Oropharyngeal dysphagia belongs to the most dangerous symptoms of stroke affecting more than 50 % of patients in the acute stage [1]. Stroke-related dysphagia is a major cause of aspiration pneumonia, malnutrition, and dehydration and is associated with a prolonged hospital stay, increased mortality, and poor long-term outcome [2, 3]. Conventional treatment strategies include behavioral interventions, adaptation of food consistencies, and tube feeding [46]. Whereas these approaches all have their place in the therapeutic armamentarium, treatment of stroke-related dysphagia remains notoriously difficult. During the last few years novel neurostimulation approaches have been evaluated. Transcranial direct current stimulation (tDCS) and repetitive transcranial magnetic stimulation (rTMS) have been explored both in acute and postacute stroke with proof-of-principle evidence for their therapeutic potential [710]. Electrical pharyngeal stimulation (EPS) has been shown to improve swallowing function and in particular decrease airway aspiration in acute stroke patients [11]. Irrespective of the modality of neurostimulation applied, most of these studies focussed on mild to moderate stroke and included only a small proportion of severely dysphagic patients, i.e., patients being tube dependent without any oral intake.

In the present clinical trial we included severely affected, tracheotomized stroke patients, who after having been successfully weaned from the respirator could not be decannulated because of severe and persistent dysphagia. Our objective was to evaluate the therapeutic value of EPS vs sham stimulation for dysphagia treatment in these patients.

Patients and methods


Between June 2013 and August 2014 consecutive acute stroke patients from our neurological intensive care unit (ICU) at the University Hospital Muenster, who were completely weaned from the ventilator and able stay alert for at least 15 min, were screened for possible inclusion in this study. Patients were eligible to be included if they were tracheotomized and suffered from severe persistent dysphagia (as defined in Sect. “Dysphagia assessment”) rendering decannulation impossible. Exclusion criteria were pre-existing dysphagia or presence of implanted electronic devices of any kind. Patient recruitment was stopped once the target sample size was reached.

Data on age, sex, National Institutes of Health Stroke Scale (NIH-SS) on admission, type of stroke, site of stroke, stroke etiology classified according to the TOAST (Trial of Org 10172 in Acute Stroke Treatment) criteria, and vascular risk factors were obtained from the patient, the relatives, or the patient’s general practitioner. We also kept records on reason for intubation, acute stroke treatment (i.e., thrombolytic therapy, mechanical recanalization, neurosurgical intervention), time from orotracheal intubation to tracheotomy, total time of artificial ventilation, and time from stroke onset as well as termination of ventilation until study inclusion.

Informed consent was obtained from all patients or, in case the patient’s communication was impaired, their next of kin. The nature of the study was approved by the local ethics committee at the University of Muenster. The study was registered as a randomized controlled trial ( NCT01956175).

Dysphagia assessment

Patients were assessed by a trained neurologist together with a speech language pathologist in accordance with our protocol for standardized endoscopic swallowing evaluation for tracheotomy decannulation in critically ill neurologic patients, which we had previously developed and evaluated [12]. In brief, the patient’s management of secretions, spontaneous swallow frequency, laryngeal sensitivity, and cough were assessed in a stepwise manner. Next, patients were exposed to puree consistency and fluids. The patient was regarded as unsafe for decannulation and the procedure was stopped as soon as he or she failed at one single step of the protocol according to our decisional flowchart (see online supplementary file 1 for further details). Investigators were blinded to the patient’s study group allocation.

Electrical pharyngeal stimulation

Stimulation was delivered via the Phagenyx™ catheter system and base station (Phagenesis Ltd, UK). The system consists of a nasogastric feeding tube housing a pair of bipolar titanium ring electrodes with a distance of 10 mm in between. The electrodes were positioned in the middle pharynx. Correct positioning of the electrodes was visually confirmed by fiberoptic endoscopic evaluation of swallowing (FEES). The catheter was connected to the base station to deliver stimuli of 0.2 ms pulse duration at a frequency of 5 Hz with 280 V, which had previously been found to be the most effective stimulation parameters [13]. The current intensity (mA) was individually adjusted in every session. Therefore prior to the actual intervention the perceptual threshold (PT) and the maximum tolerated threshold (MTT) were determined repeatedly by slowly increasing the current. The average values of three trials were taken into account for the calculation of the optimal stimulation intensity according to the formula PT + 0.75 × (MTT − PT) [13]. Thresholds as well as calculated optimal stimulation intensities were documented at each session. In the treatment condition stimulation was afterwards delivered for a total of 10 min at this intensity, whereas in the sham condition the catheter was left connected to the base station for a further 10 min without current flow between the electrodes. The intervention was repeated daily for three consecutive days. The stimulation catheter remained in place over this period of time and was used as a regular feeding tube between treatment sessions.

