The effects of noninvasive respiratory support on swallowing physiology, airway protection, and respiratory-swallow pattern in adults: A systematic review

Abstract

Purpose:

The use of noninvasive respiratory support– namely high flow of oxygen delivered via nasal cannula (HFNC), continuous positive airway pressure (CPAP), and noninvasive ventilation (NIV) – has been expanding in recent years. The physiologic mechanisms underlying each of these forms of support are generally well understood. In contrast, the effects on the sensorimotor mechanisms of swallowing movements, and of breathing and swallowing coordination – critical elements of airway protection and bolus clearance – remain unclear. The purpose of this systematic review is to assess the existing evidence about the impact of noninvasive respiratory support on swallowing mechanics, airway protection, and respiratory-swallowing patterns in adults.

Methods:

Six databases (PubMed, EMBASE, Web of Science, Scopus, CINAHL and ProQuest Dissertations & Theses) were searched using predetermined terms. Inclusion criteria were: 1) adult humans 2) use of noninvasive respiratory support, and 3) assessment of swallowing.

Results:

We identified 8727 articles for screening; 15 met the inclusion criteria. Six studies assessed noninvasive respiratory support in healthy adults, and 9 assessed participants with heterogenous respiratory diagnoses including chronic obstructive pulmonary disease (COPD), obstructive sleep apnea (OSA), acute respiratory failure, and chronic respiratory failure due to neuromuscular disease. Risk of bias was assessed using a modified NIH Quality Assessment Tool. In healthy adults, results demonstrated mixed effects of HFNC and CPAP on measures of swallowing function, airway protection, and respiratory swallowing patterns. Negative effects on respiratory-swallowing patterns were reported with NIV. In adults with heterogeneous respiratory diagnoses, six studies reported that HFNC, CPAP, or nasal NIV improved measures of swallowing and respiratory-swallowing patterns. HFNC has mixed effects on swallowing measures in ICU patients. NIV increased atypical respiratory-swallowing patterns in patients with stable COPD.

Conclusions:

Due to small sample sizes and the wide variation in study designs, the impact of noninvasive respiratory support on swallowing, airway protection, and respiratory-swallowing patterns cannot be confidently assessed based on the current evidence. Future studies using standardized, validated, and reproducible methods to assess the impact of noninvasive respiratory support on swallowing physiology and airway protection are warranted.

1. Introduction

The use of noninvasive respiratory support in the management of acute and chronic respiratory failure has increased in recent decades [1-4]. Notably, the global COVID-19 pandemic led to a surge in the number of hospitalizations associated with acute hypoxemic respiratory failure, highlighting the role of noninvasive respiratory support [5-9]. Noninvasive respiratory support modalities, namely continuous positive airway pressure (CPAP), noninvasive ventilation (NIV), and high nasal airflow oxygen delivery (HFNC), provide respiratory support using interfaces that cover the nose, or the nose and mouth [2,10,11].

In contrast to noninvasive respiratory support, invasive mechanical ventilation bypasses the upper airway using an artificial airway via endotracheal intubation or tracheostomy tube. Although invasive mechanical ventilation is highly effective in delivering respiratory support, its use is associated with risks and complications [12]. Complications specifically related to the swallowing mechanism result from the placement, presence, and removal process of an artificial airway [2]. These include laryngeal injury resulting in dysphagia [13-17], sensorimotor-related swallowing pathology [13,14,17-19], abnormal laryngeal penetration and aspiration, aspiration pneumonia [15,16,20], and prolonged use of alternate non-oral means of nutrition. In contrast, noninvasive respiratory support does not bypass the upper airway, thus preserving the physiologic functions of airway defenses and potentially allows patients to eat or drink by mouth, communicate verbally, and clear secretions [3,10,21].

The upper airway plays a central role in the movement of positive airflow pressure during the application of noninvasive respiratory support, while also intermittently accommodating the movement of foods and liquids through the aerodigestive tract during swallowing. The effects of positive airway pressure have been tested in animal models during breathing, and have been observed to influence the activity of the glottal muscles (i.e. thyroarytenoid and cricothyroid) [22], yet there is a paucity of similar studies in human subjects [23]. The impact, therefore, of noninvasive respiratory support on the sensorimotor mechanisms of swallowing movements and the coordination of breathing with swallowing remains unclear. This systematic review aims to assess the existing evidence on the potential impact of noninvasive respiratory support on swallowing physiology, airway protection, and respiratory-swallowing patterns in adults, to identify gaps in the literature, and to inform a clinical trial designed to empirically test the effects of noninvasive respiratory support on swallowing.

