Dysphagia in the intensive care unit: epidemiology, mechanisms, and clinical management

Swallowing is a complex procedure involving more than 50 muscles, and a number of cranial nerves [36] (Fig. 1). Cortical structures involved particularly include the frontoparietal operculum, the primary sensorimotor cortex and association cortices and the anterior part of the insula. Cortico-bulbar projections then target the central pattern generators located in the dorsal medulla oblongata [37‐39], and the solitary nucleus and nucleus ambiguous coordinate swallowing [40‐42]. In an awake state, individuals swallow involuntarily more than once per minute. The frequency of swallowing is increasing to about > 5 times per minute during meals and decreased to about eight times per hour during sleep [43]. Four phases of swallowing can be differentiated [44]: (1) oral preparatory, (2) oral transit, (3) pharyngeal, and (4) the esophageal phase (Fig. 2). Bolus formation takes place in phase 1. In phase 2, the bolus is placed in the middle of the tongue, pushed against the hard palate and backwards to the oropharynx. When in contact with the palatoglossal arch, the swallowing reflex is triggered and the involuntary pharyngeal phase is initiated. The epipharynx is sealed and airway closure occurs in three different stages: (a) vocal cord adduction, (b) ventricular fold adduction, and (c) contact of arytenoid cartilages with the anteriorly tilted epiglottis. Importantly, active laryngeal elevation occurs which indirectly opens the upper esophageal sphincter. The bolus then enters the epiglottic valleculae and pyriforme sinuses. In the involuntary esophageal phase, peristaltic esophageal waves transport the bolus into the stomach (Fig. 2).

Oropharyngeal dysphagia in general can be caused by either (1) severe neurological impairment, affecting (a) the central nervous system directly (e.g., stroke, Morbus Parkinson, multiple sclerosis, or amyotrophic lateral sclerosis), (b) due to traumatic peripheral nerve damage and impaired function of the neuro-muscular junction, (c) primary neuro-muscular junction abnormalities (e.g., myasthenia gravis, Lambert-Eaton myasthenic syndrome [LEMS]), or (d) a primary muscular disease (e.g., inflammatory myopathies); (2) structural damage (e.g., trauma caused by the intubation or malignancies); (3) medication or toxic/drug side-effects; (4) presbyphagia; or (5) phagophobia [45].

In critically ill patients on the ICU, the etiology of dysphagia post-extubation appears less clear. PED is considered multifactorial with underlying mechanisms unknown, and the presence of an endotracheal tube/prolonged mechanical ventilation is considered a key risk factor for dysphagia. Six potential key mechanisms for the development of ICU-acquired swallowing disorders including PED were previously suggested [4]: (1) direct trauma caused by endotracheal and tracheostomy tubes, (2) neuromyopathy resulting in muscular weakness, (3) diminished laryngeal sensory function, (4) an impaired sensorium, reflecting a more centrally located problem, (5) gastroesophageal reflux, and (6) dyssynchronous breathing and swallowing.

In detail, direct trauma (1) seems an obvious and major mechanism in ICU-acquired swallowing dysfunction (Fig. 1). Artificial tubes of any kind, e.g., endotracheal, tracheostomy, echocardiography probes, or potentially feeding tubes, could directly cause trauma to anatomic structures. This may especially be relevant in emergency diagnostic or therapeutic interventions. Mechanical irritations of the underlying mucosal tissue may lead to focal ulceration and/or triggering of localized inflammatory processes. Scarring of the vocal cords was reported also, known as “vocal cord synechiae” [46]. Cuffed airway tubes per se prohibit normal swallowing function and active laryngeal elevation and subsequently reduce the passive opening of the upper esophageal sphincter which impedes rapid esophageal passage [45]. Two smaller studies showed contradicting results: One study [47] found no effect on penetration or aspiration by cuff status (inflated vs. deflated), but a significantly reduced score for the liquid bolus when a one-way valve was placed in comparison to in- or deflated cuff conditions (n = 14). Another study with a limited sample size (n = 7) [48] observed no significant alteration in hyoid bone movement and laryngeal excursion in cases when a tracheotomy tube was placed. Further, long-term intubation may lead to dislocation or even subluxation of arytenoid cartilages, resulting in an impaired glottis closure during swallowing [49]. Further, traumatic laryngoscopy was shown to lead to hypoglossal nerve palsy and to cause dysphagia [50‐53]. Peripheral damage of the recurrent laryngeal nerve, e.g., caused by tube cuff compression (or as a complication during surgery) can result in vocal cord paresis/paralysis and may prohibit competent airway protection.

