Biofeedback using surface electromyography (sEMG) is a therapeutic technique in which muscle activity is recorded through surface electrodes placed on the skin. This muscle activity is then converted into visual and/or auditory information (Huckabee, 1998). During swallowing rehabilitation, visual biofeedback via EMG recording can be beneficial. Muscles relevant to swallowing and accessible through surface EMG include the suprahyoid muscles. These muscles are among those targeted during the Mendelsohn maneuver, which aims to improve hyolaryngeal elevation during swallowing (Logemann, 1998). However, the added value of biofeedback in the treatment of patients with dysphagia remains inconclusive.

Summary of Findings

Stroke patients

PICO (1.1)

Population: stroke patients

Intervention: Swallowing training program with biofeedback

Comparator: Swallowing training programs without biofeedback (conventional therapy)

_{^1. Risk of bias: serious. Inadequate concealment of allocation during randomization process, Inadequate/lack of blinding of participants and personnel, Baseline imbalances
Imprecision: serious. Low number of patients (<100), Data from only one study, Wide confidence intervals}

_{^2. Risk of bias: serious. Inadequate/lack of blinding of participants and personnel
Imprecision: very serious. Low number of patients (<100), Data from only one study, Wide confidence intervals}

_{^3. Risk of bias: serious. Inadequate concealment of allocation during randomization process, Inadequate/lack of blinding of participants and personnel, Baseline imbalances
Imprecision: very serious. Low number of patients (<100), Data from only one study, Wide confidence intervals}

_{^4. Risk of bias: serious. Inadequate concealment of allocation during randomization process, Inadequate/lack of blinding of participants and personnel, Baseline imbalances
Imprecision: serious. Low number of patients (<100), Data from only one study, Wide confidence intervals}

_{^5. Risk of bias: serious. Inadequate concealment of allocation during randomization process, Inadequate/lack of blinding of participants and personnel, Baseline imbalances
Imprecision: very serious. Low number of patients (<100), Data from only one study, Wide confidence intervals}

_{^6. Risk of bias: serious. Inadequate/lack of blinding of participants and personnel
Imprecision: very serious. Low number of patients (<100), Data from only one study, Wide confidence intervals}

_{* The clinical relevance could not be determined for the study that only reported medians and interquartile ranges for specific outcomes (Alyanak (2025) – FOIS, DHI, PAS).}

_{** The clinical relevance of the outcome ‘presence of enteral nutrition’ (Nordio, 2022) was not assessed, because absence of enteral nutrition is not necessarily equal to returning to a normal diet.}

Patients with Parkinson’s Disease

PICO (1.2)

Population: patients with Parkinson’s Disease

Intervention: Swallowing training program with biofeedback

Comparator: Swallowing training programs without biofeedback (conventional therapy)

_{^1. Risk of bias: serious. Inadequate concealment of allocation during randomization process, Inadequate/lack of blinding of participants and personnel
Imprecision: serious. Low number of patients (<100), Data from only one study, Wide confidence intervals}

_{^2. Risk of bias: serious. Inadequate concealment of allocation during randomization process, Inadequate/lack of blinding of participants and personnel
Imprecision: serious. Low number of patients (<100), Data from only one study}

Head and neck cancer patients

PICO (1.3)

Population: Head and neck cancer patients

Intervention: Swallowing training program with biofeedback

Comparator: Swallowing training programs without biofeedback (conventional therapy)

_{^1. Risk of Bias: very serious. Inadequate selection of participants, confounding, loss to follow up unclear
Imprecision: very serious. Low number of patients (<100), Data from only one study, Wide confidence intervals}

Description of studies

A total of five studies were included in the analysis of the literature. Findings were categorized according to disease type. Important study characteristics and results are summarized in table 2. The assessment of the risk of bias is summarized in the risk of bias tables (under the tab ‘Evidence tabellen’).

