The study was performed at a tertiary Head and Neck Oncology Department and approved by the local Medical Ethical Committee (METC21.0904/N21STL). The guidelines of the Helsinki Declaration were followed, and written informed consent was obtained from each participant before inclusion.

Participants

Between April 2022 to February 2023, individuals (≥ 18 years) who underwent TL and were experiencing dysphagia were recruited from the own institute and via a notice on the Dutch Patient Association for Head and Neck platform. Participants had to be at least six months post-surgery before enrollment and, if applicable, had completed their postoperative (chemo-) radiotherapy at least six months ago. Pharynx reconstruction method was not a selection criterion. We aimed to include 20 participants based on previous results demonstrated in the healthy senior subjects (Cohen’s d > 0.6) [31]. At the end of the enrollment period, 21 participants [17 men (81%), four women (19%)] were included and signed informed consent. One participant (S13) changed his mind after a few days and withdrew from the study before starting, leaving 20 participants for inclusion.

The median age at baseline was 71 years (range 45–78), and the median time after TL was 47 months (range 9–274). Six participants (30%) had a history of stenosis and had undergone one or more dilatations. Nineteen participants were able to maintain an adequate oral diet, and one participant was feeding tube-dependent. Participant characteristics are displayed in Table 1.

Table 1 Participant characteristicsMultidimensional Assessment Program

A multidimensional assessment program was set up to assess the specific nature and extent of the dysphagia and to investigate whether training with the SEA 2.0 improves subjective and/or objective swallowing function. Different questionnaires/patient-reported outcome measurements (PROMs) and objective assessments were combined. Since none of the measurements, except the Swallowing Outcome After Laryngectomy (SOAL) questionnaire, are developed for TL patients, tools frequently used in the Head and Neck patient population were selected. The PROMs were completed by the participants individually, and the clinician checked if all questions were answered to ensure all data were present. All objective assessments were video-recorded with the SONY ZV-E10 camera placed on a ROLLEI mini M1 tripod. The average time to complete the total assessment was approximately 90 min. All outcome parameters were assessed prior to participation (at baseline, T0), two days after the 6-week training period (T1) and after eight weeks of rest (longer term results, T2).

Patient-Reported OutcomesLaryngectomy Dysphagia Complaints Inventory

The assessment was started with a study-specific structured interview was held with each participant. The participants were asked about eating, drinking, swallowing, (type of) oral intake, adaptions, duration, anxiety, regurgitation, stenosis, and sensitivity. During the interview, the clinician scored the topics dichotomous as “problem or coping strategy present” or “problem or coping strategy absent.” With the dichotomous outcomes of the interview, a Laryngectomy Dysphagia Complaints Inventory (LDCI) was created (see Table 2).

Table 2 The Laryngectomy Dysphagia complaints inventory—baseline assessmentSubjective Perspective on General Health

Participants’ health state was assessed with the EQ-5D-5L questionnaire using health profiles. This validated questionnaire focuses on mobility, self-care, usual activities, pain/discomfort, and anxiety/depression and includes five levels (no, slight, moderate, severe, and extreme problems) for each modality. The outcome ranges from − 0.446 to 1.000, with a higher score indicating better health state. The reference score is 0.839 [33]. It also includes a vertical visual analog scale (VAS) ranging from 0 to 100 to record the participants’ self-rated health status.

Subjective Perspective on Swallowing Function and Oral Intake

The Functional Oral Intake Scale (FOIS) and International Dysphagia Diet Standardization Initiative (IDDSI) assessed oral intake status and type of diet [34,35,36]. The FOIS is a reliable and validated tool that ranges from 1 to 7, with nothing by mouth (1) and no oral restriction (7) [34]. The type of diet was classified with the IDDSI and ranges from 0 to 7, with thin liquid (0) and normal solid or easy to chew consistencies (7) [36].

The MD Anderson Dysphagia Inventory (MDADI) questionnaire, validated for Dutch, was used to evaluate the impact of dysphagia on the quality of life. The MDADI consists of twenty statements with five answer options [Strongly Agree (1), Agree (2), No Opinion (3), Disagree (4), Strongly Disagree (5)]. Total scores range from 20 (extremely low functioning) to 100 (high functioning) [37, 38]. This questionnaire is not specifically designed for TL patients and therefore it also includes questions about aspiration that are not relevant for TL patients; however, the TL patients were able to answer those questions with “Disagree” or “Strongly Disagree.”