2. Methods

A preliminary review of an international prospective register of systematic reviews (PROSPERO) was reviewed to ensure that a similar review does not exist. Keyword search was completed using the following terms: dysphagia, deglutition, swallow*, Noninvasive Ventilation, NIV, High Flow Nasal Cannula, HFNC, artificial breathing, Bilevel Positive Airway Pressure, BiPAP, adult*, participant, individual*, people, or person*. This systematic review was registered in PROSPERO on May 17th, 2021. The Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement standards were utilized during the article selection process (Fig. 1) [24].

Fig. 1.

Preferred reporting items for systematic reviews and meta-analysis (PRISMA).

2.1. Eligibility criteria

We included articles that met the following inclusion criteria: a) any type of noninvasive respiratory support and b) assessed swallowing, airway protection, or respiratory-swallowing patterns. Randomized control trials, case-control studies, case series, and dissertations were included. We excluded articles that lacked measures of swallowing, assessed pediatric populations, utilized animal models, or were published in languages other than English.

2.2. Search strategy and information sources

We searched PubMed, EMBASE, Central, Web of Science, ProQuest Dissertations & Theses, Scopus, and CINAHL databases. Using predetermined keywords (Appendix A.2) in the selected databases, an initial search was completed on April 9th, 2021, with no date limit, except for ProQuest Dissertations & Theses, which was restricted to the last 10 years. Three subsequent literature searches in the same databases were performed to detect any newly published articles up until February 1st, 2024 (Fig. 1).

2.3. Selection and data collection process

Initial results of the search and repeat searches were screened using Rayyan QCRI software, an online systematic review manager for screening of articles [25]. The software was utilized to include or exclude articles at the title and abstract level. To ensure that no relevant articles were missed, using the 15 included articles, a backward literature search of the references was conducted utilizing SnowGlobe; an online tool using a technique called “snowballing” for systematic reviews and meta-analyses [26]. Interrater reliability was completed by the first and second authors (SR and RR) on 20 % of titles, abstracts, and full-text articles from the initial, and subsequent searches. Any disagreements between the two authors were discussed and resolved until complete consensus was achieved.

2.4. Data items

After the abstract screen, included articles were assessed for further eligibility using a brief code (Appendix A.3). This brief code specified inclusion criteria based on participant demographics, type of noninvasive respiratory support, objective or subjective swallowing assessments, airway protection, and respiratory-swallowing patterns outcome measures.

2.5. Risk of bias assessment

The included articles were evaluated for risk of bias using a modified version of the NIH Quality Assessment tool for Observational Cohort and Cross-Sectional Studies. This modified assessment tool analyzed study quality related to methodology, use of blinding, sample size justification, clarity of research questions, and specifications of the patient sample. Reliability of the risk of bias assessment was completed independently by the first and second authors for 20 % of articles. Disagreements were discussed and consensus was achieved.

3. Results

3.1. Study selection

We conducted four literature searches (Fig. 1) over a three-year period, identifying a total of 8727 articles for screening, with 87 full-text articles selected for further review. Of these, 73 were further excluded due to the lack of either a swallowing measure or the use of noninvasive respiratory support, resulting in 15 articles included in this systematic review.

3.2. Reliability

RS and RR achieved a minimum of 80 % inter-rater reliability across title, abstract, and full-text screening. Additionally, 100 % agreement was achieved on inclusion or exclusion decisions of articles.

3.3. Risk of bias assessment

Per our study design, all fifteen selected studies stated clear research questions and specified a targeted population. In twelve of fifteen (80 %) studies, there was lack of blinding [27-38]. Justification related to sample size was lacking in eleven of fifteen (73 %) studies [27-30,32,33,36-40].

3.4. Participant demographics

The total number of study participants from the 15 studies was 387 (range 2–80; mean of 25.8). The sex distribution of the study samples was 54.5 % (211/387) males and 45.5 % (176/387) females. The mean age of participants was reported in 12 of 15 studies and ranged from 18 to 96 years (Table 1).

Table 1.

Participant demographics and diagnoses.