Study protocol and endpoints

Patients showing persistent severe dysphagia on initial FEES assessment preventing decannulation and fulfilling the additional inclusion criteria were randomly assigned 2:1 to receive either EPS or sham stimulation using computer-assisted randomization. The randomization schedule was kept remotely from the study environment. The study coordinator provided assignment to the treating physician by phone. Primary endpoint was ability to decannulate the patient, facilitated by improved swallowing function based on FEES assessment (i.e., passing all steps of the protocol) after three consecutive days of stimulation. Patients that could not be decannulated after 3 days of sham stimulation were immediately offered another 3 days of real EPS during unblinded follow-up and were afterwards assessed by FEES again the next day (Fig. 1). Secondary outcome parameters were feeding status assessed at discharge with the Functional Oral Intake Scale (FOIS) [14], modified Rankin Scale (mRS), length of stay (LOS) on ICU/in the hospital, and time from stimulation to discharge. Decannulation was always performed immediately after the patient passed the FEES protocol and was as regarded successful if there was no need for reintubation until patient discharge.

Fig. 1
figure 1

Study protocol

Statistical analysis

From our clinical experience, we expected a spontaneous recovery rate of swallowing function allowing decannulation of around 20 % within the target time interval of our study. On the basis of promising previous results [11], we expected to be able to increase this rate to 60 % by EPS treatment. In a sample size calculation, n = 26 patients would yield a power of 80 % to detect a trend (α = 0.1) for an increase of 40 % in decannulation rate between the EPS and sham treatment group.

Descriptive statistics were used to quantify patient characteristics. The data are presented as frequencies for categorical variables and mean ± standard deviation for continuous variables. Categorical variables were tested using the Fisher exact test. Continuous variables were tested for normal distribution by applying Kolmogorov–Smirnov statistics. Normally distributed continuous variables were compared with the Student t test. The Mann–Whitney U test was used as an analogue non-parametric test. The analyses were carried out using SPSS Statistics 22.0 (IBM Corp., USA).


Fifty-one patients were assessed for possible inclusion in this study. Twenty-one patients were excluded: in 15 the tracheal cannula could be removed according to FEES, three patients were transferred to rehabilitation before possible start of the study, two had a cardiac pacemaker, and one declined to participate. Thirty patients were randomized 2:1 to receive either EPS or sham stimulation. All study participants were alert and able to communicate at least in a basic way. All recruited patients finished the study. One patient was transferred to rehab during unblinded EPS but could be followed up.

Adverse events

There were no EPS-related or non-device-related complications.

Patient characteristics

Table 1 shows that both study groups were well matched regarding demographic data and clinical patient characteristics. There was a trend to a more severe neurological deficit at study inclusion in the stimulation group only. Prestimulation times were slightly longer in this group.

Table 1 Baseline data of the stimulation and the control group

Treatment results

After EPS 15 out of 20 patients (75 %) of the stimulation group and two out of 10 patients (20 %) of the control group could be successfully decannulated within 72 h after finishing study treatment (p < 0.01). Of those eight patients from the sham group with persistent severe dysphagia, seven were included in the unblinded follow-up, whereupon decannulation became possible in five of them (71.4 %). One patient of the control group was transferred to the rehabilitation ward and hence could not be treated. There were no significant differences in secondary outcome parameters between the stimulation and the control group (Table 2). LOS variables as well as FOIS scale and mRS at discharge were comparable. There were no cases of decannulation failure in both groups.