The Dutch validated Eating Assessment Tool (EAT-10) questionnaire is a self-administered, symptom-specific outcome instrument for dysphagia [39]. It includes ten questions that assess the initial dysphagia severity with five answer options per question [No Problems (0) to Severe Problems(4)]. A score of three or more is considered abnormal [40]. Also this questionnaire is not specifically designed for TL patients since it includes questions about aspiration; however, the TL patients were able to answer those questions with “No Problems.”

The Swallowing Outcomes after Laryngectomy (SOAL) questionnaire focuses on swallowing problems, specifically for laryngectomized patients. The English version of the SOAL has been validated [41], and the Dutch version is currently being tested for validity. The SOAL consists of 17 questions that assess issues patients may experience with their swallowing function after TL, and every question has three answer options [No (0), A Little (1), A Lot (2)]. Lower scores indicate fewer problems and better self-reported overall swallowing function [41].

Objective MeasurementsBody Weight, Body Length, and BMI

Body weight and body length were provided by the patients, and the Body Mass Index (BMI) was calculated. The BMI was interpreted as follows: < 18.5 Underweight, ≥ 18.5 and < 25.0 Normal weight, ≥ 25.0 and < 30.0 Overweight, and ≥ 30.0 Obese [42, 43]. Reference values for the distribution of the BMI in seniors (65 + years) in the Netherlands (2022) are: 43.8% Under- and Normal weight, 40% Overweight, and 16.2% Obese [42].

Swallowing Assessment

Swallowing function was assessed using two methods. First, videofluoroscopy (VFSS) was used to assess the different phases of the swallow and the amount of residue. All participants were instructed to sit in upright position and to swallow different consistencies in varying amounts of Omnipaque (350 mg I/ml). The protocol included 2 × 10 cc and 1 × 40 cc of thin liquid (IDDSI 0), 2 × 10 cc of extremely thick liquid (IDDSI 4), and 1× (Omnipaque coated) cracker (IDDSI 7) in random order, to avoid a learning effect. The videofluoroscopy videos were recorded at 25 frames per second and exported as AVI file.

For analysis, the Dynamic Imaging Grade of Swallowing Toxicity (DIGEST) was used [44, 45]. The DIGEST is a reliable, validated ordinal scoring tool for pharyngeal dysphagia. The DIGEST uses a Safety and Efficiency Grade to quantify pharyngeal bolus transit. Since laryngectomized participants have no penetration or aspiration issues (unless there is a leaking voice prosthesis), the Safety Grade is scored 0 (Normal; Material does not enter the airway). To assess the Efficiency Grade, first the percentage of pharyngeal residue remaining in the entirety of the pharynx after the initial swallow of each bolus have to be scored (< 10%, 10%–49%, 50%–90%, and > 90%). Then the estimated percentage residue can be converted into the Efficiency Grade (scored from 0 = Normal to 4 = Profound Impaired). Since all participants had a voice prosthesis, the landmarks for the (neo) pharynx were considered from base of the skull up to the vertebra where the voice prosthesis is located. The overall DIGEST Grade is based on the interaction between the Safety Grade (scored as 0 in our participants) and the Efficiency Grade (scored from 0 to 4) and can be interpreted as the severity of dysphagia (0 = Safe and Efficient, 1 = Safe and Mildly Inefficient, 2 = Safe and Moderately Inefficient, 3 = Safe and Severely Inefficient, 4 = Safe and Profoundly Inefficient). A lower score on the DIGEST was interpreted as better swallowing efficiency and reduced residue in the neopharynx. Besides the DIGEST, visual perceptual outcome variables including the presence of a pseudovallecula or pseudoepiglottis, stenosis, regurgitation (nasal, oral, or pharyngeal), pre-swallow posterior spill, tongue base contact against the posterior neopharynx wall, piecemeal deglutition, oral residue, and liquid wash were scored.

