Preventing radiation‐induced dysphagia and trismus in head and neck cancer—A randomized controlled trial

Abstract

Background

Radiation‐induced dysphagia and restricted mouth opening are common problems among patients with head and neck cancer. The aim of the present randomized controlled trial was to determine if an exercise protocol could prevent swallowing and mouth opening impairment.

Methods

Eighty‐nine participants were randomly assigned to either an active group performing preventive swallowing and mouth opening exercises (n = 45) or to a control group (n = 44). Outcome measures were collected at baseline before radiotherapy and approximately 1‐month post‐treatment. Primary endpoints were changes in swallowing function according to the Penetration Aspiration Scale and mouth opening ability measured in millimeters. Intention‐to‐treat analysis was used.

Results

Swallowing function and mouth opening deteriorated in both groups, with no statistically significant positive effect of the protocol detected at follow‐up. Among patients who completed >75% of exercises, there was a trend toward better outcomes.

Conclusions

Preventive exercises did not improve short‐term swallowing function and mouth opening after radiotherapy.

1. INTRODUCTION

The incidence of head and neck cancer (HNC) has increased in the Western world in recent years mainly as a result of the aging population and an increase in cancers caused by human papillomavirus (HPV). Globally, the survival rate after HNC treatment is around 50% depending on the size and location of the tumor. ¹ One of the primary curative treatment options for HNC is radiotherapy either alone or in combination with chemotherapy or surgery. ² Unfortunately, radiation can cause both acute and long‐term side effects with negative impacts on the health and quality of life of patients. ³

Two common side effects of HNC treatment are dysphagia and trismus, that is, swallowing difficulties and restricted mouth opening. Both ailments affect the ability to eat and drink, leaving patients at risk of malnutrition, dehydration, and aspiration pneumonia. ⁴ , ⁵ During treatment, acute reactions in the irradiated tissues can cause mucositis, pain, and edema. These reactions often restrict oral intake to mainly nutritional drinks and lead to partial or full feeding tube dependency. Moreover, reduced oral intake is considered a risk factor for atrophy of muscles involved in swallowing and mouth opening and can, along with fibrosis and cranial neuropathy, cause chronic impairment. ⁶ Persistent swallowing difficulties and trismus affect around 40%–70% and 30% of patients, respectively, 2 years after chemoradiation. ⁴ , ⁷ , ⁸

There is a knowledge gap regarding exercises aimed at reducing swallowing difficulties and trismus in HNC patients, including the most beneficial type of exercise, whether intervention should be offered preventatively or after the onset of impairment, the optimal amount of exercise required to achieve beneficial effects, and strategies to enhance adherence to exercise regimens. ⁵ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ Exercises to improve swallowing function usually focus on improving strength and range of motion in the muscles and other structures involved in swallowing. ⁵ Similarly, interventions for restricted mouth opening frequently involve passive and active stretching with or without a jaw‐opening device. ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸

Structured exercise with a mouth opening device has been found to improve mouth opening ability, reduce trismus‐related symptoms, and augment health‐related quality of life (HRQL) both in the short‐ and long‐term for patients with established trismus after radiotherapy. ¹³ , ¹⁷ However, results from randomized controlled trials (RCTs) evaluating the effect of preventive exercises for restricted mouth opening are not consistent. ¹⁶ , ¹⁹ , ²⁰ , ²¹ , ²² The available research concerning swallowing function suggests better outcomes of preventive swallowing exercises and of encouraging patients to maintain oral intake with as few modifications as possible during radiotherapy. ⁶ However, due to shortcomings of existing studies, such as small populations, retrospective designs, and the absence of control groups, more research is needed to fully support these notions. ⁵ , ⁹ , ¹⁰ , ¹¹

Poor adherence to exercise protocols among patients suffering from acute reactions to irradiation is a challenge in interventional studies. Preventive protocols are typically comprehensive and consist of several different exercises; adherence to such protocols has generally been low. ²³ , ²⁴ Therefore, there is a need for a simple and convenient, but effective intervention protocol targeting both swallowing function and mouth opening. The aim of this RCT was to determine if a simplified exercise protocol could prevent radiation‐induced swallowing and mouth opening impairment in HNC patients.

2. MATERIALS AND METHODS

2.1. Study population

HNC patients who were offered radiotherapy with curative intent, with or without chemotherapy for a tumor located in the oropharynx, hypopharynx, and larynx were approached for inclusion in the study (n = 355) at Sahlgrenska University Hospital in Sweden 2019–2022. Patients who were planned to undergo surgical treatment for the tumor were not eligible for inclusion. Participants were excluded if they had a previous history of HNC, had trismus at the time of diagnosis, were edentulous, had a tracheostomy, neurological or neuromuscular disease, or had dysphagia unrelated to the current cancer diagnosis. Ninety‐one patients fulfilled the inclusion criteria and agreed to participate. The participants were randomized into either active treatment (intervention group, n = 45) or standard treatment (control group, n = 46). The flowchart in Figure 1 illustrates enrollment, allocation, and drop‐outs in the study.

FIGURE 1.

Flowchart over study participation and analysis. HNC, head and neck cancer. [Color figure can be viewed at wileyonlinelibrary.com]

2.2. Oncological treatment

All participants received External Beam Radiation Therapy (EBRT) delivered by volumetric modulated arc therapy (VMAT) according to the regional cancer treatment program. The target dose for the primary tumor was typically 68 Gy. The treatment period was 5.5 weeks and followed a moderately accelerated fractionation schedule, that is, 2 Gy delivered 1–2 times daily and 6 treatments per week. Seventy‐one participants (83%) also received chemotherapy (cisplatin) concomitant to radiotherapy.

2.3. Study design

Randomized allocation was stratified by age, sex, HPV status, tumor localization, tumor stage, and level of comorbidity according to the Adult Comorbidity Evaluation‐27 (ACE‐27) index, ²⁵ and mouth opening and swallowing ability.