Table 2 Secondary outcome data

Indicators of treatment success

When looking for variables associated with treatment success in the stimulation group only few differences were found between successfully and unsuccessfully treated patients (Table 3). There was a trend towards an older age in successfully treated patients (66.3 vs. 53.2 years) and the time interval “end of weaning to stimulation” was shorter in that group (148 vs. 326 h).

Table 3 Comparison of patient characteristics from the stimulation group with successful versus unsuccessful EPS

Calculated stimulation intensities did not differ significantly between treatment and control group. However, there was a significant decline of maximum tolerated threshold and calculated optimal stimulation intensities between the first and the second session in the treatment group that was not observed in the control group (Fig. 2a). When comparing patients with treatment success vs. failure within the stimulation group only, a significant decline of maximum tolerated threshold was found from session to session in successfully treated patients. Also the applied stimulation intensities went down significantly in the second session as opposed to the first. No significant changes were observed in patients with treatment failure (Fig. 2b).

Fig. 2
figure 2

Stimulation thresholds, a stimulation thresholds of the stimulation and control group; b stimulation thresholds of successfully and unsuccessfully treated patients of the stimulation group only


The main finding of this study was that EPS successfully treated dysphagia in severely affected, tracheotomized stroke patients, thereby enabling rapid decannulation in 75 % of the stimulation group. Furthermore, this success rate was nearly replicated during the unblinded follow-up treatment of seven patients of the control group. Although there were no clear predictors of treatment success, a decline in stimulation intensities over the three treatment episodes was related to remission of dysphagia.

Previously, EPS has been studied in a randomized trial in moderate to severely dysphagic stroke patients showing aspiration on an initial videofluoroscopic swallowing study (VFSS) [11]. Stroke patients in that study were older (mean age 75 years) and less severely affected (mean NIH-SS 10 points) than in the present trial. In keeping with our findings, Jayasekeran and co-workers reported a substantial improvement in swallowing function following EPS: the proportion of aspirative swallows on VFSS more than halved and a clinical swallowing score showed substantial improvement after EPS while both parameters remained essentially unchanged in the sham condition. Taken together, these two studies provide preliminary evidence that EPS may improve swallowing function in moderate to most severely affected stroke patients.

A tracheotomy is one of the most common surgical procedures on the intensive care unit (ICU) and is reported to be performed in 10–15 % of cases in mixed patient collectives [15]. In stroke patients treated on the ICU a tracheotomy seems to be performed even more frequently with numbers ranging from 15 to 35 % [16, 17]. Typical indications for a tracheotomy in general critical care are long-term ventilation due to prolonged respiratory failure, need for airway protection because of dysphagia with an increased aspiration risk, functional/mechanical obstruction, or prolonged need for endotracheal suctioning of secretions [15]. Although there is still some debate surrounding this topic, today a tracheotomy in ventilated ICU patients is supposed to prevent laryngeal and tracheal damage, to shorten the duration of mechanical ventilation, to reduce the length of stay on the ICU, and to cut down the related hospital costs [18, 19]. In spite of these beneficial aspects provided by the tracheal cannula during the acute stage of the illness, this device turns into a liability as soon as the patient has been successfully weaned from the respirator. The prolonged presence of a tracheotomy tube can delay rehabilitation, cause complications, reduces patient comfort, and is associated with longer hospitalization and higher costs [2023]. Even more important, several studies have shown that a tracheotomy tube in place at discharge from the ICU is predictive of a poor outcome in those patients. In particular, both Martinez et al. [24] and Clec’h et al. [25] reported significantly increased post-ICU mortality in this group of patients even after correcting for several confounders. This finding fuels the assumption that tracheotomized patients treated on non-ICU wards are particularly prone to suffer from cannula-related complications with potentially hazardous consequences. In summary, an efficient treatment option enabling timely decannulation would bring large benefit.

