Effectiveness of the sequential 4-channel NMES compared with conventional 2-channel NMES for the treatment of dysphagia in a prospective double-blind randomized controlled study

Background: To date, conventional swallowing therapies and 2-channel NMES are standard treatments for dysphagia. The precise mechanism of 2-channel neuromuscular electrical stimulation (NMES) treatment has yet to be determined, and controversy remains over the ecacy. The sequential 4-channel NMES was newly developed based on the normal contractile sequence of swallowing-related muscles. Objective: To evaluate and compare the rehabilitative effectiveness of sequential 4-channel NMES with that of conventional 2-channel NMES. Methods: In this prospective randomized case-control study, 23 subjects with dysphagia were enrolled. Twelve subjects with 4-channel NMES group and eleven subjects with 2-channel NMES group completed the intervention. Pretreatment and posttreatment evaluations were performed with the videouoroscopic dysphagia scale (VDS), penetration-aspiration scale (PAS), MD Anderson dysphagia inventory (MDADI), functional oral intake scale (FOIS), and Likert scale. Results: The sequential 4-channel NMES group signicantly improved the VDS (oral, pharyngeal, and total), PAS, FOIS, and MDADI (emotional, functional, and physical scale) compared with pretreatment data. The 2-channel NMES group signicantly improved the VDS (oral, pharyngeal, and total) and MDADI (emotional, physical scale), but not the PAS and FOIS compared with pretreatment data. When the two groups were directly compared, the 4-channel NMES group showed signicant improvement in oral and total VDS. Conclusions: Sequential 4-channel NMES activating the suprahyoid, thyrohyoid, and other infrahyoid muscles with proper interval time can be a new effective treatment for dysphagia. clinicaltrial.gov, The Wilcoxon signed-rank test to compare and The Mann-Whitney test to differences of VDS, and PAS NMES The results ± Scale, PAS: Penetration Aspiration Scale, MDADI: M.D. Anderson Dysphagia Inventory, FOIS: Functional Oral intake Scale.


Introduction
Dysphagia is a common and serious problem in patients with stroke and the prevalence ranges from 37 to 78%. [1] Decreased laryngeal elevation caused by pharyngeal muscles' weakness is a main cause of dysphagia in these patients, which can result in aspiration and pharyngeal residue during swallowing. [2,3] Till now, diverse methods, such as oropharyngeal exercises, compensatory maneuvers, neuromuscular electrical stimulation (NMES), and diet control, are used for dysphagia treatment.
To date, most of the clinical studies regarding NMES evaluated the rehabilitative effects after distinct treatment sessions, and 2-channel NMES has been gaining attention for its muscle strengthening effect by motor stimulation and facilitation of swallowing re ex by sensory stimulation. [4] Freed et al. and Blumenfeld et al. indicated that transcutaneous electrical stimulation is superior to conventional dysphagia management probably due to stimulation of the sensory cortex of the cerebrum, recruitment of more motor units rather than volitional contractions, and an increase in local blood ow. [5,6] However, the precise mechanism of 2-channel NMES treatment has yet to be determined, and controversy remains over the e cacy and method of stimulation. [7] No previous studies have provided the reason for the effectiveness of co-stimulation of suprahyoid and infrahyoid muscles, and a recent randomized controlled trial failed to prove the e cacy of 2-channel NMES in patients with stroke. [8,9] Moreover, conventional 2-channel NMES does not stimulate the muscles as a physiological sequence of muscle activation during swallowing. [7,10] In our previous study, activations of the suprahyoid muscles develop about 150-300 ms earlier than those of the infrahyoid muscles. [10] These sequential contractions of the suprahyoid and infrahyoid muscles induce a circular motion of the hyoid bone during the normal swallowing process, which moves forward-upwardly in the beginning and then backward-downwardly. [11] This result may suggest that simultaneous stimulations of the suprahyoid and infrahyoid muscles could result in the cancellation of positive effects. [12][13][14] But the previous 2-channel NMES stimulated swallowing related muscles simultaneously, which is different from a physiologic motion. Stimulation of these muscles via the 4channel NMES may lead to better modi cations of the abnormal hyoid and laryngeal motion in patients with dysphagia.
Therefore, we hypothesize that the sequential 4-channel NMES based on normal physiology would improve the hyoid and laryngeal motions during swallowing, as well as the swallowing function in general. The purpose of this study was to compare and prove the superiority of sequential 4-channel NMES to 2-channel NMES at pretreatment and posttreatment in a randomized double-blind clinical trial.

