Introduction

Dysphagia is a common and significant symptom following stroke. Dysphagia, or swallowing difficulties, affect a third to over two thirds of patients after stroke [1, 2]. Dysphagia causes medical complications, including increased hospitalisation, morbidity, and risk of aspiration pneumonia [3]. It is associated with poor psychosocial health outcomes, such as reduced nutrition, hydration and quality of life [4]. Patients with dysphagia have longer lengths of hospital stay and higher healthcare costs [5, 6].

Current management of dysphagia involves compensation and rehabilitation. Compensatory techniques—such as chin tuck or modifying diet and fluid consistencies—enable safe swallowing but do not alter long-term function [7]. Rehabilitative exercises, however, can improve swallowing function and resumption of oral intake or normal food and drink [7]. Rehabilitative exercises can be indirect (motor without swallow) or direct (motor with swallow) [8]. Indirect exercises aim to strengthen muscles involved in swallowing, and include the Shaker head-lift and tongue strengthening exercises [9]. Direct exercises involve the action of swallowing and include the Mendelsohn manoeuvre and effortful swallow [9]. Studies have shown the positive effects of rehabilitative exercises on reducing the severity and symptoms of post-stroke dysphagia [10].

However, there is limited understanding of the optimal way to conduct swallowing exercises, particularly, the optimal dosages of swallowing exercises. Dosage is an important factor which can impact on intervention efficacy and efficiency [11]. According to the American College of Sports Medicine’s (ACSM) FITT framework, dosage consists of Frequency (how often), Intensity (how hard), Time (how long) and Type (what kind) of exercise [12]. Altering or increasing these components of dosage can optimise exercise or intervention outcomes, as seen in sports medicine and stroke rehabilitation [13, 14]. However, there is limited knowledge on what dosages to use for swallowing exercises. Previous scoping and literature reviews have highlighted the paucity of data regarding dosage recommendations in dysphagia intervention [9, 15]. This is reflected in surveys of speech pathologists which indicate variability in the exercises and dosages used to treat dysphagia [16].

The aim of this systematic review was to investigate what dosages of swallowing exercises are reported in studies in post-stroke dysphagia. To our knowledge, only one scoping review has specifically examined the dosages of swallowing exercises [15]. Our review was conducted to systematically update the search with new studies. Given that intervention dosage may vary depending on diagnosis, this review focused on one of the most common causes of acquired dysphagia: stroke [17]. This systematic review aimed to investigate dosage reporting in research and describe current swallowing exercise dosages in intervention studies to guide clinicians when considering dosage prescription. The findings can be used to identify areas for future research in optimising dosage of swallowing exercises to facilitate more cost-effective intervention, increased patient engagement and improved outcomes.

Methods

This systematic review was conducted according to PRISMA guidelines [18]. Prior to conducting the study, a protocol was registered on PROSPERO (https://www.crd.york.ac.uk/PROSPERO/, registration number: 131294).

Eligibility criteria

Studies were included if: (i) they included adult participant/s with dysphagia due to stroke, (ii) they examined rehabilitative, behavioural swallowing exercises, (iii) they were a published intervention study, where pre/post outcomes were reported, and (iv) they provided a detailed description of the dosage of the rehabilitative exercise. Only full-text published studies accessible through online databases were included in this review. For the purpose of this review, the ACSM FITT framework was used to specify the minimal elements required in a detailed description of dosage [12]. Dosage description needed to include the frequency (or number) of sessions, intensity or dose (at a minimum, the number of repetitions of each exercise), intervention duration and type of exercise. Stating the type of exercise required the name and reference of a well-known exercise, or a detailed description of the materials, procedures, activities and/or processes involved in the exercise (as per points 1–4 of the Template for Intervention Description and Replication checklist) [19]. Studies were included even if only a subset of participants matched inclusion criteria (e.g., participants with dysphagia due to stroke and other conditions). This review focused on methodology reporting, not intervention effect, so type of outcome measure was not an inclusion criterion. Outcome measures were collected as reported, without limit. Studies in all languages were included. Studies were excluded if they only applied passive interventions, such as acupuncture, thermal–tactile stimulation, compensatory strategies, or electrical stimulation (i.e., when not combined with active exercise), as these require different dosages to behavioural exercises and were not the focus of this review.