Another relevant aspect is the presence of ICU acquired weakness (ICUAW) [54]. In critically ill patients with ICUAW, general muscular weakness and muscular atrophy was reported, which may affect the swallowing apparatus [54‐56]. ICUAW may be a consequence of “disuse” in patients receiving long-term intubation, long-term (analgo)-sedation, and/or neuromuscular blocking agents [28, 54, 57, 58]. Further, specific swallow-related muscular weakness was recently suggested in previously orally intubated acute respiratory distress syndrome patients (n = 11, median duration of intubation 14 days) after examination by videofluoroscopic swallowing study (VFSS) [59]. In addition, in ICU-acquired ventilator-induced diaphragmatic dysfunction (VIDD) [55], cough strength could be diminished leading to limited glottic clearance. Importantly, reduced local sensation appears as an additional key problem in ICU-acquired dysphagia. Either caused by direct mechanical damage, local inflammation/edema, or by critical illness polyneuropathy (CIP) [54, 56], afferent sensory pathways may be impaired leading to swallowing dysfunction [60‐63]. Clinically, this may become apparent when a bolus reaches the reflex trigger zone in the palatoglossal arch but the afferent input is impaired resulting in a delayed swallow response and pre-deglutitive aspiration. However, the exact role of sensory impairment in critically ill patients appears unclear. In a recent study, no nerve conduction abnormalities were demonstrated, questioning the role of CIP in dysphagia [27].

Central (cerebral) problems in ICU-acquired swallowing disorders are mostly caused by direct damage to the central nervous system, e.g., in traumatic brain injury, stroke/hemorrhage, and/or inflammatory disorders. Contributing to this, reduced qualitative (e.g., in delirium) or quantitative level of consciousness further increases the risk for aspiration [64] and may delay therapeutic measures for dysphagia. Moreover, drug-induced effects (e.g., (analgo-) sedatives or various neurotropic medications) may affect swallowing either centrally (mostly via reduced consciousness) or peripherally (mostly at the neuro-muscular junction). In this context, another potential mechanism was suggested [4], i.e., exact coordination of laryngeal closure, apnea, and opening of the upper esophageal sphincter may be impaired. In critically ill patients, this is referred to as “dyssynchrony” between respiration and swallowing [4]. Furthermore, in critically ill patients with respiratory distress, the apnoeic period during swallowing is shortened with potential premature opening of the larynx before the bolus has passed into the esophagus [65].

Different terms are used to assess dysphagia. In the 10th revision of the International Statistical Classification of Disease and Related Health Problems (ICD-10, WHO-Version 2016), dysphagia (R13) is listed in “Chapter XVIII: symptoms, signs and abnormal clinical and laboratory findings,” not elsewhere classified and more specified under R10–19 in “Symptoms and signs involving the digestive system and abdomen.” Dysphagia, swallowing disorder, or deglutition disorder/dysfunction are often used synonymously. In 2013, the term ICU-acquired swallowing disorder was introduced [4] suggesting multiple potential pathomechanisms in critical illness leading to acquired dysphagia in a previously dysphagia-naïve patient. International consensus on dysphagia definitions is lacking which may negatively impact on data comparability. We therefore recently proposed a delphi procedure with the aim to harmonize respective terminology [66].