Measurements of swallowing function, mouth opening, and saliva production were made before radiotherapy (baseline) and at follow‐up approximately 1 month after the completion of radiotherapy by a speech‐language pathologist (SLP). At these time points, the SLPs also collected information about sociodemographic characteristics, nutrition, and weight, and distributed a questionnaire concerning HRQL after treatment for HNC.

A one‐point difference in the Penetration Aspiration Scale (PAS) score was considered a clinically relevant change in the follow‐up assessment between the two groups assuming a standard deviation (SD) of 1.3 and a 25% dropout rate. The included sample size was determined to be 40 participants in each group for a predicted power of 80% (Mann–Whitney U test, alpha = 0.05).

Considering the change in interincisal mouth opening (MIO) before and after treatment, a difference of 5 mm was expected with an SD of 5.8 and a 25% dropout rate. Thirty‐one participants were needed per treatment group to achieve 80% power (Mann–Whitney U test, alpha = 0.05).

2.4. Intervention

SLPs instructed participants on how to perform the preventive exercises that targeted swallowing and mouth opening and distributed written and video instructions. Participants were instructed to exercise daily and to start directly after inclusion, approximately 1–2 weeks before the start of radiotherapy, and continue the daily exercise until the follow‐up.

The required dosage for exercise‐based dysphagia intervention is unclear. ²⁶ Existing recommendations for achieving hypertrophy are based on effects on limb muscles in healthy individuals. ²⁷ , ²⁸ One study on healthy men found that 1 set 3 times weekly was enough to accomplish hypertrophy in limb musculature compared to a control group. ²⁹ Therefore, it was hypothesized that 1 set of 10 repetitions of the tongue hold (also called Masako's maneuver), 7 days per week would achieve at least enough hypertrophy to maintain the swallowing strength and ensure adherence.

The tongue hold exercise is performed by anchoring the tip of the tongue between the front teeth during repeated saliva swallows and is designed to improve contact between the posterior pharyngeal wall and the base of the tongue. ³⁰ , ³¹ The tongue hold exercise is hypothesized to decrease problems of residue buildup in the pharynx, ³¹ which is a common complaint among HNC survivors. Moreover, residue accumulating in the pharynx is linked to spillage into the airway, that is, aspiration. The tongue hold exercise is frequently included as one of several exercises in preventive protocols targeting swallowing ability in HNC patients. ³² We chose to include it as the single swallowing exercise in the present study because it meets two important principles of training: it is task‐specific (swallowing), and it provides overload by forcing muscles in the pharyngeal wall and base of the tongue to work extra hard. ³³

In addition, participants in the intervention group performed passive and active jaw exercises with a JawTrainer©. ³⁴ To perform the passive exercise, each participant was instructed to place the trainer between their front teeth and press down until they felt tension but no discomfort in the jaw. The stretch was maintained for 30 s and repeated 3 times with a short pause in between. To perform the active exercise, each participant was instructed to open the trainer to a width of 1.5 cm and bite down on the clasp for a few seconds. This motion was repeated 5 times with a pause between repetitions.

Patients reported daily if they had performed the exercise or not in an exercise diary. Reports were made separately for each exercise. It was possible to report reasons for not doing a particular exercise in a free‐text column of the diary.

During treatment and up to 4 weeks post‐radiotherapy, both the intervention group and the control group had weekly contact (usually by telephone) with the SLPs and were encouraged to eat and drink as much as possible by mouth, even just a small amount, to prevent long‐term inactivity of the swallowing musculature. All participants were to report daily in their diary whether they had something to eat or drink. At the weekly contact, the SLPs supported the participants in the intervention group in performing the exercise.

3. ASSESSMENT AND ENDPOINTS

3.1. Instrumental evaluation of swallowing function

Flexible endoscopic evaluation of swallowing (FEES) was performed at baseline and the follow‐up by experienced SLPs. Participants were offered liquid and solid consistencies in accordance with the International Dysphagia Diet Standardization Initiative (IDDSI): mildly thick, 3, 5, and 10 mL; thin liquid, 3, 10, and 20 mL; pureed, 5 and 10 mL; easy to chew biscuit ¼ (i.e., IDDSI levels 2, 0, 4, and 7, respectively). ³⁵ All boluses were colored with green food coloring and administered by spoon or cup. Swallowing was done on command for liquid boluses except the 20 mL of thin liquid where the instruction was to drink freely. If a bolus was deemed to pose a risk of gross aspiration, it could be excluded from the protocol.

Blinded analysis of FEES films was performed by two SLPs with more than 10 years of FEES experience, who had no other involvement in the study. Both SLPs were accustomed to using the PAS. To allow for analysis of intra‐rater reliability, 25% of the films were duplicated. PAS was used as a primary endpoint for swallowing function ³⁶ as it is validated and widely used in research studies where swallowing is evaluated by instrumental assessment. ³⁷ The scale is ordinal and considers how far the bolus material enters the airway and the patient's sensory response. PAS scores range from 1 (material does not enter the airway) to 8 (material enters the airway, passes below the vocal folds, and no effort is made to eject). Max PAS rating during the FEES was used as representation of the swallowing function.

Swallowing function was also evaluated by comparing Dynamic Imaging Grade of Swallowing Toxicity for Flexible Endoscopic Evaluation of Swallowing (DIGEST‐FEES) scores. This protocol is developed and validated for the HNC population ³⁸ and serves as a standardized overall outcome measure for swallowing ability in both research and clinical settings. It consists of three scales where the DIGEST‐FEES grade represents the overall function and is based on a summary of the DIGEST‐FEES safety and DIGEST‐FEES efficiency ratings. All DIGEST‐FEES grades are scored from 1 (mild) to 4 (life‐threatening/profound). Raters in the present study used a validated Swedish translation that has not yet been published. Prior to performing study evaluations, the SLP raters participated in a one‐day training session in the DIGEST‐FEES protocol held by an SLP who had participated in the online DIGEST course held by the MD Anderson Cancer Center.