Stroke patients

Alyanak (2025) conducted a randomized controlled trial (RCT) to study the effects of game-based electromyography (EMG)-biofeedback therapy on swallowing functions and quality of life in patients with post-stroke dysphagia. Adult patients (>18y old) with a history of hemorrhagic or ischemic stroke lasting more than 3 months, with swallowing complaints after stroke, with a Mini-Mental State Examination (MMSE) score of at least 24, without serious systemic comorbidities, with a Functional Oral Intake Scale (FOIS) score of less than 6, with detection of pathology in the oropharyngeal phase of swallowing during videofluoroscopic evaluation and without treatment related to dysphagia in the last three months, were enrolled in the trial.

A total of 33 participants were randomized into two groups: an intervention group receiving game-based EMG-biofeedback (n=16), and a control group receiving only verbal feedback during clinical swallowing assessments (n=17). In total, the participants received 15 sessions of 30 minutes. The outcomes were the Functional Oral Intake Scale (FOIS), Penetration–Aspiration Scale (PAS), Dysphagia Outcome and Severity Scale (DOSS), and Dysphagia Handicap Index (DHI).

Benfield (2023) conducted a randomized controlled feasibility trial aimed at assessing the feasibility, safety, and preliminary efficacy of swallow strength and skill training using surface electromyographic (sEMG) biofeedback in acute stroke patients with dysphagia.

The study included 27 adult patients who were within 4 weeks post-stroke and had clinically confirmed dysphagia (FOIS<6). Exclusion criteria included a history of previous dysphagia and being medically unstable or having a poor medical prognosis.

Participants were randomly assigned to either usual care (n=15) or usual care plus sEMG-guided swallow training (n=12) delivered in up to 10 sessions over approximately two weeks.

Outcome measures included measures of dysphagia severity, quality of life, functional oral intake, swallow strength and skill, acceptability, tolerability and feasibility.

Nordio (2022) conducted an RCT to study the effect of sEMG biofeedback as an adjunctive treatment for post-stroke dysphagia. Seventeen patients diagnosed with dysphagia following stroke were enrolled. Inclusion criteria were first stroke, onset from stroke > 6 weeks, FOIS<6 and preserved comprehension. Exclusion criteria were hearing or visual impairments, severe concomitant medical conditions and other neurological diseases.

Participants were randomized into two groups: an intervention group receiving sEMG biofeedback therapy alongside standard treatment (n=9), and a control group receiving standard treatment alone (n=7). The program was 5 weeks in total, with 25 sessions of one hour (including recovery periods).

Outcome measures included the Functional Oral Intake Scale (FOIS) as the primary endpoint, and assessments of pharyngeal clearance and swallowing safety using fiberoptic endoscopic evaluation of swallowing (FEES).

Parkinson’s Disease patients

Manor (2013) conducted an RCT to investigate the effectiveness of video-assisted swallowing therapy (VAST) on swallowing function and quality of life in patients with Parkinson’s Disease. The study enrolled 42 non-demented patients with primary Parkinson’s Disease who experienced swallowing difficulties (confirmed via FEES). Exclusion criteria were a history of other uncontrolled neurological or medical disorders interfering with swallowing, brain surgery, dementia, or vision and hearing impairments.

Participants were randomly assigned to either an intervention group receiving VAST (n=21), which involved exposure to general and personalized video recordings of the swallowing process during therapy sessions, or a control group receiving conventional swallowing therapy without video assistance (n=21). Both groups underwent six therapy sessions conducted by the same speech and swallowing therapist.

Outcome measures included swallowing function assessed by FEES, quality of life measured through questionnaires, patient-reported quality of care, and enjoyment of eating evaluated with specific scales.

Head and neck cancer patients

Denk (1997) conducted an observational cohort study to evaluate the effect of videoendoscopic biofeedback on swallowing rehabilitation efficacy in patients after head and neck surgery. The study included 33 patients with prolonged postoperative aspiration following radical resections of malignant tumors of the oropharyngolaryngeal swallowing structures. Nineteen patients received videoendoscopic biofeedback as part of their swallowing therapy, while the 14 patients underwent standard swallowing rehabilitation without biofeedback. The therapy in both groups was 45 mins for 2-5 days/week, up to 6 months in total. Outcome measures were restoration of exclusively oral nutrition and tube status post therapy.