Study	Sample n	Mean age (years) Age range (years)	Sex	Diagnosis
Allen & Galek, 2021 [41]	29	<60 years Mean and range n/a	Female: 23 Male: 6	Healthy
Eng et al., 2019 [40]	80	Mean age n/a Range 35–65	Female: 49 Male: 31	Healthy
Sanuki et al., 2017 [35]	9	Mean age 32.1 ± 5.9 Range n/a	Male: 9	Healthy
Arizono et al., 2021 [27]	30	Mean age 29.9 ± 6.7 Range n/a	Female: 19 Male: 11	Healthy
Hori et al., 2016 [30]	22	2 groups means: Young group: 28 ± 11.5 Elder group: 73.9 ± 5.8 Range n/a	Female: 10 Male: 12	Healthy
Nishino et al., 1989 [33]	8	Mean age n/a Range 28-48	Male: 8	Healthy
Hori et al., 2019 [31]	20	Mean age 75.8 ± 6.6 Range n/a	Male: 18 Female: 2	Stable COPD
Leder et al., 2016 [32]	50	Mean age 70 Range 27-96	Male: 16 Female: 34	ICU patients with acute hypoxemic respiratory failure
Flores et al., 2019 [28]	10	Mean age 71 Range 44-92	Male: 7 Female: 3	Heterogeneous respiratory diagnoses with HFNC requirements (e.g. post-op respiratory failure, tachycardia, COPD exacerbations, pneumonia, hypoxic respiratory failure, acute asthma exacerbation)
Terzi et al., 2014 [38]	15	Mean age 64.6 Range 48-82	Male: 10 Female: 5	COPD with acute exacerbation
Sato et al., 2011 [36]	10	Mean age 48 ± 11 Range n/a	Male: 10	Obstructive sleep apnea
Caparroz et al., 2019 [39]	70	Mean age 48.9 ± 11.2 Range n/a	Male: 49 Female: 21	Moderate obstructive sleep apnea
Sato et al., 2021 [37]	2	Mean age 68.5 Ages: 61 and 76	Male: 2	Severe obstructive sleep apnea with intractable pneumonia
Rattanajiajaroen & Kongpolprom, 2021 [34]	22	Mean age 56 ± 12 Range 18-80	Male: 15 Female: 7	Post-extubation patients within 48 h
Garguilo et al., 2016 [29]	10	Mean age 33 ± 15.2 Range 20-67	Male: 7 Female: 3	Stable severe restrictive respiratory failure due to a neuromuscular disease (e.g., Duchenne Muscular Dystrophy, Limb Griddle Muscular Dystrophy and Congenital Myasthenia)

3.5. Methods of measurement

3.5.1. Measures of noninvasive respiratory support

All studies assessed the impact of noninvasive respiratory support modalities of varying flow rates and pressures on either swallowing physiology [27,28,32,33,39-41], respiratory-swallowing patterns [29-31,36-38], or both [34,35]. In seven studies investigators assessed the effect of high flow nasal cannula (HFNC) [27,28,32,34,35,40,41], four assessed the effects of CPAP [33,36,37,39], two compared the effect of both CPAP and NIV [30,31], and two assessed nasal NIV equipped with a patient-controlled off-switch [29,38] (Table 2).

Table 2.

Noninvasive respiratory support types and settings, methods of PO presentations, and methods of signal acquisition.