3.2. Mouth opening

The primary endpoint for mouth opening was the MIO, which is the maximal distance between the edges of the incisors of the mandible and the maxilla. MIO was measured using a ruler with the patient in an upright position. The cut‐off for trismus was set at an MIO of ≤35 mm as defined by Dijkstra et al. ³⁹

3.3. European Organization for Research and Treatment of Cancer Quality of Life questionnaire head and neck module (EORTC QLQ‐H&N35)

HRQL was evaluated using the European Organization for Research and Treatment of Cancer (EORTC) Quality of Life Head and Neck Module (QLQ‐H&N35), a valid and reliable instrument ⁴⁰ that was developed to capture symptoms specific to HNC. The instrument consists of 35 items divided into seven symptom domains and 11 single items. The ratings are based on Likert scales, with responses from one (not at all) to four (very much), which are converted into scores ranging from 0 to 100. Lower scores reflect a lower symptom burden and a ≥10‐point change is considered a minimal clinically important change.

4. STATISTICAL ANALYSIS

The two primary and all secondary variables were analyzed for the intention to treat (ITT) population, that is, all eligible randomized participants (Figure 1). Missing data were imputed applying multiple imputation. To assess the robustness of the multiple imputation, per protocol analyses were performed. The per protocol population was defined as all randomized participants who adhered to the preventive exercises ≥50% and participated at follow‐up after oncological treatment. An additional sensitivity analysis was performed on complete cases. For the primary analysis the significance level was set to 0.025 due to two primary variables being tested. All other outcome variables were tested and evaluated exploratively and a p‐value of 0.05 was applied.

Multiple imputations with 50 study samples were performed for the primary analyses (PAS and MIO). Baseline patient characteristics that were related to the outcome variables change in PAS and MIO and their missingness were included in the imputation regression model. General linear models were applied for normally distributed outcome variables adjusted for randomization strata and baseline value of the outcome. For test between two groups, Fisher's exact test was used for dichotomous variables, Mantel–Haenszel chi‐square trend test for ordered categorical variables, chi‐square test for non‐ordered categorical variables, t test for normally distributed variables or Mann–Whitney U test in case of skewed data. The relationship between continuous variables was analyzed by the Pearson correlation coefficient for normally distributed variables. Otherwise, the Spearman correlation coefficient was used.

For descriptive statistics, categorical variables were described by number and percentages and continuous variables by the mean, standard deviation (SD), median, minimum and maximum.

All tests were two‐tailed. All analyses were performed by using SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA).

4.1. Ethical considerations

The study was approved by the Swedish Ethical Review Authority (reference number 1151‐18/2019‐00752) and was conducted according to the Declaration of Helsinki of 1975 as revised in 1983. Before inclusion in the study, all participants gave their written informed consent.

5. RESULTS

5.1. Participant characteristics

The intervention and control groups were similar in terms of disease characteristics, oncological treatment, and sociodemographic factors, with no statistically significant differences prior to treatment (Table 1).

TABLE 1.

Treatment and sociodemographic data for the participants in the intervention and control group at baseline.

Variable	Intervention (n = 45)	Control (n = 44)	p‐value
	Mean ± SD Median (Min; Max)	Mean ± SD Median (Min; Max)
Age (years)	64.5 ± 9.7 64 (43–81)	63.5 ± 9.2 65 (42–83)	0.68
Weeks since completion of radiotherapy to first follow‐up after treatment	7.8 ± 3.1 7.1 (2.9–16.9)	8.7 ± 4.3 8.0 (3.3–21.1)	0.58
MIO at baseline (mm)	49.7 ± 6.4 49.0 (37.0–64.0)	49.7 ± 6.4 50.0 (39.0–67.0)	0.97
	n (%) a	n (%) a
Sex			0.81
Male	35 (78%)	33 (75%)
Female	10 (22%)	11 (25%)
Tumor location			0.88
Tonsil	25 (56%)	25 (57%)
Base of tongue	13 (29%)	13 (30%)
Hypopharynx	2 (4%)	3 (7%)
Larynx	5 (11%)	3 (7%)
Tumor stage			0.51
I	23 (51%)	23 (53%)
II	5 (11%)	2 (5%)
III	15 (33%)	12 (27%)
IV	2 (4%)	7 (16%)
Radiotherapy	45 (100%)	44 (100%)
Chemotherapy (concomitant)	34 (76%)	37 (84%)	0.32
Swallowing dysfunction at baseline	4 (9%)	2 (5%)	0.68
Comorbidity according to ACE‐27			0.21
None	28 (62%)	20 (46%)
Mild	10 (22%)	14 (32%)
Moderate	6 (13%)	10 (23%)
Severe	1 (2%)	0 (0%)
HPV positive	39 (87%)	38 (88%)	0.92
Not tested	3 (7%)	3 (7%)
Missing	0	1
Living alone	11 (24%)	9 (21%)	0.80
Education			0.54
Elementary school	10 (26%)	11 (26%)
High school 2–4 years	14 (36%)	19 (45%)
College/university	15 (39%)	12 (29%)
Missing	6	2
Occupation			0.63
Working	16 (36%)	15 (34%)
Sick leave	6 (13%)	9 (21%)
Retired	22 (49%)	20 (46%)
Unemployed	1 (2%)	0 (0%)
Smoking			0.47
Never smoked	17 (46%)	16 (38%)
Stopped smoking >12 months	11 (30%)	17 (41%)
Stopped smoking <12 months	9 (24%)	5 (12%)
Smoking	0 (0%)	4 (10%)
Missing	8	2

Note: For test between two groups with respect to dichotomous variables Fisher's exact test was used, for ordered categorical variables Mantel–Haenszel chi‐squared trend test, for non‐ordered categorical variable chi‐squared test, and for continuous variables Mann–Whitney U test (not normally distributed) and two‐sample t test (normally distributed) were used.