Severe dysphagia with related insufficient airway protection is given as the main reason why in successfully weaned patients the treatment goal of decannulation is not promptly achieved and patients need to remain tracheotomized [15]. Thus, depending on the case mix and the assessment method, disordered swallowing is found in 30–70 % of tracheotomized patients with the neurologically ill probably being the most vulnerable [2629]. As recently summarized by Macht et al. apart from direct damage to the swallowing network, several other mechanisms not restricted to neurocritical care may play a prominent role in the pathophysiology of ICU-acquired swallowing disorders [30]. First, endotracheal, tracheal, and nasogastric tubes themselves can cause direct trauma to anatomic structures involved in swallowing. Second, long-lasting ICU treatment may result in critical illness neuromyopathy impairing both the motor function of swallowing muscles and the feedback of sensory nerves. Third, an impaired sensorium, either due to effects of sedating medication or ICU-acquired delirium, also interferes with swallowing execution. Therefore the etiology of dysphagia in critically ill, tracheotomized stroke patients is complex. It involves both a disruption of the central swallowing network originating from the initial brain lesion and, as a result of the intensive care treatment and its complications, damage to different peripheral structures. In particular, sensory nerves providing indispensable feedback to coordinating centers are affected, but also motor nerves and the swallowing musculature itself, which are both immediately linked to the act of deglutition. Intriguingly, EPS may have effects on several neuroanatomical levels. Since the first studies in this field by Hamdy and co-workers it is known that EPS enhances the excitability and reorganization of the human pharyngeal motor cortex, thereby promoting the notion of stimulation-associated effects on the level of the central nervous system [31, 32]. While we assume that part of the stimulation effect observed in the present study is due to this mechanism, a second, more peripheral impact of EPS is also conceivable. We observed a decline of stimulation intensities in successfully treated patients of the stimulation group, whereas stimulation intensities slightly increased both in patients showing no improvement after EPS and in patients randomized to the sham group. Thus, we may speculate that EPS, apart from its central effects, also enhances restoration of peripheral sensory feedback finally leading to improved airway protection. Moreover we can state that treatment should start as soon as possible, as we found that the time from weaning to stimulation was shorter in successfully decannulated patients of the treatment group.

In this study, readiness for decannulation was used as primary endpoint. Evidence-based practical guidelines on when decannulation can safely be performed are lacking. Clinicians usually take into account a variety of parameters such as level of consciousness, cough effectiveness, ability to tolerate tracheotomy capping, amount of secretions, and comorbidities [33] all of which have recently been transformed into a formal albeit still subjective and so far untested score [34]. In the present study we therefore decided to employ an objective endoscopic decannulation protocol that used a stepwise algorithm to assess airway protection and swallowing safety [12]. Previously this protocol had been applied and validated against clinical decision-making in 100 neurocritical care patients. Fifty-four patients were decannulated after fulfilling endoscopic criteria and recannulation became necessary in only one of them. In keeping with these findings, there was no case of decannulation failure in the present study.

Today, tracheotomized patients are frequently treated by multidisciplinary teams consisting of specialized nurses, speech language pathologists, physiotherapists, ENT specialists, intensive care and respiratory physicians [35]. Efficacy of this approach has been evaluated in several studies invariably using historical controls as a comparator [36, 37]. In their recent meta-analysis of seven trials, Speed and Harding found low-quality evidence that this collaborative and holistic treatment provided by the multidisciplinary team may shorten total tracheotomy time with a mean reduction of 8 days. Apart from that, length of stay on the ICU might also be reduced and the use of speaking valves enhanced [38]. In the context of this less standardized treatment option the present study provides the first evidence that a short and well-defined intervention may reasonably help to achieve the ultimate goal of decannulation. However, as a result of the differences in study design and patient collectives a comparison of the respective efficacies is hardly possible.

Our study has several limitations. First, the number of included patients is rather low and all patients were recruited at one institution. Therefore the results should be replicated in a larger study employing a multicenter design. Second, although both study groups were well matched for all important parameters, standard deviations were partly large reflecting a substantial heterogeneity within the patient collective. Third, as a result of specific aspects of the study design, in particular the unblinding after three sessions of study treatment and the delayed treatment offered to patients of the control group, conclusions with regard to secondary outcome measures, which did not differ between groups, could not reliably been drawn in terms of treatment effect.

In conclusion pharyngeal stimulation was significantly associated with improvement of airway protection and remission of dysphagia, thereby enabling decannulation in the majority of patients in this randomized clinical trial.