Study	Sample n	Type of noninvasive respiratory support	Airflow, pressure, and O2 settings	PO presentation/task; Participants position	Measures of swallowing	Measures of respiratory- swallow patterns
		Healthy individuals
Arizono et al., 2021 [27]	30	HFNC (Optiflow™) (Fisher & Paykel Healthcare) Interface: nasal cannula	0, 10, 20, 30, 40, 50 LPM FiO2 at 0.21	30 mL water swallow test Repetitive saliva swallow Participant position: upright	Subjective observations of number of “choking occurrences” under each airflow condition.	n/a
Allen & Galek, 2021 [41]	29	HFNC (AIRVO™) (Fisher & Paykel Healthcare) Interface: nasal cannula	0, 10, 20, 30, 40, 50, 60 LPM FiO2 at 0.21 Temperature 37 °C	20 mL thin liquid trials only Participant position: upright	Videofluoroscopic Swallow Study (VFSS)	n/a
Eng et al., 2019 [40]	80	Adult Optiflow (Fisher & Paykel Healthcare) Interface: nasal cannula	0, 20, 40, 50 LPM HFNC system Fisher & Paykel RT 202. Heated humidification by Fisher & Paykel MR 850 FiO2 at 0.21	MBSImP protocol bolus consistencies Participant position: upright	Videofluoroscopic Swallow Study (VFSS)	n/a
Sanuki et al., 2017 [35]	9	HFNC AIRVO (Fisher & Paykel Healthcare) Interface: nasal cannula	15, 30, 45 LPM FiO2 at 0.21 Temperature 37 °C	5 mL water boluses infused over 3 s via a polyethylene catheter Participant position: supine	Submental Electromyogram (sEMG)	Respiratory Inductance Plethysmography (RIP)
Hori et al., 2016 [30]	22	CPAP and Bilevel PAP applied using NIV artificial ventilator (V60, Philips Respironics) Interface: full face mask	Control condition CPAP level (4 cmH2O) Bilevel PAP (IPAP 8 cmH2O, EPAP 4 cmH2O)	None. 5 series of repetitive saliva swallows under each condition Participant position: supine (head tilted upward at 30°)	Piezoelectric pressure sensor	Respiratory flow measured via nasal flow sensor cannula and a differential pressure transmitter, and piezoelectric pressure sensor
Nishino et al., 1989 [33]	8	CPAP Interface: nasal mask	0, 5, 10, 15 cmH2O CPAP	Thin polyethylene catheter inserted nasally into the pharynx. 0.5 mL water bolus injection and continuous infusion of 3 mL of water Participant position: supine	Submental electromyogram (sEMG)	Respiratory Inductance Plethysmography (RIP)
		Individuals with Respiratory Diseases
Rattanajiajaroen & Kongpolprom, 2021 [34]	22	HFNC (Optiflow™) (Fisher & Paykel Healthcare) Interface: nasal cannula	High airflow: 50 LPM Temperature of 34 °C FiO2 at 0.35 Low airflow: 5 LPM	Continuous infusion of 10-mL water for 1 min over 3 trials Participant position: n/a	Submental electromyogram (sEMG)	Electrocardiogram (EKG)-derived respiratory signals
Leder et al., 2016 [32]	50	HFO2-NC (Care Fusion Bird Air-Oxygen Blender, Viasys Healthcare) Interface: nasal cannula	Airflow range:10 to 50 LPM O2 requirements range 30–100 %	Bolus presentation: unspecified Participant position: n/a	Yale Swallow Screening Protocol and/or via FEES	n/a
Flores et al., 2019 [28]	10	HFNC (unspecified model) Interface: nasal cannula	Airflow range: 30 to 50 LPM FiO2 range: 35 %–99 %	MBSImP protocol bolus consistencies Participant position: n/a	Retrospective chart review of Videofluoroscopic Swallow Study (VFSS)	n/a
Hori et al., 2019 [31]	20	CPAP and Bilevel PAP applied using NIV artificial ventilator (V60, Philips Respironics) Interface: full face mask	Control condition CPAP level (4 cmH2O) Bilevel PAP (IPAP 8 cmH2O/EPAP 4 cmH2O) Airway pressure monitored using analog output from the V60 ventilator and recorded at 10 kHz	None. 5 series of repetitive saliva swallows under each condition. Participant position: supine (head tilted upward at 30°)	Piezoelectric pressure sensor	Respiratory flow measured via nasal flow sensor cannula and piezoelectric pressure sensor
Terzi et al., 2014 [38]	15	Two types used: 1 A standard mechanical ventilator delivering NIV with Bilevel PAP mode 2 NIV device equipped with an off-switch for use during swallowing Interface: nasal mask	Pressure support was titrated to obtain a tidal volume between 6 and 8 mL/kg of predicted body weight Flow trigger sensitivity was set low. (IPAP/PSV 13cmH2O/EPAP/PEEP 6cmH2O)a FiO2 ranged between 0.30 and 0.60	With and without NIV (conventual NIV or NIV with off-switch), five sets of water blouses (5 mL and 10 mL) administered via a syringe Participant position: upright	Electromyogram (EMG) and piezoelectric sensor	Respiratory Inductance Plethysmography (RIP)
Garguilo et al., 2016 [29]	10	NIV able to provide volumetric and assisted modes of ventilation (Elysee 150, ResMed SAS) equipped with an off-switch to allow patient-controlled ventilation deactivation Interface: nasal mask	The patient’s usual home mechanical ventilation settings were applied to the prototype ventilator (volume adjusted targeted ventilation without EPAP/PEEP)	With and without NIV (with off-switch), five sets of water boluses (5 mL and 10 mL) administered via a syringe 5 trials of 5 mL pudding presented (in absence of aspiration with water boluses) Participant position: upright	Electromyogram (EMG) and piezoelectric sensor	Respiratory Inductance Plethysmography (RIP)
Sato et al., 2011 [36]	10	CPAP Interface: unspecified	Unspecified settings	None. Saliva swallowing during sleep (under CPAP therapy) Participant position: sleep	Electromyogram (EMG)	Polysomnography-derived respiratory variables (nasal and oral airflow) and breathing efforts (chest and abdomen movements)
Caparroz et al., 2019 [39]	70	CPAP Interface: nasal mask	Unspecified CPAP set at automatic pressure mode	Consistencies were assessed (5 and 10 mL of thin liquid and thick liquid, puree and solids aFor thick liquid: 4.5 g thickener/100 mL water For puree: 9.0 g thickener/100 mL water) Participant position: n/a	Fiberoptic endoscopic evaluation of swallowing before and after CPAP therapy	n/a
Sato et al., 2021 [37]	2	CPAP Interface: unspecified	Titration performed and proper pressure were measured	None. Saliva swallowing during sleep before and under CPAP therapy Participant position: sleep	Electromyogram (EMG)	Polysomnography-derived respiratory variables (nasal and oral airflow) and breathing efforts (chest and abdomen movements)