Abbreviations: ACE‐27, Adult Comorbidity Index; HPV, human papillomavirus; MIO, Maximal interincisal opening.

^a Percentages rounded, therefore does not always sum to 100.

Overall, 31% of participants still had a nasogastric tube at the follow‐up (Table 2), and 67% of participants had a weight loss of ≥7.5%. The proportion of participants with trismus was lower in the intervention group (6%) than in the control group (18%) at the follow‐up (not significant). There were no statistically significant differences between the two groups at baseline or follow‐up for any of the variables presented in Table 2.

TABLE 2.

Nutritional and eating related variables in the intervention and control group at baseline and follow‐up. and post intervention.

	Intervention group (n = 45)		Control group (n = 44)
	Baseline	Follow‐up	Baseline	Follow‐up
Variable	n (%)	n (%)	n (%)	n (%)
BMI classification
Under weight (<18.5)	0 (0%)	1 (3%)	0 (0%)	1 (3%)
Normal weight (18.5–24.9)	13 (30%)	19 (50%)	14 (32%)	19 (49%)
Over weight (25–29.9)	20 (46%)	15 (40%)	21 (48%)	14 (36%)
Obese ≥30	12 (27%)	3 (8%)	9 (21%)	5 (13%)
Missing	0	7	0	5
Salivary flow
Hyposalivation (≤0.7 mL/min)	4 (11%)	14 (48%)	10 (28%)	10 (37%)
Missing	8	17	8	16
Mouth opening
Trismus (≤35 mm)	0 (0%)	2 (6%)	0 (0%)	7 (18%)
Missing	0	9	0	6
Nasogastric tube
Nasogastric tube user	0 (0%)	12 (31%)	0 (0%)	12 (31%)
Missing	0	6	0	5
PEG
PEG user	0 (0%)	2 (5%)	0 (0%)	0 (0%)
Missing	0	6	0	5
Pneumonia
Yes	0 (0%)	0 (0%)	2 (5%)	0 (0%)
Missing	0	7	0	6

Note: Percentages rounded, therefore does not always sum to 100. There were no statistically significant differences between the groups at baseline or follow‐up. Number of missing data varies at follow‐up depending on variable.

Abbreviations: BMI, body mass index; PEG, percutaneous endoscopic gastrostomy.

5.2. Swallowing function and mouth opening

There were no statistically significant differences between the two groups concerning change in max PAS score and MIO from baseline to follow‐up (p = 0.60 and p = 0.16, respectively), Table 3. Swallowing function measured by PAS deteriorated from baseline, where mean scores increased to 4.1 and 3.5 in the intervention and control groups, respectively (mean change 1.9 and 1.4, respectively). Mouth opening decreased somewhat less in the intervention group, −4 mm, compared with the control group, −7 mm. Robustness analysis using the per‐protocol population revealed a statistically significantly less deteriorated mouth opening in the intervention group (p = 0.03), but not swallowing function (p = 0.77).

TABLE 3.

Results of the primary variables Penetration Aspiration Scale score (PAS) and maximum interincisal mouth opening (MIO) measurements for the intervention and control group at baseline and follow‐up and group comparison of change after radiotherapy.

	Intervention group (n = 45)			Control group (n = 44)			Comparison of intervention and control group a
	Baseline	Follow‐up	Change (Δ) baseline to follow‐up	Baseline	Follow‐up	Change (Δ) baseline to follow‐up	Change (Δ) baseline to follow‐up
	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean (95% CI)
Variable	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	p‐value
PAS	2.1 ± 1.5 1.0 (1.0–5.0)	4.1 ± 2.3 5.0 (1.0–8.0)	1.9 ± 2.8 2.0 (−4.0 to 7.0)	2.0 ± 1.3 1.0 (1.0–5.0)	3.5 ± 2.3 3.0 (1.0–8.0)	1.4 ± 2.3 0.0 (−4.0 to 7.0)	0.39 (−1.07 to 1.86) p = 0.60
MIO (mm)	49.7 ± 6.4 49.0 (37.0–64.0)	45.1 ± 7.3 45.0 (30.0–68.0)	−4.2 ± 5.2 −4.0 (−19.0 to 6.0)	49.7 ± 6.4 50.0 (39.0–67.0)	42.1 ± 9.6 41.5 (20.0–61.0)	−7.0 ± 8.7 −6.0 (−31.0 to 11.0)	2.73 (−1.11 to 6.58) p = 0.16

Note: The scale ranges from 1 (material does not enter the airway) to 8 (material enters the airway, passes below the vocal folds, and no effort is made to eject).

^a Primary analysis was performed applying general linear models adjusting for randomization strata and baseline value of the outcome. Missing data was handled by multiple imputation, including all baseline variables related to the outcome and its missingness.

Analysis of DIGEST‐FEES safety and efficiency ratings showed no statistically significant effects of the intervention on swallowing safety and the amount of residue left in the pharynx after swallowing, respectively. There were no statistically significant differences in the comparison of the overall DIGEST‐FEES grade, Table 4.

TABLE 4.

Comparison of Dynamic Imaging Grade of Swallowing Toxicity (DIGEST) scales: Safety, efficiency and overall DIGEST‐score at baseline and follow‐up for control and intervention group as well as sub‐groups by level of adherence to preventive exercises.

