developing targeted preventive intervention protocols, researchers must focus on identifying the longitudinal effects in reducing the swallow and respiratory/cough morbidity rates to support evidence-based practice. Future studies should use strong methodology and design to reduce the above-mentioned limitations, where possible, and provide additional substantive evidence for the currently available therapies.

Conclusion

PD is a multidimensional, complex, progressive neurodegenerative disease that provides a challenge in optimising rehabilitation. A diverse range of evidence-based manual therapies have been reported in the literature and evaluated in this review. They show broad positive therapy effects, often involving or extending to more than one body system. The range of therapies identified suggests multiple options are available to improve swallow and respiratory/cough function. Varied modalities were used, such as muscle strength and resistance training exercises, biofeedback-based therapies, electromagnetic stimulation therapies, device-driven treatments, and voice training singing therapies. Combined therapeutic approaches are also supported by this literature, and effective combinations should be considered and evaluated further. Given the progressive nature of PD, future studies should focus on well-designed randomised controlled trials using larger samples sizes that include more severe disease levels, long-term outcomes, and maintenance therapy.

Disclosure statement

No potential conflict of interest was reported by the author(s).