The second objective method was the Swallowing Proficiency for Eating and Drinking (SPEAD) test, which measures the swallowing capacity [46]. The SPEAD test is reliable, feasible, and valid to objectify the transport capacity of the upper digestive tract (in grams per second) and has been developed to evaluate and monitor the swallowing capacity in head and neck cancer patients. The SPEAD test contains three subtasks covering the normal range of food consistencies, including thin liquid, thick liquid, and solid texture. The participant was instructed to sit straight in a chair at the table and asked to swallow three food consistencies as quickly and comfortably as possible, with at least 60 s of rest between the different consistencies. The observer kept track of the time and videotaped the participant during the SPEAD test. The videotape was analyzed on total duration (time between substance touching lips until the end of the last swallow), grams swallowed, number of swallows, and number of chews. Different outcomes were calculated: (1) speed of ingestion per consistency (g/s), (2) average swallow volume (g/swallow), and (3) the SPEAD rate (g/s), combining the mean ingestion speed of the three consistencies. Karsten et al. found ‘normal’ SPEAD rate values of 6 g/s (range 2–11 g/s) for healthy participants and 2 g/s (range 0–10 g/s) for Head and Neck cancer patients diagnosed with dysphagia [46].

To ensure blinded analysis, the order of all videofluoroscopy and SPEAD test recordings at baseline, T1, and T2 measurements was randomized by giving each video a random number between 1 and 60. Both key documents were secured with passwords and saved in an independent folder. A pre-analysis consensus meeting between two Speech and Language Pathologists (SLPs) was held. After the consensus meeting, one trained SLP (MN) analyzed all randomized VFSS and SPEAD test video recordings independently. A second SLP (LvdM) scored 10% of the VFSS recordings, and a third SLP (MLA) scored 10% of the SPEAD test video recordings to assess interrater reliability. After three weeks, the first SLP (MN) again scored 10% of the VFSS and SPEAD videos to assess intra-rater reliability.

Mouth Opening Assessment

The clinician measured the Maximum Interincisor Opening (MIO) in millimeters using the TheraBite Range of Motion (ROM) scale (Atos Medical, Hörby, Sweden). A mouth opening of ≤ 35 mm was considered trismus [47,48,49].

Tongue and Strength Assessment

The Iowa Oral Performance Instrument (IOPI) was used to measure tongue strength [50]. With IOPI, measuring the maximum tongue pressures (at anterior and posterior locations) and endurance is possible using a small air-filled bulb. Participants were instructed to sit up straight and press the tongue upwards on the air-filled bulb to squeeze the bulb against the hard palate. The IOPI digitally measures pressures in kilopascal (kPa). The normal values of the tongue elevation strength (P-max) fall in about 40–80 kPa with an average of about 63 kPa [51]. After one familiarization session, three maximum tongue pressure trials are obtained for each participant, with approximately 2-min rest period between the tests. The mean maximum pressure of the highest two of three values was calculated and used as the participants’ maximal (anterior) tongue strength. To ensure the exact positioning of the bulb in the mouth, a small rubber band was attached around the silicone tube at the level of the lips.

CTAR and JOAR strengths were measured in Newton with a digital dynamometer (MicroFET™, Biometrics, Almere, the Netherlands), mounted in an adapted ophthalmic examination frame, used to avoid head and chin position variations and ensure consistent measurement, as previously described by Kraaijenga et al. [31, 32]. A superior fixed belt stabilized the participants’ heads, and the height of both the chin rest and the superior belt were adjustable per participant. Participants were instructed to sit upright and press their chin down on the dynamometer as effortful as possible, once with their mouth and teeth closed (like the CTAR) and once by opening their jaw/mouth (like the JOAR). The dynamometer measured the maximal isometric chin tuck and jaw opening strength (in Newton). Both measurements were preceded by one familiarization session to exclude learning curve effects and improve the reliability of the values obtained. After the familiarization session, both measures are repeated three times, with a 60-s rest period between the trials. The highest value of three was considered as the 1 Repetition Maximum (1RM). The 1RM is the maximum amount of force that can be generated in one maximal contraction [52]. The mean pressure of the highest two values was calculated and considered as the participants’ mean strength of both exercises [50].

The Swallow Exercise Aid (SEA) 2.0

The previously extensively described SEA is further developed into the new SEA 2.0 (see Fig. 1) [31, 32]. The SEA 2.0 consists of the Chest Bar with Chest Pad (1), Chin Bar (2), Chin Pad (3), Resistance adjustment knob (4), and Handle (5) (see Fig. 2). The range of motion of the Chin Bar is 30 mm in total, resulting in enough space for the tracheostoma and breathing. The SEA 2.0 device provides quantifiable adjustable resistance training for swallowing and jaw exercises, and audible and tactile feedback. Unique to this SEA 2.0 is the mechanism that allows a stepwise resistance force in Newton raising from level 1 (20 Newton) to level 8 (150 Newton), which was determined on 100 devices by Atos Medical (Hörby, Sweden) (see Fig. 3). This mechanism makes it possible to adapt the resistance to participants’ capacity and/or performance.