Table 2. Characteristics of included studies

_{*For further details, see risk of bias table in the appendix}

Results

Stroke patients

Three studies evaluated the impact of swallowing training with biofeedback on outcome measures in stroke patients (Alyanak, 2025; Benfield, 2023; Nordio, 2022).

Functional oral intake of food as experienced by the patient (crucial)

Three studies reported the outcome measure functional oral intake of food as experienced by the patient (Alyanak, 2025; Benfield, 2023; Nordio, 2022).

All studies reported this outcome measure with the use of the FOIS (Functional Oral Intake Scale, scale 1-7, higher is better). Results could not be pooled as two studies only reported medians and interquartile ranges.

Alyanak (2025) reported that in total 16 patients received swallowing training with biofeedback (game-based EMG), and 17 patients received conventional swallowing therapy. Baseline FOIS (median [IQR]) was 5.5 [3.5-6] in the biofeedback group versus 5 [5-6] in the control group. Post-intervention FOIS was 6 [5-6] in the biofeedback group versus 6 [5-6] in the control group. The median change was 0 [0-1] versus 0 [0-0]. No statistically significant differences in FOIS score change and post-treatment FOIS scores were reported. Clinical relevance could not be determined (Alyanak, 2025).

Benfield (2023) reported that in total 12 patients received swallowing training with biofeedback (sEMG), and 15 patients received conventional swallowing therapy. Baseline FOIS (median [IQR]) was 3 [1] in the biofeedback group versus 4 [1] in the control group. At 2-week follow-up, FOIS scores were 5 [3] in the biofeedback group versus 4 [2] in the control group. At 90-day follow-up, FOIS scores were 7 [1] versus 7 [2]. The unadjusted mean differences were 0.29 (95% CI: 0.1 to 1.2) at 2-week follow-up, and 0.4 (95% CI: 0.3 to 7.2) at 90-day follow-up, both in favor of biofeedback. These differences were not considered clinically relevant (Benfield, 2023).

Nordio (2022) reported that in total 9 patients received swallowing training with biofeedback (sEMG), and 7 patients received conventional swallowing therapy. Baseline FOIS was 4.14 ± 1.29 in the biofeedback group versus 4.06 ± 1.53 in the control group. At 5-week follow-up, FOIS scores were 5.71 ± 1.38 in the biofeedback group versus 5.56 ± 1.63 in the control group. At 2-month follow-up, FOIS scores were 6.42 ± 0.9 versus 6.08 ± 1.5. At 5-week follow-up, the mean difference was 1.57 (95% CI: 0.51 to 2.63) in the biofeedback group and 1.50 (95% CI: -0.01 to 3.01) in the control group. At 2-month follow-up, the mean difference was 2.28 (95% CI: 1.59 to 2.97) in the biofeedback group and 2.02 (95% CI: 0.63 to 3.41) in the control group.

The mean differences between groups were 0.07 (95% CI: -1.61 to 1.75) and 0.25 (95% CI: -1.15 to 1.67) for 5-weeks and 2-months, respectively, in favor of biofeedback. These differences were not considered clinically relevant (Nordio, 2022).

Swallowing-related quality of life (crucial)

Two studies reported the outcome measure swallowing-related quality of life (Alyanak, 2025; Benfield, 2023). Both studies reported the outcome swallowing-related quality of life with the use of the Dysphagia Handicap Index (DHI, scale 0-100, lower is better). Alyanak (2024) reported medians and interquartile ranges.

Alyanak (2025) reported that in total 16 patients received swallowing training with biofeedback (game-based EMG), and 17 patients received conventional swallowing therapy. Baseline DHI (median [IQR]) was 57 [48-65] in the biofeedback group versus 64 [50-75] in the control group. Post-intervention DHI was 41 [32-50] in the biofeedback group versus 60 [45-73] in the control group. The median change was -12 [-21.5 - -6) versus -4 [-6 - -2]. Statistically significant differences for both DHI score change and post-treatment DHI scores were reported, both in favor of biofeedback. Clinical relevance could not be determined (Alyanak, 2025).