Study design
This study was a multicenter, prospective, double-blind, randomized controlled clinical trial from Oct 1, 2018 to August 4, 2019. It was performed at the rehabilitation unit of ve teaching hospitals (Seoul National University Bundang Hospital, Seoul National University Hospital, Hallym University Dongtan Sacred Heart Hospital, Daegu-Patima general Hospital, and the Jeju University Hospital). The study protocol was approved by the institutional review board of each hospital (IRB No.: E-1806/475-002, 2018-07-012, JEJUNUH 2018-07-010, J-1810-064-979, DFH19DPOS033) and all methods were performed in accordance with the relevant guidelines and regulations. It was also approved by the Ministry of Food and Drug Safety in Republic of Korea and registered at clinicaltrial.gov (Registration number: NCT03670498, Initial release: 09/13/2018, Actual study start date: 10/01/2018, Actual study completion date: 08/04/2019, Last release: 07/23/2020). All patients or their representatives provided written informed consent prior to participation. A steering committee (Seoul National University Bundang Hospital) was responsible for the design, conduct, and reporting of the study. Data and safety were monitored every 6 months. The datasets generated and/or analyzed during the study are available from the corresponding author upon reasonable request.

Participants
As the 4-channel NMES was rst developed, there was no study before to nd out the rehabilitative effect of this equipment. Therefore, we randomly calculated 13 subjects in each group, considering 10% of the dropouts. [15] Participants were recruited from the rehabilitation unit of ve teaching hospitals.
Patients were eligible for study participation if they were older than 19 years, presented with cerebral infarction or hemorrhage within 3 months, had at least one symptom of dysphagia such as food sticking, cough with eating, and globus sensation, had con rmed dysphagia by video uoroscopic swallowing study (VFSS), had stable vital signs, agreed to participate in the present study, and gave informed consent. VFSS criteria included de nite presence of aspiration (presence of de nite aspiration (penetration-aspiration scale (PAS) ≥ 6) or presence of penetration (PAS scale, ≥ 2 and ≤ 5) with residual material at vallecular pouch or pyriformis sinus) to prevent ceiling effect. [16] The following patients were excluded: those who had severe cognitive dysfunction who could not perform 1 step follow command, serious psychiatric disorder, previous cervical surgery, respiratory di culty, and cervical surgery. Patients who were pregnant and breast-feeding and those who had cancer and allergic reaction to electrodes of NMES were excluded.

Randomization and masking
Patients were randomly assigned to receive 4-channel NMES or 2-channel NMES (1:1) via computerized random allocation sequences that were prepared by a statistician not involved in participant recruitment. The randomization schedule was only accessible by 2 individuals: the statistician and the primary investigator (JS Ryu). Group allocation was only accessible to occupational therapists and primary investigator. All participants were blinded to the allocation and four designated areas of the electrodes to guarantee masking. The study group received sequential 4-channel NMES, and the control group received conventional 2-channel NMES. In addition, the investigators involved in outcome assessment (JY Jang, SY Lee, D Park, KH SEO) were blinded to group allocation. This study conforms to all CONSORT guidelines and reports the required information accordingly (see Supplementary Checklist).
[10] This device has four channels that are adjustable for amplitude of current, latency, and duration of electrical stimulation. The device uses four pairs of electrodes for electrical stimulation. The electrodes are round in shape and 22 mm in size, and the gap between electrodes consists of two forms: 0.5 cm (type 1 electrode) and 1 cm (type 2 electrode). Type 1 electrode was used for channels 1, 2 and 4, and type 2 electrode was used for channel 3 (Supplementary Fig. 1(B); One Bio Medic Co., Ltd, Bucheon-si, The location of the electrodes was determined using both anatomical landmarks (attachment of each muscle) and manual palpitation. Channel 1 (right) and channel 2 (left) electrodes were placed superior to the hyoid bone and posterior to the mandible with 1-cm interval from the midline, which were targeted for bilateral digastric and mylohyoid muscles. Channel 3 electrodes were placed on bilateral superior pole of the thyroid cartilage for bilateral thyrohyoid muscles, and channel 4 electrodes were placed medial to the sternocleidomastoid muscle and inferior to the thyroid cartilage, which were targeted for the other infrahyoid muscles (sternohyoid, omohyoid, and sternothyroid muscles) ( Fig. 1(A)). In a previous study, a 2-channel NMES on submental and throat stimulations showed better outcome than submental stimulation. The real electrical stimulation was applied to two sets of electrodes attached to the suprahyoid and thyrohyoid muscles ( Fig. 1(B)). [9] Other electrodes used for sham were only attached, but electrical stimulation was not applied.