A systematic review on the incidence of PED published in 2010 included a total of 14 studies with a total of 3520 individuals (mean of approximately 251 patients per study, median of 67) and concluded that the incidence rate ranges from 3 to 62% [7]. Study design, patient selection (e.g., assessment of patients post-aspiration), and/or limited patient numbers in respective included studies introduced a high risk of bias and showed reduced quality of evidence [1‐3, 5, 6, 67]. In a subsequent retrospective observational cohort study, a dysphagia prevalence of up to 84% was reported [67]. A recent larger study (DYnAMICS) performed by us included 1304 medical and surgical ICU patients with potential PED risk reported an incidence rate of 12.4% (18.3% in unselected emergency admissions) after systematic screening [8]. In DYnAMICS, the incidence was likely underestimated due to exclusion of patients leaving the ICU alive with tracheostomy (no extubation/decannulation) [8].

Risk factors for dysphagia might theoretically be inferred from the abovementioned pathomechanisms. However, studies focusing on risk factors for dysphagia following endotracheal intubation are scarce and provide conflicting results. Studies are mostly of limited sample size and either supporting or rejecting respective factors, including factors such as age [12, 19, 20, 68‐75], decreased cardiac output [19, 70], intubation duration [19, 20, 22, 32, 70, 71, 73, 74], postoperative pulmonary complications [70, 74], tube feeding [19, 22, 32, 74], sepsis [6, 32, 72], transesophageal echocardiogram (TEE) [70, 71], perioperative stroke [3, 32, 69‐71], or gastroesophageal reflux [1, 19, 76]. More consistently rejected as potential risk factors are APACHE II and SOFA scores [1, 3, 6, 9, 22, 67]; BMI [1, 6, 9]; gender [1, 6, 9, 32, 70, 75, 77, 78]; comorbidities such as arterial hypertension, kidney disease, diabetes, COPD, myocardial infarction, or heart failure [3, 19, 32, 67, 69, 70, 72], as well as smoking [32, 72]; and endotracheal tube size [3, 67, 75]. These contradicting results reflect bias due to patient selection, differing study/screening protocols, and limited patient numbers. However, despite controversial discussions, it appears that most presumed dysphagia risk factor would be duration of intubation/mechanical ventilation [1, 3, 4, 12, 32, 67, 68, 70, 78‐80]. In addition, rather accepted risk factors [4] may include the presence of pre-existing dysphagia, local malignancy/post-surgical medical conditions affecting anatomic structures of the swallowing tract, and/or considerable quantitative/qualitative reduction of consciousness. Overall, large-scale clinical data is missing and strongly warranted in order to potentially reduce the number of patients affected by preventional measures.

In stroke patients, early dysphagia detection minimizes the risk of aspiration [81] and systematic dysphagia screening reduces stroke-associated pneumonia rates [82]. This suggests that a systematic routine screening approach should be performed in all patients at risk without limitation to selected patient cohorts (e.g., stroke patients). Non-instrumental [66] and instrumental measures are available for timely assessment of dysphagia in the critically ill [83]. Non-instrumental assessments for dysphagia are typically performed by trained specialists (e.g., speech-language therapists, physiotherapists, or occupational therapists). The following clinical examinations were previously proposed for general populations of hospitalized (mostly non-ICU) patients: the bedside swallowing evaluation (BSE) [84], the Volume Viscosity Swallowing Test (V-VST) [85], the Mann Assessment of Swallowing Ability (MASA, K-MASA, MASA-C, MMASA) [86‐89], the McGill Ingestive Swallowing Assessment (MISA, MISA-DK) [90, 91], the Gugging Swallowing Screen (GUSS) [92], the Northwestern Dysphagia Patient Check Sheet (NDPCS) [93], the Dysphagia Disorder Survey (DDS) [94], the Practical Aspiration Screening Scheme (PASS) [95], the Kuchi-Kara Taberu Index (KT Index) [96], and the Practical Assessment of Dysphagia [97] test (reviewed in [66]).