Benfield (2023) reported that in total 12 patients received swallowing training with biofeedback (sEMG), and 15 patients received conventional swallowing therapy. Baseline DHI was 29.6 ± 20.1 in the biofeedback group versus 25.9 ± 16.7 in the control group. At 2-week follow-up, DHI scores were 27.2 ± 21.9 in the biofeedback group versus 18.9 ± 13.7 in the control group. At 90-day follow-up, DHI scores were 24.1 ± 27.4 versus 24.3 ± 25.4. The unadjusted mean differences were 0.23 (95% CI: -6.2 to 22.9) (in favor of control) at 2-week follow-up, and -0.3 (95% CI: -21.2 to 20.7) (in favor of biofeedback) at 90-day follow-up. These differences were not considered clinically relevant (Benfield, 2023).

Functional swallowing ability (important)

The outcome measure functional swallowing ability was not reported.

Number or percentage of patients who return to a normal diet (important)

Two studies reported the outcome measure number or percentage who return to a normal diet (Benfield, 2023; Nordio, 2022).

One study reported this outcome measure by the feeding route (oral diet – normal) (Benfield, 2023). In total, 12 patients received swallowing training with biofeedback (sEMG), and 15 patients received conventional swallowing therapy. After two weeks, 41.7% (5/12 patients) of the biofeedback group was on a normal oral diet, and 13.3% (2/15 patients) of the control group was on a normal oral diet. After 90 days, 81.9% (9/11 patients) of the biofeedback group was on a normal oral diet, and 71.4% (10/14 patients) of the control group was on a normal oral diet. The risk differences were 0.28 (95% CI: -0.043 to 0.611) and 0.10 (95% CI: -0.222 to 0.432), respectively, in favor of biofeedback. The difference after two weeks (28.4%) was considered clinically relevant. The difference after 90 days (10.5%) was not considered clinically relevant (Benfield, 2023).

One study reported this outcome measure by the presence of enteral nutrition (Nordio, 2022). In total, 9 patients received swallowing training with biofeedback (sEMG), and 7 patients received conventional swallowing therapy. After five weeks, 11.1% of the biofeedback group was on enteral nutrition, and 33.33% of the control group was on enteral nutrition. After two months, these percentages were 12.5% versus 20%, respectively. The clinical relevance of this outcome was not assessed, because absence of enteral nutrition is not necessarily equal to returning to a normal diet (Nordio, 2022).

Aspiration (important)

Three studies reported the outcome measure aspiration (Alyanak, 2025; Benfield, 2023; Nordio, 2022). As the studies did not report a total PAS score, the semi-solid PAS score was described (scale 1-8, lower is better). This represents a standardized texture commonly used in dysphagia management, while balancing the challenges of both liquids and solid foods. Results could not be pooled as one study only reported medians and interquartile ranges.

Alyanak (2025) reported that in total 16 patients received swallowing training with biofeedback (game-based EMG), and 17 patients received conventional swallowing therapy. The baseline semi-solid PAS score (median [IQR]) was 4 [2-5] in the biofeedback group versus 3 [2.5-4.5] in the control group. The post-intervention PAS was 2 [1-2] in the biofeedback group versus 2 [2-4] in the control group. The median change was -1.5 [-2.75-0.25] versus 0 [-1-0], which was reported as a statistically significant change. Clinical relevance could not be determined (Alyanak, 2025).

Benfield (2023) reported that in total 12 patients received swallowing training with biofeedback (sEMG), and 15 patients received conventional swallowing therapy. The baseline PAS score (5 ml puree) was 1.4 ± 1.2 in the biofeedback group versus 1.6 ± 0.9 in the control group. At 2-week follow-up, PAS scores were 1.6 ± 1.2 in the biofeedback group versus 2.0 ± 1.9 in the control group. The unadjusted mean difference was -0.13 (95% CI: -1.75 to 0.92) in favor of biofeedback (Benfield, 2023). This difference was not considered clinically relevant (Benfield, 2023).