Search strategy and selection process

A comprehensive search of studies was conducted from inception until 10 February 2022 using the electronic databases: MEDLINE, Embase via Ovid, CINAHL, Web of Science and SpeechBITE. The Medical Subject Heading terms: “Deglutition, Deglutition Disorders, Pharynx OR Pharyngeal Muscles” were combined with “Stroke OR Cerebrovascular Disorders”; and “Exercise, Exercise Therapy, Neurological Rehabilitation, Stroke Rehabilitation OR Rehabilitation” along with free key word searches of specific swallowing exercises. The search strategy was developed in conjunction with a university librarian using candidate search terms from two relevant studies. See Online Appendices 1–3 for full search strategies for each database. The reference lists of included studies were hand searched to identify further studies.

Covidence software was used to remove duplicates and double checked by the lead author [20]. Two reviewers independently screened titles, abstracts, and eligible full text articles against inclusion criteria using Covidence software. Conflicts were resolved through discussion with a third reviewer. Abstracts or full text studies which were not in English were translated by bilingual speakers. All members of the research term agreed on the final studies included for review.

Data collection

Data was extracted from included studies by the lead author using an Excel form and checked by a second reviewer. The following data was extracted:

  • Study author, year, and source of publication.

  • Participant demographics (sample size, age, sex, inclusion/exclusion criteria, length of time since stroke) and participant diagnoses (cause of dysphagia, stroke type and severity).

  • Setting and study design (including study aims and intervention groups).

  • Dosage of swallowing exercises (type of exercises, any reported intensity, frequency of sets/sessions and duration).

  • All outcomes pre and post intervention (excluding follow-up timepoints).

Risk of bias assessment

Included studies were assessed for quality by two independent reviewers using the American Speech–Language–Hearing Association (ASHA) Levels of Evidence framework on an online spreadsheet tool [21, 22]. The framework involved rating studies against eight quality markers: blinding of assessors, random sampling/allocation, group/participant comparability, treatment fidelity, validity and reliability of an outcome measure, whether significance was reported, precision of effect size and/or confidence interval and analysis by intent-to-treat. Each quality marker contributed to one point in an overall quality score. A quality score of 7–8 was considered high quality, 5–6 good quality and ≤ 4 low quality [21]. Conflicts were resolved through discussion with a third reviewer.

Summary measures

Information about participants, swallowing exercises, reported dosages and outcome measures was collated into two summary tables. Tables were organised alphabetically by exercise type to allow comparison of swallowing exercise dosages. When summary data was missing or in a different form, means and standard deviations were estimated using Hozo et al. or Wan et al.’s methods [23, 24]. Effect sizes (Hedges’ g) were calculated for pre–post changes in continuous data for swallowing intervention groups using an online effect size calculator [25]. Hedges’ g effect sizes can be interpreted as 0.2 = small effect, 0.5 = medium effect, 0.8 = large effect [26]. Meta-analysis could not be conducted due to heterogeneity of study designs, interventions, dosages, and outcome measures, and was not necessary to address study objectives.

Results

The initial search resulted in 7263 studies. After duplicates were removed, 4835 studies were screened for inclusion. Of these, 54 studies passed full text review (Fig. 1).