		Intervention group (n = 45)	Control group (n = 44)	Adherence <50% (n = 11)	Adherence 50%–75% (n = 11)	Adherence ≥75% (n = 13)
		(%) a	(%) a	(%) a	(%) a	(%) a
Baseline b	DIGEST safety
	Normal (0)	34 (76%)	36 (82%)	10 (91%)	8 (73%)	9 (69%)
	Mild (1)	8 (18%)	8 (18%)	0 (0%)	3 (27%)	3 (23%)
	Moderate (2)	3 (7%)	0 (0%)	1 (9%)	0 (0%)	1 (8%)
	Severe (3)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
	Profound (4)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
	DIGEST efficiency
	Normal (0)	10 (22%)	11 (26%)	6 (55%)	0 (0%)	2 (15%)
	Mild (1)	23 (51%)	23 (54%)	2 (18%)	7 (64%)	8 (62%)
	Moderate (2)	3 (7%)	1 (2%)	0 (0%)	1 (9%)	2 (15%)
	Severe (3)	9 (20%)	8 (19%)	3 (27%)	3 (27%)	1 (8%)
	Profound (4)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
	DIGEST‐score
	Normal (0)	9 (20%)	10 (23%)	6 (55%)	0 (0%)	1 (8%)
	Mild (1)	25 (56%)	25 (58%)	2 (18%)	8 (73%)	10 (77%)
	Moderate (2)	10 (22%)	8 (19%)	2 (18%)	3 (27%)	2 (15%)
	Severe (3)	1 (2%)	0 (0%)	1 (9%)	0 (0%)	0 (0%)
	Profound (4)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Follow‐up b	DIGEST safety
	Normal (0)	11 (33%)	16 (50%)	3 (33%)	3 (30%)	5 (42%)
	Mild (1)	11 (33%)	4 (13%)	2 (22%)	3 (30%)	5 (42%)
	Moderate (2)	9 (27%)	9 (28%)	4 (44%)	2 (20%)	2 (17%)
	Severe (3)	2 (6%)	3 (9%)	0 (0%)	2 (20%)	0 (0%)
	Profound (4)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
	DIGEST efficiency
	Normal (0)	0 (0%)	3 (9%)	0 (0%)	0 (0%)	0 (0%)
	Mild (1)	15 (46%)	15 (47%)	4 (44%)	3 (30%)	7 (58%)
	Moderate (2)	5 (15%)	1 (3%)	2 (22%)	2 (20%)	0 (0%)
	Severe (3)	12 (37%)	13 (41%)	2 (22%)	5 (50%)	5 (42%)
	Profound (4)	1 (3%)	0 (0%)	1 (11%)	0 (0%)	0 (0%)
	DIGEST‐score
	Normal (0)	0 (0%)	3 (9%)	0 (0%)	0 (0%)	0 (0%)
	Mild (1)	13 (39%)	11 (34%)	2 (22%)	4 (40%)	6 (50%)
	Moderate (2)	14 (42%)	10 (31%)	5 (56%)	3 (30%)	5 (42%)
	Severe (3)	6 (18%)	8 (25%)	2 (22%)	3 (30%)	1 (8%)
	Profound (4)	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)

Note: Percentages rounded, therefore does not always sum to 100. There were no statistically significant differences between control and intervention group at baseline or follow‐up. Group comparisons of DIGEST‐scale score were made by Mantel–Haenszel chi‐squared trend test for ordered categorical variables.

^a p ≤ 0.05 between adherence group for DIGEST efficiency and DIGEST‐score at baseline.

^b Missing: 0 in all groups at baseline and 12 at follow‐up, in both control and intervention group. Two in the <50% adherence group and 1 in the 50%–75% and 1 ≥75%.

Further, describing the participants as a whole, most presented with residue at baseline, as only 24% had a normal DIGEST‐FEES efficiency rating while 79% had a normal DIGEST‐FEES safety rating. Both safety and efficiency ratings deteriorated after oncological treatment. Only 5% of participants had a normal overall DIGEST‐FEES grade at follow‐up compared with 21% at baseline.

5.3. European Organization for Research and Treatment of Cancer Quality of Life questionnaire head and neck module (EORTC QLQ‐H&N35)

After oncological treatment, both groups had statistically significant worsening of several HNC‐related symptoms, for example, social eating, dry mouth, sticky saliva, and coughing, Table 5. Concerning swallowing function, the intervention group reported a deterioration in function from baseline to follow‐up of 22.7 points and the control group 14.0 points. For mouth opening change from baseline was, 6.2 points and 9.7 points, respectively. However, no statistically significant effect of the preventive intervention could be detected in HNC related symptoms between the groups when comparing the change from baseline to follow‐up.

TABLE 5.

Results of the EORTC QLQ‐H&N35 for the intervention and control group at baseline and follow‐up and group comparison of change after radiotherapy.