Nordio (2022) reported that in total 9 patients received swallowing training with biofeedback (sEMG), and 7 patients received conventional swallowing therapy. (Nordio, 2022). Baseline PAS (semi-solid) was 2.07 ± 1.39 in the biofeedback group versus 2.25 ± 1.65 in the control group. At 5-week follow-up, PAS scores were 1.21 ± 0.8 in the biofeedback group versus 1.47 ± 1.25 in the control group. At 2-month follow-up, PAS scores were 1.08 ± 0.29 versus 1.38 ± 1.12. At 5-week follow-up, the mean difference was -0.86 (95% CI: -2.09 to 0.37) in the biofeedback group and -0.78 (95% CI: -2.69 to 1.13) in the control group. At 2-month follow-up, the mean difference was -0.99 (95% CI: -2.07 to 0.09) in the biofeedback group and -0.87 (95% CI: -2.71 to 0.97) in the control group.

The mean differences between groups were -0.08 (95% CI: -2.35 to 2.19) and -0.12 (95% CI: -2.26 to 2.02) for 5-weeks and 2-months, respectively, both in favor of biofeedback. These differences were not considered clinically relevant (Nordio, 2022).

Patients with Parkinson’s Disease

One study evaluated the impact of swallowing training with biofeedback on outcome measures in patients with Parkinson’s Disease (Manor, 2013).

Functional oral intake of food as experienced by the patient (crucial)

The study did not report the outcome measure functional oral intake of food as experienced by the patient (Manor, 2013).

Swallowing-related quality of life (crucial)

The study reported the outcome measure swallowing-related quality of life with the use of three questionnaires (Manor, 2013). In total 21 patients received swallowing training with biofeedback (video-assisted), and 21 patients received conventional swallowing therapy.

The SDQ (Swallowing Disturbances Questionnaire, scale 0.5-44.5, lower is better) was administered at baseline, immediately post-treatment and at 4-week follow-up. At baseline, the SDQ score was 14.65 ± 5.81 in the biofeedback group versus 14.27 ± 7.17 in the control group. Immediately post-treatment, the SDQ score was 12.73 ± 7.6 in the biofeedback group versus 13.43 ± 7.03 in the control group. At 1-month follow-up, the SDQ score was 9.05 ± 5.30 versus 13.08 ± 7.2. Immediately post-treatment, the mean difference was 1.92 (95% CI: -2.43 to 6.27) in the biofeedback group and 0.84 (-3.73 to 5.41) in the control group. At 1-month follow-up, the mean difference was 5.60 (95% CI: 2.02 to 9.18) versus 1.19 (95% CI: -3.44 to 5.82).

The mean differences between groups were 1.08 (95% CI: -5.23 to 7.39) and 4.41 (95% CI: -1.44 to 10.26) immediately post-treatment and at 1-month follow-up, respectively, both in favor of biofeedback. These differences were not considered clinically relevant. However, the between-group difference at 1-month approaches clinical relevance (Manor, 2013).

The SWAL-QOL (scale 0-100, higher is better) was administered at baseline, 4-week follow-up and 6-month follow-up. (Manor, 2013). No total scores for quality of life were reported. However, at 4-week follow-up, statistically significant differences on two of the nine domains (burden and symptoms frequency) were reported in favor of biofeedback. At 6-month follow-up, statistically significant differences on five domains (burden, eating desire, social functioning, mental health and symptoms frequency) were reported in favor of biofeedback. Clinical relevance could not be determined (Manor, 2013).

The SWAL-CARE (Swallowing Quality of Care questionnaire, scale 0-100, higher is better) was administered immediately post-treatment to assess the effectiveness of the advice of the speech and language pathologist. The SWAL-CARE score was 26.26 ± 5.86 in the biofeedback group, and 22.34 ± 5.7 in the control group. The mean difference was 3.92 points (95% CI: 0.42 to 7.42) in favor of biofeedback. This difference was not considered clinically relevant (Manor, 2013).

Functional swallowing ability (important)

The study did not report the outcome measure functional swallowing ability (Manor, 2013).

Number or percentage of patients who return to a normal diet (important)

The study did not report the outcome measure number or percentage who return to a normal diet (Manor, 2013).

Aspiration (important)

The study did not report the outcome measure aspiration (Manor, 2013).

Head and neck cancer patients

One study evaluated the impact of swallowing training with biofeedback on outcome measures in head and neck cancer patients (Denk, 1997).