Fig. 1
figure 1

PRISMA flow diagram

Full size image

Participant characteristics

Within the 54 included studies were a total of 1501 participants with an average age of 65.8 years. Study sample sizes ranged from one [27,28,29,30] to 90 [31] participants. Seven studies had mixed caseloads, including patients with dysphagia due to cancer, brain injury, degenerative and/or cardiac conditions [32,33,34,35,36,37,38]. The remaining studies included participants with stroke-related dysphagia only, with a variety of stroke types, locations, and severities. Eleven studies included ischemic stroke only [30, 31, 37, 39,40,41,42,43,44,45,46] and sixteen studies included both ischemic and haemorrhagic stroke [47,48,49,50,51,52,53,54,55, 57,58,59,60, 71, 72, 81]. Nine studies examined supratentorial strokes [41, 42, 46,47,48, 57, 58, 61, 62], six studies examined infratentorial strokes [27,28,29,30, 56, 63] and twelve studies included both supratentorial and infratentorial strokes [40, 45, 51, 59, 60, 64,65,66,67,68,69, 81]. Seven studies did not report stroke type or location. Only two studies reported stroke severity scores [30, 37].

The length of time between stroke and commencement of intervention was reported in 50 studies. Twenty-five studies were conducted within 6 months after stroke, 16 conducted more than 6 months after stroke and nine studies included participants across a range of time periods post-stroke. See Tables 1 and 2 for sample sizes, participant ages, and length of time between stroke and onset of intervention as reported in each study.

Table 1 Single swallowing interventions in included studies (including participants, exercise descriptions and dosages, outcome measures and pre–post Hedges g effect sizes in exercise-based intervention group) grouped by exercise type
Full size table
Table 2 Combined intervention programs in included studies (including participants, exercise descriptions and dosages, outcome measures and pre–post Hedges g effect sizes in exercise-based intervention group)
Full size table

Study characteristics

There were 28 randomised controlled trials, eight non-randomised controlled trials, three retrospective case controls, twelve pre/post case series and three case studies. Studies were published from 2002 [38] to 2022 [46]. See Online Appendix 4 for study designs.

Dosages of swallowing exercises

Exercise type

Fourteen different swallowing exercises and twelve different swallowing programs were described in the included studies. Eleven studies investigated indirect oral exercises, including lip exercises, tongue exercises and an orofacial exercise program. Twenty-eight studies investigated indirect pharyngeal exercises, including Shaker head lift, expiratory muscle strength training and chin tuck against resistance. Ten studies investigated direct swallowing exercises, most commonly the Mendelsohn manoeuvre and effortful swallow. Twelve studies examined a combined swallowing program. Some studies examined more than one intervention. The most reported exercise was Shaker head lift (investigated in ten studies). See Table 1 for single swallowing exercises in included studies (including swallowing exercises, dosages, outcome measures and effect sizes), Table 2 for combined swallowing programs and Table 3 for definitions of common exercises.

Table 3 Descriptions of common swallowing exercises
Full size table

In general, exercise type was well-reported but varied between studies. Studies named specific exercises or provided detailed descriptions of how to conduct exercises (as per inclusion criteria). Consistent descriptions were mostly used when replicating the same exercises. For example, studies examining Shaker head lift all involved patients lifting their heads to look at their feet while in supine position. However, there was variation in devices and variation between combined exercise programs. In studies investigating single exercises, such as chin tuck against resistance, jaw opening and lip training, the main variations were in the use of devices. For example, the jaw opening exercise was described with four different types of resistance: none, against a trainer’s hand, against a jaw opening device, or against a ball. In studies investigating combined intervention programs, there was variation in which exercises were included. While all combined programs included orofacial exercises, they varied in whether they included pharyngeal exercises, swallowing with or without real boluses and breathing exercises.

Frequency

Frequency of intervention was well-reported but varied between studies. Frequency was consistently described by the number of sets per day and/or number of days per week of intervention. Most studies conducted intervention five times per week (30 out of 54 studies). However, frequency ranged from one to ten sets per day, and from one time per fortnight to seven times per week [31, 36].