	Intervention group (n = 45)			Control group (n = 44)			Comparison between intervention and control group
	Baseline	Follow‐up	Change (Δ)	Baseline	Follow‐up	Change (Δ)	Difference in Δ from PRE to POST
			PRE to POST			PRE to POST
			Follow‐up			Follow‐up
	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD
Variable	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	p‐value
Pain	20.8 ± 25.5 12.5 (0.0–100.0)	35.4 ± 28.6 33.3 (0.0–100.0)	13.6 ± 32.7 16.7 (−50.0 to 100.0)	18.2 ± 22.9 8.3 (0.0–91.7)	22.9 ± 18.6 16.7 (0.0–83.3)	5.9 ± 27.4 8.3 (−66.7 to 58.3)	0.52
Swallowing	11.9 ± 21.5 0.0 (0.0–83.3)	33.8 ± 25.1 33.3 (0.0–91.7)	22.7 ± 31.3 29.2 (−50.0 to 75.0)	10.3 ± 18.8 0.0 (0.0–66.7)	23.9 ± 25.3 11.1 (0.0–91.7)	14.0 ± 31.8 8.3 (−50.0 to 91.7)	0.15
Senses	13.3 ± 22.4 0.0 (0.0–83.3)	42.9 ± 30.1 50.0 (0.0–100.0)	31.3 ± 28.9 33.3 (−16.7 to 100.0)	10.1 ± 20.6 0.0 (0.0–100.0)	36.1 ± 23.9 33.3 (0.0–100.0)	27.5 ± 26.3 33.3 (−16.7 to 100.0)	0.63
Speech	10.8 ± 14.9 0.0 (0.0–55.6)	20.5 ± 22.4 11.1 (0.0–88.9)	8.0 ± 24.1 0.0 (−33.3 to 55.6)	8.5 ± 10.9 5.6 (0.0–44.4)	16.5 ± 16.7 11.1 (0.0–66.7)	7.0 ± 19.5 11.1 (−44.4 to 55.6)	0.98
Social eating	9.7 ± 17.5 0.0 (0.0–83.3)	31.6 ± 24.5 33.3 (0.0–100.0)	23.0 ± 25.9 25.0 (−33.3 to 100.0)	9.5 ± 14.2 0.0 (0.0–50.0)	24.1 ± 22.0 23.6 (0.0–100.0)	15.2 ± 24.0 16.7 (−16.7 to 100.0)	0.08
Social contact	5.3 ± 15.2 0.0 (0.0–80.0)	12.1 ± 17.4 0.0 (0.0–73.3)	7.5 ± 17.4 0.0 (−26.7 to 73.3)	14.8 ± 61.2 0.0 (−13.3 to 393.3)	12.0 ± 16.6 6.7 (0.0–53.3)	6.2 ± 20.5 0.0 (−60.0 to 53.3)	0.99
Sexuality	34.3 ± 36.5 33.3 (0.0–100.0)	51.0 ± 36.6 66.7 (0.0–100.0)	17.3 ± 45.0 8.3 (−100.0 to 100.0)	23.4 ± 34.1 0.0 (0.0–100.0)	41.4 ± 42.8 33.3 (0.0–100.0)	21.7 ± 33.9 0.0 (−33.3 to 100.0)	0.99
Teeth problems	15.8 ± 29.2 0.0 (0.0–100.0)	9.1 ± 22.5 0.0 (0.0–100.0)	−10.4 ± 28.6 0.0 (−100.0 to 33.3)	11.6 ± 25.1 0.0 (0.0–100.0)	11.8 ± 25.2 0.0 (0.0–100.0)	0.0 ± 28.5 0.0 (−66.7 to 66.7)	0.23
Mouth opening	10.8 ± 24.3 0.0 (0.0–100.0)	17.2 ± 20.6 0.0 (0.0–66.7)	6.2 ± 28.6 0.0 (−66.7 to 66.7)	5.4 ± 14.4 0.0 (0.0–66.7)	15.1 ± 20.8 0.0 (0.0–66.7)	9.7 ± 21.4 0.0 (−33.3 to 66.7)	0.78
Dry mouth	16.7 ± 30.2 0.0 (0.0–100.0)	58.6 ± 36.4 66.7 (0.0–100.0)	43.8 ± 31.0 33.3 (0.0–100.0)	13.2 ± 26.4 0.0 (0.0–100.0)	58.1 ± 31.0 66.7 (0.0–100.0)	44.1 ± 30.3 33.3 (−33.3 to 100.0)	0.91
Sticky saliva	17.9 ± 29.5 0.0 (0.0–100.0)	59.6 ± 36.1 66.7 (0.0–100.0)	45.2 ± 32.8 33.3 (0.0–100.0)	13.2 ± 24.3 0.0 (0.0–100.0)	60.2 ± 34.9 66.7 (0.0–100.0)	49.5 ± 33.2 33.3 (0.0–100.0)	0.85
Coughing	18.3 ± 25.0 0.0 (0.0–100.0)	24.2 ± 25.4 33.3 (0.0–66.7)	6.3 ± 35.4 0.0 (−100.0 to 66.7)	17.8 ± 22.2 0.0 (0.0–100.0)	33.3 ± 27.2 33.3 (0.0–100.0)	16.1 ± 29.7 33.3 (−66.7 to 66.7)	0.16
Feeling ill	19.3 ± 29.0 0.0 (0.0–100.0)	35.4 ± 31.1 33.3 (0.0–100.0)	16.5 ± 34.8 30.0 (−66.7 to 66.7)	18.6 ± 26.5 0.0 (0.0–100.0)	19.4 ± 24.0 0.0 (0.0–100.0)	3.2 ± 27.7 0.0 (−66.7 to 66.7)	0.06

Note: Lower scores represent better functioning. For test between two groups with respect to continuous variables Mann–Whitney U test (not normally distributed) was used.

Abbreviation: EORTC QLQ‐H&N35, European Organization for Research and Treatment of Cancer Quality of Life questionnaire head and neck module.

5.4. Adherence to preventive protocol and maintenance of eating and drinking

Throughout the exercise period, the intervention group performed 61% of the prescribed training sessions. Adherence was somewhat higher for jaw exercises (passive and active; 62%) than for the swallowing exercise (58%). Adherence to the protocol was lowest during weeks 4–6 of radiotherapy and varied from 41% to 82% during the exercise period.

There were no statistically significant differences between the two groups concerning self‐reported daily eating and drinking. The intervention group reported swallowing some food or drink on 77% of days during the treatment period compared with 81% for the control group. Oral intake was lowest at the end of radiotherapy in both groups.