Functional oral intake of food as experienced by the patient (crucial)

The study did not report the outcome measure functional oral intake of food as experienced by the patient (Denk, 1997).

Swallowing-related quality of life (crucial)

The study did not report the outcome measure swallowing-related quality of life (Denk, 1997).

Functional swallowing ability (important)

The study did not report the outcome measure functional swallowing ability (Denk, 1997).

Number or percentage of patients who return to a normal diet (important)

The study reported the outcome measure restoration of exclusively oral nutrition (with food of all consistencies without moderate or severe aspiration). In total 19 patients received swallowing training with biofeedback, and 14 patients received conventional swallowing therapy. Exclusively oral nutrition was restored in 73.7% of the biofeedback group (14/19 patients) and in 78.6% of the control group (11/14 patients). The risk difference was -0.049 (95% CI: -0.34 to 0.24) in favor of the control. The difference (-4.9%) was not considered clinically relevant (Denk, 1997).

Aspiration (important)

The study did not report the outcome measure aspiration (Denk, 1997).

A systematic review of the literature was performed to answer the following question:

What is the effectiveness of intensive swallowing training programs with or without biofeedback as treatment for patients with dysphagia?

Table 1. PICO

Relevant outcome measures

The guideline panel considered the functional oral intake of food as experienced by the patient and swallowing-related quality of life as a critical outcome measure for decision making; and functional swallowing ability, the number or percentage of patients who return to a normal diet and aspiration as an important outcome measure for decision making. 

A priori, the guideline panel did not define the outcome measures listed above but used the definitions used in the studies.

The guideline panel used the GRADE standard limits of 25% (a relative risk (RR) of <0.80 and >1.25) for dichotomous outcomes and 10% for continuous outcomes as a minimal clinically (patient) important difference. For dichotomous outcomes with very few events, a risk difference of 25% was used as a minimal clinically (patient) important difference.

Search and select (Methods)

A systematic literature search was performed by a medical information specialist using the following bibliographic databases: Embase.com and Ovid/Medline. Both databases were searched from 2015 to April 3^rd, 2025 for systematic reviews, RCTs and observational studies. Systematic searches were completed using a combination of controlled vocabulary/subject headings (e.g., Emtree-terms, MeSH) wherever they were available and natural language keywords. The overall search strategy was derived from four primary search concepts: (1) dysphagia; (2) biofeedback; (3) electromyography; and (4) modified barium swallow. Duplicates were removed using EndNote software. After deduplication a total of 2380 records were imported for title/abstract screening. Studies were selected based on the criteria described in Table 1. Titles and abstracts were screened using the ASReview software version 1.6.2. The settings TF-IDF and Naïve bayes were used. One study (Martin-Harris, 2015) was used as prior knowledge for inclusion and one study (Moghadam-Kia, 2017) was used as prior knowledge for exclusions.

The first 10% of hits were screened by the guideline development group and the guideline methodologist. The remaining articles were subsequently screened by the guideline methodologist, using the following stopping rule: stop after 200 subsequent exclusions.

Initially, 21 studies were selected based on title and abstract screening, and 292 studies were excluded. After reading the full-text articles, a total of five studies were included, and 16 studies were excluded (see the exclusion table under the tab ‘Evidence tabellen’).

Among these five studies were three RCTs (Alyanak, 2025; Benfield, 2023; Nordio, 2022) and two systematic reviews (SR) (Battel, 2021; Benfield, 2019). From the SR of Battel, one relevant study was included (Manor, 2013). From the SR of Benfield, one relevant study was included (Denk, 1997). The studies by Li (2016) and McCullough (2012 and 2013) from the latter SR were not included, as they did not have a control group in a comparative study design. Moreover, the study by Huimin (2015) from the SR by Benfield was not included, as no outcome measures relevant for this module were reported and the full-text of the article is in Chinese. This resulted in a total of five studies to be included in the analysis of the literature.

The included studies are described separately per disease category in the results section. These categories comprise stroke patients, patients with Parkinson’s Disease, and patients with head and neck cancer (HNC).