Intensity

Studies varied in intensity and in how they reported intensity. All studies reported the number of repetitions of each exercise (as per inclusion criteria). Nine studies reported both number of repetitions and length of sessions [28, 34, 39,40,41,42, 70, 80, 88]. Total exercise repetitions ranged from 3 to 411 repetitions per session [48, 57, 75, 89] and from 7 to 500 repetitions per day [50, 60].

Seventeen studies reported intensity as a percentage of one-repetition maximum (with one-repetition maximum [1-RM] defined as the maximum resistance that can be applied to one exercise repetition through full range of motion) [12]. Indirect strengthening or accuracy exercises using devices—most commonly expiratory muscle strength training and tongue strengthening exercises—were more likely to report a specific intensity level. All expiratory muscle strength training studies specified exercise intensity, but targets ranged from 30% to > 70% of 1-RM. Six tongue strengthening exercises specified a target intensity of either 60% or 80% of 1-RM for strengthening and between 25% and 85% of 1-RM for accuracy training. For chin tuck against resistance, game-based biofeedback, jaw opening, Mendelsohn and Intensive Dysphagia Rehabilitation, one study per intervention specified a target intensity, e.g., 70% of 1-RM with a device.

Intensity of swallowing exercises was otherwise reported and measured in different ways. Many studies used qualitative descriptions of effort (e.g., “as hard as possible”, “to maximum extent”) [39, 73]. Biofeedback (e.g., surface electromyography) with set targets was used in four studies to set specific intensity levels [28, 66, 72, 78]. External resistance was used in around 12 studies with varying forms of resistance depending on the exercise type. Intensity or task difficulty was increased in some studies through increasing length of holds, number of repetitions or changing the amount or substance being swallowed [33, 36, 57]. See Table 4 for different ways swallowing exercise intensity was reported in studies.

Table 4 Ways that intensity of swallowing exercises was reported or varied in studies
Full size table

Time

Time, or duration, of intervention was reported as the number of days, weeks, or months of intervention. Most studies had an intervention duration of 4 weeks (22 studies), or 6 weeks (12 studies). However, across all studies, duration ranged from 6 days to 12 weeks [41, 42, 69, 70]. Three studies reported duration based on performance (e.g., once patients were no longer tube feeding) [36, 43, 63]. See Tables 1 and 2 for detailed exercise dosages reported in included studies.

Standard care

Around half (27) of included studies provided some form of standard care additional to their experimental intervention, either given to a control group for comparison or used as a co-intervention in both groups. Synonymous terms were used to describe this baseline intervention, including “conventional dysphagia therapy”, “traditional dysphagia therapy”, “regular” or “routine” training. While all included studies described the dosage of the experimental intervention in detail, the same level of detail was not used when describing the dosage of interventions in standard care. Most studies used general terms (e.g., orofacial muscle exercises, therapeutic manoeuvres) to describe what was involved in standard care rather than naming specific exercises. Most studies stated the length of intervention time provided in standard care groups but not the number of exercise repetitions and no other measures of intensity. Standard care was typically provided for 30 min, 5 days per week for 4 weeks. However overall, in studies that described exercises, there was variation in which exercises were included in “standard care” and in their dosages. See Table 5 for details on interventions and dosages used in “standard care” groups.

Table 5 Interventions and dosages of “standard care” (SC) for dysphagia used in control groups or as co-interventions
Full size table

Reported rationales for dosage prescription

Only five studies specifically described an evidence-based rationale for their dosage. The recommended dosage for strength training drawn from limb rehabilitation research (i.e., ten repetitions, three times per day, 3 days per week for 8 weeks) [90] was applied to tongue strengthening exercises [27, 68] and effortful swallowing [62]. Two studies provided a high dosage of intervention, derived from principles of neural plasticity and/or exercise physiology [33, 78]. One crossover study investigating a high-intensity program with Mendelsohn manoeuvre showed improved outcomes on treatment weeks compared to non-treatment weeks, and with 2 weeks of intervention compared to 1 week [78]. No other studies in this review specifically compared different dosages of the same intervention.