Electrical stimulation algorithm is based on a previous study. [10,17] Channels 1 and 2 start electrical stimulation rst, and channels 3 and 4 start their stimulations 150 ms and 250 ms later, respectively. The stimulation of channels 1 and 2 lasts 1200 ms, and the stimulations of channels 3 and 4 last 1050 ms and 950 ms, respectively. Therefore, channel 1-4 stimulation ends simultaneously in one sequence. [10] Two-channel NMES (Vitalstim®; Chattanooga Group, Hixson, TN, USA) and 4-channel NMES had the same stimulation parameters. The electrical stimulus of NMES device had a continuous symmetric biphasic waveform. The pulse frequency was 80 Hz, and the pulse interval was 700 µs. The amplitude for each channel could be independently adjusted between 0 and 25 mA. [18] Interventions All participants received 2-or 4-channel NMES for 2-3 weeks (minimal session: 7 times, treatment duration: 420 min-840 min). Treatment durations and daily sessions were different among the 5 rehabilitation units and general patients' condition. For example, some patients received once or twice daily 30 min, once or twice daily 40 min, or once daily 60 min treatment. Some patients had di culty in once daily or twice daily treatment for fatigue or general medical condition. When the participants received once daily 40 min treatment for 2 weeks (10 times) and once daily 60 min treatment for 2 weeks, the treatment durations were 400 and 800 min, respectively. We considered a treatment margin of 20 min. Therefore, we set 420 min and 840 min as the minimal and maximal treatment durations, respectively.
Each participant was familiarized with the expected sensations upon use of the surface electrical stimulation unit. Before starting a recording, the stimulation intensity was gradually increased until the participants felt a tugging sensation. The intensity level was then further increased until the participants felt that any additional increase would not be sensed, yielding the maximum tolerance level similar to previous studies. [7,18,19] This maximum tolerance level was used and recorded for all electrode pairs. In addition to NMES, the patients also performed conventional swallowing therapies or maneuvers such as chin tuck, multiple swallowing, effortful swallowing, supraglottic swallowing, Shaker's exercise, and the Mendelsohn maneuver depending on the clinical symptoms and VFSS ndings. NMES and conventional therapies were given simultaneously by the same occupational therapist for each subject.

Outcome measures
Clinical and VFSS evaluations were performed before and after intervention within 1 week. The maximal duration between initial and follow-up evaluation was 4 weeks to minimize natural recovery. For clinical evaluations, MD Anderson dysphagia inventory (MDADI) was administered to assess quality of life in dysphagia and Likert scale (0-5) to measure satisfaction. [20] For VFSS, subjects were seated upright in a neutral head position under a uoroscopic machine. All uoroscopic images of swallows were digitally recorded. Each VFSS was performed using the following boluses to swallow sequentially: extremely thick uid (International dysphagia diet standardization initiative (IDDSI) 4); dysphagia I diet (IDDSI 4, pureed); dysphagia II diet (IDDSI 5, minced and moist); dysphagia III (IDDSI 7, regular); mildly thick (IDDSI 2); and thin uid (IDDSI 0). [21] Each patient received an initial 3-mL bolus, followed by two 5-mL boluses. Fluids (thick, nectar-like, and thin) were delivered using 10-mL syringes; patients with dysphagia of grades I, II, or III were fed with spoons.
We analyzed the VFSS video using the video uoroscopic dysphagia scale (VDS), functional oral intake scale (FOIS) and PAS. VDS quanti es the severity of dysphagia; VDS is a 14-item scale representing oral and pharyngeal functions that can be observed by VFSS. [22] PAS is an 8-point, equal-appearing interval scale for describing any penetration and aspiration events. [16] By analyzing these two scales, quantitatively analyzing the deglutition function is possible. All VFSS evaluations were performed by two researchers in two groups who certi ed the modi ed barium swallow impairment pro le (MBSImP™).
Thus, our results for each group are mean scores by two researchers.