Instrumental tests, such as the flexible endoscopic evaluation of swallowing (FEES) or the VFSS, may be regarded the gold standard of dysphagia assessment in the critically ill. FEES can be performed at the ICU bed using a small flexible endoscope passing through a nostril into the epipharynx so that the oro-/ hypopharynx and the glottic area can be visualized. Using a multicolor dye technique [98], testing of different food consistencies can be performed. Further, in selected patients (depending upon availability), sensation testing can be performed using short blasts of air to the supraglottic mucous membrane for assessment of vocal cord adduction, a technique known as flexible endoscopic evaluation of swallowing with sensory testing (FEESST) [99, 100]. Apart from research, FEESST was mainly abandoned and replaced by touching the aryepiglottic area with the endoscope tip, rendering the sensation normal, absent, or reduced. In FEES, severity of penetration or aspiration is assessed using a penetration and aspiration (PAS) scale (1 indicating no penetration and 8 indicating aspiration without coughing, i.e., silent aspiration) [101]. VFSS, also referred to as “modified barium swallow,” requires patient transfer to a radiology suite, which limits feasibility in larger cohorts of critically ill patients. Although exposure to radiation may be a disadvantage, VFSS investigates the entire swallowing act, i.e., all four stages of swallowing [102, 103]. Different barium-containing food consistencies can be visualized and recorded using high-resolution imaging devices. Intra-deglutitive aspiration can be visualized. This is not possible in FEES due to a “white out”-effect caused by velum elevation. Besides providing proof for the diagnosis, effects of compensatory maneuvers and diet modifications can be studies in a real-time manner using VFSS or FEES. Further “instrumental” methods include ultrasonography [104], tissue Doppler imaging [105], high-resolution manometry [106‐110], and oropharyngo-esophageal scintigraphy (OPES) [111]. Whereas manometry can be used to assess pharyngeal propulsion and upper esophageal sphincter performance, OPES allows detailed analysis of transit times and potential retention of a food bolus in the various anatomical areas. However, in critically ill patients post-extubation, this appears not feasible.

We recently proposed a feasible, pragmatic approach for systematic dysphagia assessment in the ICU [8]. This includes a two-step approach with systematic bedside screening for dysphagia by trained ICU nurses within few hours post-extubation, followed by an expert exam that is optimally complemented by a confirmatory FEES investigation [8].

In the critically ill, only few studies analyzed the impact of PED on patient-centered clinical outcomes. A recent retrospective study found an independent association of dysphagia with a composite endpoint of pneumonia, reintubation, or death [67]. In addition, dysphagia was associated with longer hospitalization, more discharges to a nursing home, and increased need for placement of a feeding tube [67, 77]. In a recent large prospective observational study (DYnAMICS) in a mixed population of 1304 critically ill patients with systematic bedside screening for dysphagia post-extubation, we observed an independent association of PED with 28-day and 90-day mortality after adjustment for typical confounders. An excess of 9.2% of a 90-day mortality rate was observed in patients with dysphagia [8].

In summary, the clinical consequences of dysphagia in the critically ill are important, with prolonged length of hospitalization, increased resource use, increased treatment costs, and increased mortality [8, 31, 67, 112]. Early identification of patients at risk seems warranted in an effort to minimize respective burdens.

In the light of the fact that dysphagia is not routinely screened for in critically ill patients on most ICUs, it appears that post-extubation dysphagia awareness should be increased [137, 138] and systematic bedside screening should be implemented on the ICU [8]. Further, after identification of patients with PED on the ICU, rehabilitation measures should be started. The promising results of recent randomized controlled trials assessing PES, rTMS (repetitive transcranial magnetic stimulation), and/or tDCS (transcranial direct current stimulation) (Fig. 3) pointed to positive effects in patients with severe post-stroke dysphagia. In the future, a novel therapeutic interventions using peripheral afferent approaches (e.g., via PES) and/or central efferent stimulation of pathways of the swallowing network may be of particular interest.