Outcome measures

There was a wide range of different outcome measures used. Within 54 studies, 52 different outcome measures were used. The most commonly reported outcome measure was the Penetration–Aspiration Scale [91] (used in 23 studies) which rates depth and severity at which food or drink is aspirated into the airway. The next most commonly used measures were the Functional Oral Intake Scale [92] (12 studies) which rates level of oral intake, hyoid displacement/elevation (ten studies) and maximal tongue strength (nine studies). See Tables 1 and 2 for outcome measures used in included studies.

Risk of bias

There were 12 studies of high quality, 25 good quality and 17 low quality. Studies generally performed well on having similar groups and participants, using a valid and reasonable outcome measure, and reporting significance, effect size and confidence interval. The quality markers that were least often observed were blinding of assessors, randomised sampling, treatment fidelity and intention-to-treat analysis. See Online Appendix 4 for quality ratings for each study.

Discussion

This systematic review found that overall, the dosages of swallowing exercises in post-stroke dysphagia studies were poorly reported, and when reported, varied significantly. Most post-stroke dysphagia studies were excluded from review due to under-reporting of exercise dosage, particularly intensity. While swallowing intervention was often provided five times per week for 4 weeks, there was a wide breadth of different exercises, frequencies, durations and intensities of intervention, with a range of different study designs and outcome measures. Variation between studies may have been due to arbitrary selection of intervention dosage, with few studies reporting a rationale for their dosage. Due to the heterogeneity of results, it was difficult to determine optimal dosages of swallowing exercises. Despite a growing awareness of the importance of dosage, more work is needed to improve consistency of dosage reporting and identify evidence-based principles to optimise prescription of swallowing exercise dosages.

To investigate dosages of swallowing exercises, we needed to identify if dosages were reported in studies. One hundred and forty-six studies were excluded due to not providing a detailed description of swallowing exercises. Only 27% of the 200 studies which matched the first three inclusion criteria (i.e., rehabilitative intervention studies in post-stroke dysphagia) reported frequency, intensity, time, or type of intervention in detail, and were included for review. Even within included studies, most studies only described the dosage of the experimental intervention in detail but not the dosage of the non-experimental or control intervention. These findings reveal an under-reporting of dosage in post-stroke dysphagia studies. Poor reporting of interventions and dosages prevents reliable implementation, comparison and replication of interventions [19]. Evidence suggests that this is an issue that extends beyond dysphagia research [93, 94]. The disproportionate under-reporting of control group interventions has also been highlighted in stroke rehabilitation studies [95]. While checklists exist to improve reporting of interventions [19, 96], specific guidelines around comprehensive dosage reporting would further improve study reporting.

Better reporting and measurement of intensity of swallowing exercises is needed. Most studies excluded for not providing a detailed description of dosage did not report intensity. Within included studies with dosage descriptions, exercise type, frequency and duration were relatively well-reported compared to intensity. Intensity is defined as how hard or how much effort an exercise involves [12]. The methods used by studies in this review may provide a starting point when considering how to measure, change and report on swallowing exercise intensity (see Table 4). Number of exercise repetitions provides some indication of intensity but does not describe amount of effort [97]. Providing both exercise repetitions and session length (as reported in nine studies) allows calculation of dosage rate, which contributes to intensity [98]. Qualitative descriptors (e.g., “as hard as you can”) or patient rating scales [99] can indicate subjective level of effort. Exercises using devices can set specific intensity levels as percentages of 1-RM, in a similar format to limb training. However, there was still significant variation in how intensity was measured, and many swallowing exercises did not have routine ways to set or measure intensity, such as a Masako manoeuvre, or head lift exercise (see Table 3 for explanations of these exercises). To properly quantify dosages, further work is needed to identify consistent methods to set and report exercise intensity for the wide range of swallowing exercises.