Exploratory analyses were made comparing PAS scores and MIO within the intervention group based on level of adherence: low (<50% adherence; n = 11), moderate (50%–75%; n = 11), and high (>75%; n = 13). ⁴¹ , ⁴² The high and moderate adherence groups had less deterioration in MIO than the low adherence group (−2.8 and −3.5 vs. −5.7 mm, respectively), however, there was no statistically significant difference (p = 0.43). For PAS scores, the high adherence group had less deterioration than the moderate and low adherence groups (0.4 vs. 2.2 and 3.3, respectively), however, there was again no statistically significant difference between the adherence groups (p = 0.06), Table 6 and Figure 2. Based on PAS scores, 78% of participants in the low adherence group experienced deterioration compared with only 42% in the high adherence group. Conversely, no participants in the low adherence group improved over time, while 33% in the high adherence group improved.

TABLE 6.

Comparison of primary variables Penetration Aspiration Scale score (PAS) and maximum interincisal mouth opening (MIO) by levels of adherence to preventive exercises.

	Adherence <50% (N = 11)			Adherence 50%–75% (N = 11)			Adherence ≥75% (N = 13)			Comparison of change between adherence groups
	Baseline	Follow‐up	Change (Δ) baseline to follow‐up	Baseline	Follow‐up	Change (Δ) baseline to follow‐up	Baseline	Follow‐up	Change (Δ) baseline to follow‐up	Change (Δ) baseline to follow‐up
	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD	Mean ± SD
Variable	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	Median (Min; Max)	p value
PAS	1.4 ± 1.2 1.0 (1.0–5.0)	4.8 ± 2.0 5.0 (1.0–8.0)	3.3 ± 2.4 4.0 (0.0–7.0)	2.1 ± 1.4 1.0 (1.0–5.0)	4.4 ± 2.6 5.0 (1.0–8.0)	2.2 ± 2.9 1.5 (−2.0 to 6.0)	2.7 ± 1.8 3.0 (1.0–5.0)	3.3 ± 2.3 3.0 (1.0–8.0)	0.4 ± 2.9 0.0 (−4.0 to 4.0)	0.06
MIO (mm)	47.5 ± 6.7 45.0 (37.0–62.0)	40.4 ± 5.2 40.0 (30.0–48.0)	−5.7 ± 3.6 −5.5 (−14.0 to 0.0)	51.5 ± 7.4 52.0 (40.0–64.0)	47.9 ± 9.2 46.0 (35.0–68.0)	−3.5 ± 7.3 −1.0 (−19.0 to 6.0)	47.9 ± 5.2 46.0 (40.0–57.0)	45.5 ± 5.9 44.0 (37.0–54.0)	−2.8 ± 3.9 −2.5 (−11.0 to 2.0)	0.43

Note: The scale ranges from 1 (material does not enter the airway) to 8 (material enters the airway, passes below the vocal folds, and no effort is made to eject). Data are presented as mean ± standard deviation (SD), median (range). For test between three groups with respect to continuous variables overall test from a general linear model was used.

Abbreviations: MIO, maximum interincisal mouth opening; PAS = Penetration Aspiration Scale.

FIGURE 2.

Comparison of change in maximum interincisal mouth opening (MIO) and Penetration Aspiration Scale score (PAS) between the control group and by level of adherence to preventive exercises. [Color figure can be viewed at wileyonlinelibrary.com]

Concerning overall swallowing function measured by the DIGEST‐FEES grade, there was a statistically significant difference between the adherence groups at baseline (p = 0.01). the low adherence group had a higher percentage of normal ratings than the moderate and high adherence groups (55% vs. 0% and 8%, respectively), Table 4. After oncological treatment, no participant had a normal DIGEST‐FEES grade and there were no statistically significant differences between the groups. The high and moderate adherence groups had a higher percentage of participants with mildly impaired swallowing function than the low adherence group (50% and 40% vs. 22%, respectively).

5.5. Intra‐ and inter‐rater reliability

Weighted kappa (kw) analysis of intra‐rater reliability was substantial (kw = 0.61–0.80) on all swallowing measures except for the DIGEST‐FEES grade and the efficiency rating for one of the raters, where agreement was almost perfect (kw = 0.81–1.00). ⁴² Inter‐rater reliability varied from moderate (kw = 0.41–0.60) for the DIGEST‐FEES grade and the efficiency rating to substantial for the DIGEST‐FEES safety rating and PAS score.

6. DISCUSSION

The present study is one of few RCTs comparing the effect of preventive exercises on swallowing function ¹⁹ , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ and mouth opening ability ¹⁵ , ¹⁶ , ¹⁹ to a non‐active control group. The preventive protocol evaluated is unique as it is simplified, including only one swallowing exercise in addition to active and passive jaw‐training. Our results did not reveal any statistically significant differences between the intervention and control groups.

On a group level, we detected no effect on short‐term swallowing function measured by PAS. To our knowledge, no studies in HNC have determined a significant effect on PAS when comparing preventive exercises to a control group, but a meta‐analysis has identified a positive effect on swallowing safety. ⁹ , ¹¹ , ⁴³ Nevertheless, a retrospective study found that preventive exercise during oncologic treatment led to better PAS outcomes compared with reactive intervention at the onset of swallowing difficulties. ⁴⁷ Also, it has been discussed that preventive exercises may influence long‐term function; therefore, it is important to evaluate the potential effect in the present study over time. ⁴⁸

In the present study, we did not see any effects of the preventive exercises on mouth opening between the intervention and control groups. A previous review covering trismus prevention concluded that there is not enough evidence to determine the effectiveness of such exercises. ¹² However, in studies where efforts were made to increase adherence to exercise, statistically better mouth opening outcomes have been observed. ¹⁸ , ⁴⁹ This indicates that low adherence might be undermining the effects of preventive mouth opening interventions in HNC. A recent meta‐analysis found positive effects on both short‐ and long‐term mouth opening by pooled data from the preventive mouth opening interventions. In addition, protocols where exercise was combined with a jaw‐opening device and telephone supervision had better results. ¹⁴ In the present study, a jaw‐opening device was used, and exercises were supervised once a week by the SLPs. Yet, we found no positive effects on mouth opening compared with the control group.