Statistical analysis SPSS 21.0 software (SPSS Inc, Chicago, IL, USA) was used for all statistical analyses. Because the number of patients is small, we used non-parametric statistics. The Wilcoxon signed-rank test was used to compare the pretreatment and posttreatment evaluations. The Mann-Whitney test was used to compare the differences of VFSS variables, VDS, and PAS variables between the 4-and 2-channel NMES groups. The results are presented as the mean ± standard deviation. p values of less than 0.05 were considered statistically signi cant.

Results
A total of 26 participants (13 for 4-channel groups and 13 for 2-channel groups) were initially enrolled; 1 participant in the 4-channel NMES group dropped out due to aggravated dizziness, and 2 participants dropped out due to too much sweating during NMES session and stoke aggravation. Therefore, 12 participants in the 4-channel NMES group and 11 participants in the 2-channel NMES group completed the clinical trial (Fig. 2).
The demographic data of the participants are presented in Table 1. The average ages of the 4-and 2channel NMES group were 64.9 ± 16.5 years and 60.6 ± 14.2 years, respectively. The disease durations of the 4-and 2-channel NMES groups were 49.2 ± 109.1 and 32.6 ± 24.8 days, respectively. Initial VDS scores of the 4-and 2-channel groups were 63.6 ± 15.10 and 51.3 ± 15.7 and initial PAS scores were 5.7 ± 2.2 and 4.4 ± 2.7, respectively. In all areas examined before we started the NMES, both groups of pre-clinical trial evaluations showed no signi cant differences (p > 0.05) (Tables 1, 2).  In the comparison between before and after treatment, oral, pharyngeal, and total VDS scores were signi cantly improved in both groups (p < 0.05, Fig. 3(A)). However, PAS and FOIS were signi cantly improved only in the 4-channel NMES group after treatment (p < 0.05, Fig. 3(B)). Although subsets of MADI (emotional, functional, physical subsets) were signi cantly improved in the 4-channel NMES group, only emotional and physical subsets were signi cantly improved in the 2-channel NMES group (Table 2) When we compared the changes of improvements between the two groups, oral VDS and total VDS score signi cantly improved in the 4-channel NMES group than in the 2-channel NMES group (p < 0.05). PAS and FOIS showed higher improvement in the 4-channel NMES group than in the 2-channel NMES group, but did not reach a signi cant level (p values of PAS and FOIS: 0.21 and 0.051, respectively). In MDADI, no signi cant difference was found between the two groups (Table 3).

Discussion
This is the rst randomized, double-blind, parallel group, controlled trial that directly compared the rehabilitative effect of sequential 4-channel NMES with conventional 2-channel NMES. In the present study, clinical improvement was observed via sequential 4-channel NMES and 2-channel NMES in VDS.
Only the sequential 4-channel NMES group showed signi cant improvement in PAS, FOIS, and subset of MDADI (functional) after treatment. When we directly compared the improvement between the two groups, the 4-channel group was superior to the 2-channel group in oral and total VDS.
PAS improvement in sequential 4-channel NMES is a very important nding. Previous studies on 2channel NMES showed con icting results. Some studies have proven effectiveness, [23,24] but others have shown no effect. [8,25] The reason might depend on the application site and methods of NMES. In a previous study, NMES on submental placement alone does not change the PAS and National Institutes of Health-swallowing safety scale (NIH-SSS). However, submental and throat placements show a signi cant improvement in the NIH-SSS, but PAS is unchanged. [9] Two-channel NMES improves pharyngeal peristalsis and cricopharyngeal functions at the esophageal entry but does not affect the elevation of the hyolaryngeal complex, which has correlation with aspiration or penetration. [9,13] In a previous study that used 4-channel NMES by connecting two sets of 2-channel NMES suggested that the sequential 4-channel NMES facilitates the movement of hyolaryngeal structures during swallowing.
[17] However, this system was made by combining two sets of 2-channel NMES; therefore, we could not adjust each channel. Thereby, we activated the rst two channels for 1400 ms, then 300 ms later, and the third and fourth channels were activated for 1100 ms. The biggest difference from the previous study is that we stimulated the third channel, which is attached to the thyrohyoid muscle in 150 ms after stimulation of the rst and second channels. A previous study evaluated during the effect, but the present study evaluated rehabilitative effect, which is commonly used for conventional 2-channel NMES.
The stimulation algorithm of sequential 4-channel NMES is based on normal contractile sequence. In the EMG analysis, the activations of the suprahyoid muscles developed about 150 ms and 350 ms earlier than those of the thyrohyoid and other infrahyoid muscles (sternohyoid and sternohyoid muscles). After 1400 ms of suprahyoid muscles' contraction, all of these muscles stop their contractions simultaneously.