Examining studies that did report dosage showed significant variation in dosages of swallowing exercises in both experimental and standard care interventions. A wide range of different exercises were used, and there was variation in dosage across different swallowing exercises, similar to findings in a previous review [15]. This variation was also observed in standard care, or “conventional” or “traditional dysphagia therapy”. While a similar repertoire of exercises was used, the combination of exercises and dosages involved in standard care varied between studies. This variation in exercise selection and dosage is reflected in surveys of speech pathologists, which show no true “standard care” in dysphagia management [16]. While some variation is inherent to dysphagia intervention, using different forms of standard care in studies impacts on the ability to compare and determine relative intervention efficacies [96, 100]. Consistency in “standard care” is needed for a stable baseline in research. Evidence-based guidelines and rationales for setting intervention dosages are also needed to reduce unwarranted variation in clinical care.

Studies in this review appeared to have different rationales for dosage prescription. Only five studies specifically provided a rationale for their dosage prescription, drawn from strength-training in limb exercises and neural plasticity principles. Most commonly, studies appeared to replicate dosages from previous studies. This was seen in the Shaker exercise, where seven out of nine studies used the same 30 isotonic and three 60-s isometric head lifts used in the pioneering article [38]. Convenience also appeared to be a contributing factor. Most studies provided intervention five times per week, likely catering to typical working days rather than rehabilitation need. Overall, there was lack of evidence-based rationales for dosage prescription, similar to findings in stroke rehabilitation literature [101]. Transparent reporting of the rationale for dosage prescription in studies could facilitate better consideration of dosage selection and allow readers to understand the reasoning of researchers. The use of strength-training and neural plasticity principles offer some direction when considering rationales for swallowing exercise dosages. These principles include specificity (targeting swallow-specific exercises), increasing the volume of intervention and introducing resistive loading to swallowing exercises [102,103,104].

This study had several strengths and weaknesses. The strengths included the systematic approach following PRISMA guidelines. The review followed ethical guidelines, including pre-registering a protocol. The comprehensive search strategy, and inclusion of studies without limitations on language or study design, generated a high number of studies. Study selection included blinded screening and quality assessment of studies by two independent authors to reduce bias. Further, findings included measures of effect size despite the heterogeneity of results. The limitations in this systematic review were related to the quality and heterogeneity of studies. To capture all swallowing exercises and dosages used in stroke rehabilitation, all study designs and quality ratings were included and there was no specification of outcome measures. This allowed a more thorough investigation of the topic, but may have introduced biases [105]. Using a less well-known quality appraisal tool to cater for various study designs may have also impacted on quality assessment. Finally, the focus on intervention methodology rather than outcome precluded in-depth statistical analysis or data synthesis.

Conclusions

Dosage is important in exercise-based intervention. There is increasing awareness of the importance of intentional dosage prescription and reporting in research. However, this review indicates that further work is needed to improve consistent dosage reporting and evidence-based dosage prescription in post-stroke dysphagia studies.

Uniform terminology and frameworks are needed to improve consistent and comprehensive dosage reporting across the field. Current frameworks can be used when prescribing and reporting on dosage [11, 12, 15, 106]. Methods used by studies in this review could help guide setting and reporting of intensity in clinical practice. Given the variable reporting of exercise types and dosages, clinicians should pay careful attention to the descriptions of exercises and dosages in studies when replicating or evaluating new interventions. More consistent dosage reporting within studies will improve quality and useability of studies and facilitate reproducibility, comparison, and synthesis of research.

Further work is also needed to improve evidence-based dosage prescription. Current evidence (such as strength training and neural plasticity principles) can be considered along with clinical reasoning to guide dosage prescription. However, more research is needed to examine which principles are applicable to dysphagia rehabilitation. Future research could investigate the impact of altering different components of dosage, such as comparing similar interventions provided at different dosages. Improved dosage reporting, and evidence-based dosage prescription has the potential to improve intervention efficacy and outcomes for patients with post-stroke dysphagia.