Based on DIGEST‐FEES and HRQL data, participants were clearly impacted by the adverse effects of oncological treatment. Swallowing function had deteriorated from baseline in the majority of participants, with increased events of penetration and aspiration, and more participants presenting with moderate to severe DIGEST‐FEES grades at the follow‐up. Approximately a third of participants in both groups were still feeding tube‐dependent at the follow‐up, and 67% had a weight loss of more than 7.5% from baseline. A previous study demonstrated that a weight loss of 7.5% predicted worse swallowing outcomes. ⁵⁰ These results highlight the importance of managing symptoms that affect eating negatively. The significant symptom burden after oncological treatment is also reflected in patient‐reported HRQL measured by the EORTC QLQ‐H&N35 albeit no statistically significant effects of the intervention were detected. A somewhat surprising result was that the intervention group reported worse swallowing function than the control group at the follow‐up. As objective measures show equal ability to swallow in both groups, this result might be a consequence of increased focus placed on swallowing function due to exercises performed by participants in the intervention group. In studies evaluating intervention by patient reported outcomes this should be considered in the interpretation of results.

Feeding tube dependence can be regarded as an important indicator of severe swallowing problems in HNC patients. The majority of studies evaluating the effects of preventive exercises on feeding tube dependence have found a statistically significant difference in favor of the intervention group. ⁵¹ In the current study, we could not detect such a trend among participants who performed preventive exercises. However, all participants were encouraged by SLPs to keep eating and drinking what they could throughout radiotherapy, an approach that is considered to reduce atrophy of the swallowing musculature and benefit swallowing function. ⁶ Barbon et al. evaluated the effect of maintained oral intake and preventive exercises during radiotherapy on feeding tube outcomes from prospectively collected registry data. At the end of radiotherapy, approximately a third of patients had ongoing feeding tube use, similar to levels in the present study. In addition to feeding tube dependency, Barbon et al. reported that patients who exercised without any oral intake or maintained an oral diet during radiotherapy were at a lower risk of a modified oral diet compared with those who neither maintained an oral diet nor exercised. ⁵² This suggests that encouraging patients to eat and drink to the best of their ability might have had a positive impact on swallowing function in both the intervention and control groups.

In the present study, swallowing function and mouth opening reflected the level of adherence to the exercises; the low adherence group had the worst swallowing and mouth opening outcomes, and the high adherence group had the best. Interestingly, the low adherence group had the best swallowing ability at baseline, that is, better PAS score and a higher proportion of participants with a normal swallowing function according to the overall DIGEST‐FEES grade. It is possible that participants were less motivated to perform the exercises if their swallowing function was not impacted at baseline.

Even though the study protocol was simplified in hope to improve adherence, adherence was similar (62%) to other RCTs with rates reported from 38% to 78%. ¹⁹ , ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ Further, several facilitating factors for intervention adherence were implemented, including providing clear instructions with the prescription of exercise dose, monitoring of exercises through diary entries, and weekly contacts with SLPs. ²³ , ²⁵  Adherence was at its lowest from week four until the end of radiotherapy, suggesting that even the simplified protocol was challenging to perform during this critical period. The explorative analysis indicated that ≥75% adherence to the protocol, that is, exercising approximately 5 days a week, could help maintain swallowing function. If further research could establish that this dosage does have positive effect on swallowing function, we believe this knowledge might help motivate patients during the most challenging period. Another possibility is to increase the exercise dose to two sets per day in hope that more participants reach levels totaling up to ≥75% over the whole treatment period. Of course, possible benefits of this measure would have to be further investigated. One effort to improve adherence could be to offer more frequent contact with a SLP during this critical period and to adapt exercises depending on the symptom burden. For instance, some patients reported that swallowing exercises were difficult to perform due to pain. The low adherence group could have experienced side effects from treatment to a greater extent. In such cases, a patient might be recommended a modified exercise protocol until the acute effects of radiotherapy subside.

A limitation of this study was that participation was not blinded and participants knew if they were part of the active group or not which might have influenced self‐reported function scores on the EORTC QLQ‐H&N35. Second, the chosen follow‐up time point of approximately 1‐month might be too early to evaluate mouth opening outcomes since restriction has been reported to develop around 3 months after oncological treatment. ⁸ Similarly, acute side‐effects as lymphedema might impact the swallowing function and possible effects of the intervention might become evident over time. ⁵³ Another limitation is that it is not guaranteed that each repetition of the tongue hold and the active jaw exercise will be executed in the same manner, that is, the tongue protrusion might vary slightly and the instruction to press down until tension is felt but not discomfort is subjective. Further, the variation in PAS scores was greater than assumed in the power calculation; therefore, the results may have differed using a larger sample size. Nevertheless, results for mouth opening corresponded to the assumptions in the initial power calculation and should not be underpowered.

In summary, the present RCT evaluated a preventive exercise protocol for swallowing and mouth opening ability in a relatively large HNC cohort treated with radiotherapy. Swallowing function and MIO were evaluated by instrumental assessment and validated scales as well as by self‐reported function, which is requested in the research field. The study is strengthened by the randomized design and the balanced groups with similar treatment and sociodemographic characteristics at baseline.

7. CONCLUSION

We found no significant differences between patients performing preventive exercises and non‐active controls in terms of swallowing function and mouth opening. Among patients who had high adherence to the preventive exercise protocol (>75% of sessions), there was a trend toward better swallowing and mouth opening outcomes. However, the high adherence group was small and results comparing outcomes by adherence were not statistically significant.

Petersson K, Finizia C, Pauli N, Tuomi L. Preventing radiation‐induced dysphagia and trismus in head and neck cancer—A randomized controlled trial. Head & Neck. 2025;47(1):159‐174. doi: 10.1002/hed.27886

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.