Fig. 1

schematically drawing of the mechanism (left) and design (right)of the Swallow Exercise Aid (SEA) 2.0

Fig. 2

Picture of the novel Swallow Exercise Aid (SEA 2.0). The SEA 2.0 consists of the Chest Bar with Chest Pad (1), Chin Bar (2), Chin Pad (3), Resistance adjustment knob (4), and Handle (5)

Fig. 3

Load graph of the resistance (in Newton) of the novel Swallow Exercise Aid (SEA 2.0) based on 100 pieces measured at ATOS Medical (Hörby, Sweden)

Due to the altered anatomy and the lack of knowledge about how the remaining and reattached muscles would react to the exercises, it was decided that the exercises should be performed at a level that was strenuous but well tolerated according to patients. To obtain the training load, the initial resistance was set at 60–75%1RM as measured with the MicroFET and was then adjusted to achieve a load that was rated as at least strenuous at 30 repetitions in a first practice round. The final starting resistance in Newton was then deducted from the Load graph.

Exercise Protocol and Logbook

The six-week exercise protocol (identical to the one published by Kraaijenga et al., 2015) of the CTAR and JOAR exercises consisted of an isokinetic and isometric part. During the isokinetic part, the participant was asked to perform the exercise 30 times. During the isometric part, the participant performed the exercise three times for 60 s, with at least 60 s of rest between the sets. The ESAR was performed ten times consecutively after another 60 s of rest. The total duration of the exercises was estimated to be 15–20 min per session. The exercises are displayed in Fig. 4.

Fig. 4

The Swallow Exercise Aid (SEA 2.0) exercises (printed with permission of the participant). Top left: start position, Top right: Chin Tuck Against Resistance (CTAR), Bottom left: Jaw Opening Against Resistance (JOAR), Bottom right: Effortful Swallow Against Resistance (ESAR) with 50% of maximum range of motion

All participants were instructed to perform the SEA 2.0 exercises three times a day, seven days a week, for six weeks in total. They received a written instruction sheet with pictures before starting their ‘six-week training period.’ Moreover, they received daily WhatsApp messages as a reminder to perform the exercises and to ensure that they could quickly get in touch if necessary. Once every one or two weeks, participants came to the hospital for a check-up. During this check-up, the clinician checked if they were doing well, whether he or she was performing the exercises correctly, whether they could increase the training resistance of the SEA 2.0 with one step, and to evaluate the Chin Tuck and Jaw Opening strength with the MicroFET. Participants were instructed to decrease the resistance or stop the exercises if they felt discomfort or pain in the chest, chin, neck, or in/around their temporomandibular joint during or after the exercises.

Feasibility and Adherence

The feasibility of training with the SEA 2.0 and the content of the multidimensional assessment program was assessed using a study-specific questionnaire at the T1 measurements (Online Appendix: Questionnaire 1). To assess the adherence to the SEA 2.0 exercises, participants received a paper exercise logbook before their six-week training period started. All participants were asked to fill in the logbook after every training session and to keep the logbook up to date (Online Appendix: Fig. 14). Exercise adherence was considered good if participants completed for at least 70% of all training sessions.

Patients were evaluated again eight weeks later (T2). For the T1-T2 period, participants received a study-specific questionnaire (Online Appendix: Questionnaire 2) in which they were asked if they had trained with the tool during their eight-week resting period. Other questions were if they experienced differences in swallowing and strength and whether they planned to continue training with the SEA 2.0.

Statistical Analysis

All statistical analyses were performed in R (version 4.2.1). [53] Descriptive statistics were used to characterize the sample. Continuous variables were summarized using median and range. Inter- and intra-observer reliability were assessed using the Two-way random effects model Intraclass Correlation Coefficient (ICC) and interpreted as follows: < 0.30 (negligible), 0.30 to < 0.50 (low), 0.50 to < 0.70 (moderate), 0.70 to < 0.90 (high), and ≥ 0.90 (very high, positive, or negative) correlation.

Linear mixed-effects models (LME) were used to summarize the changes over time of continuous outcomes. Time was used as a categorical variable. We used a random intercept per patient to account for the repeated observations and added a random slope if this improved model fit according to the AIC. This model takes missing outcome data (NA) into account. The estimated marginal means from the model, with a corresponding 95% confidence interval, were plotted along with the individual data. For discrete ordinal outcomes, we present the number and percentages patients improving, worsening, or staying the same.