[10] These sequential contractions of the suprahyoid and infrahyoid muscles accomplish the circular motion of the hyoid bone. The thyroid muscle assists laryngeal elevation, and other infrahyoid muscles, such as sternohyoid, sternothyroid, and omohyoid muscles, assist in prolonged laryngeal elevation and upper esophageal sphincter opening. [10,26] This concept was veri ed by using kinematic and pressure analyses in a previous study. [17] Thus, the contractions of the thyrohyoid and other infrahyoid muscles, which have proper interval time with those of the suprahyoid muscles, may be important for dysphagia treatment, and our results veri ed the improvement of laryngeal complex for aspiration or penetration. [10] The only muscle that elevates the larynx to the hyoid is the thyrohyoid muscle, which lies beneath the strap muscles, such as sternohyoid, sternothyroid, and omohyoid muscles. [9] In 2-channel NMES, simultaneous stimulation of submental and throat regions showed hyolaryngeal descent because sternohyoid and omohyoid stimulations exceeded the hyolaryngeal elevation effects. In other words, electrodes over the anterior neck might activate the sternohyoid and omohyoid rather than thyrohyoid and suprahyoid muscles. This hyolaryngeal descent is not a physiologic motion; therefore, the main mechanism of 2-channel NMES is strengthening of swallowing-related muscles. [27] However, theoretically, as 4-channel NMES uses normal contractile algorithm, the main mechanism of 4-channel NMES is not only strengthening of the supra and infrahyoid muscles but also increasing coordination of swallowing-related muscles.
In the present study, oral VDS and total VDS were signi cantly improved in the 4-channel NMES group compared with the 2-channel NMES group. Tongue-based pressure has been previously turned out to be important when the tongue comes in contact with the posterior pharyngeal wall, squeezing out the bolus through the pharynx. [28] Alterations in tongue coordination, strength, and pressure generation may result in a disruption of bolus movement from the oral cavity to the pharynx and result in increased risk of aspiration before or after swallowing. [29] Also, a previous study showed that a greater tongue strength results in a greater activation of the suprahyoid muscle during swallowing. [30] In the present study, one channel was attached to the suprahyoid muscles in 2-channel NMES, and the rst and second channels were attached to the bilateral suprahyoid muscles in 4-channel NMES. Considering that the effective depth of NMES is directly proportional to the distance between stimulation electrodes, [31] a wider placement of the coupled electrodes might more effectively stimulate not only the suprahyoid muscles but also the genioglossus or tongue muscles. Therefore, more channels and more covered lesion of the suprahyoid lesion in the sequential 4-channel NMES seem to induce effective contraction of the genioglossus and tongue muscles, which is highly correlated with tongue motion and is able to improve the strength of the tongue base, thereby resulting in superior e cacy in oral VDS scores.
In the present study, one participant in 2-channel NMES dropped out due to discomfort during NMES. In the 4-channel NMES group, aggravated dizziness was not related to NMES; therefore, no complication was developed. As sequential 4-channel NMES stimulated swallowing-related muscles with normal contractile algorithm, thereby inducing physiologic motion, less discomfort developed. Although the sample size was small and complication rate was negligible, 4-channel NMES is a safe and well-tolerated treatment method for dysphagia.
Originally, the purpose was to calculate the number of subjects in a future clinical trial to con rm the superiority of 4-channel NMES for future clinical trials. However, as 4-channel NMES sequentially stimulates swallowing-related muscles functionally and uses multi-channel for strengthening effect, our study veri ed the superiority of 4-channel NMES to 2-channel NMES with a small sample size. Follow-up study to verify the effectiveness of 4-channel NMES for pharyngeal phase is required.
This study has some limitations. First, a small number of subjects were included as this trial was a pilot study. As a result, pharyngeal VDS and FOIS were not signi cantly different between the two groups. However, the small number was enough to prove the superiority of 4-channel NMES in oral and total VDS.
A follow-up study is required to verify the signi cant improvement of pharyngeal VDS and FOIS. Second, treatment duration and daily sessions were different among the 5 rehabilitation units and general patients' condition. To manage this limitation, we did not use treatment frequency and time for the amount of NMES treatment, but used 420 min and 840 min as minimal and maximal treatment durations, respectively. Consent for publication