^avalues were obtained from personal communication with the corresponding author

3.6. Measures of swallowing physiology and respiratory-swallowing pattern

3.6.1. Measures of swallowing

In thirteen studies, investigators assessed swallowing physiology using swallow movement relative to bolus flow (visual assessment), submental muscle activity, sensor-detected laryngeal motion and swallow behaviors, or observations of the presence or absence of signs of aspiration. Three studies used videofluoroscopic evaluation of swallowing [28,40,41] and two used fiberoptic endoscopic evaluation of swallowing (FEES) [32,39]. Surface electromyogram (sEMG) or piezoelectric sensors were used in nine studies [29-31,33-38], and one study used subjective observations of participants’ behaviors during swallow screening [27].

3.6.2. Measures of respiratory-swallowing pattern

In four of nine studies, investigators assessed respiratory-swallow patterns using respiratory inductance plethysmography [29,33,35,38]. In the remaining five studies [30,31,34,36,37], investigators monitored synchronized respiratory and swallowing movements using varied methods of signal acquisition, recording, and analysis (Table 2).

3.6.3. Bolus types and participants’ position during swallowing tasks

Variable bolus volumes, consistencies (solids and liquids), and methods of administration were used to assess swallowing and respiratory phase patterns across the fifteen studies. In two studies [28,40], investigators used the standardized and validated protocol of bolus administration as defined in the Modified Barium Swallow Impairment Profile (MBSImP^™) [42]. In thirteen studies, investigators used varied methods for bolus administration; in nine of those studies, investigators provided the rationale for bolus selection [27,29-31,34,35,38,39,41]. In the remaining studies, investigators did not provide a rationale for bolus selection or swallow tasks used [32,33] or assessed spontaneous saliva swallow in sleeping participants [36,37].

Investigators employed varied seating positions during bolus administration. Participants were in an upright position in five studies [27,29,38,40,41], and a supine position with or without an upward head tilt at 30° in four studies [30,31,33,35]. Participants’ positions were not specified in six studies [28,32,34,36,37,39] (Table 2).

3.6.4. Outcome measures of swallowing, airway protection, and respiratory-swallowing patterns

Swallowing outcome measures included impairment scores of swallowing physiology obtained from the MBSImP [28,40], varied temporal measures of swallowing [33,35,38], and descriptions of visual observations from fiberoptic endoscopic evaluation of swallowing (FEES) [32,39]. Airway protection measures included duration of laryngeal vestibular closure [41], penetration aspiration scale (PAS) scores [28,40,41,43], descriptions of airway visualization during FEES [32,39], and non-quantitative observations of overt signs of aspiration or penetration [27,29,31,34,36-38]. Respiratory-swallowing patterns measures only included the respiratory phase surrounding the swallow event [29-31,34-38] (Table 2).

3.6.5. Results on effects of noninvasive respiratory support in healthy adults on measures of a) swallowing physiology, b) airway protection, and c) respiratory-swallowing patterns

3.6.5.1. Swallowing physiology.

In two studies, increasing levels of HFNC rates (10–60 LPM; FiO2 0.21) resulted in increased duration of the laryngeal vestibular closure [41], and increasing levels of HFNC rates (15–45 LPM; FiO2 0.21) decreased the latency of the swallowing initiation [35]. In contrast, in two studies, the highest level of HFNC rate (60 LPM; FiO2 0.21) increased impairments in lip closure and oral residue on MBSImP [40], and high-level CPAP (15 cmH2O) resulted in an increased delay of pharyngeal swallow initiation [33].

3.6.5.2. Airway protection.

Decreased airway safety was reported in two studies with high-level HFNC (40 LPM; FiO2 0.21) [27] and high-level CPAP (15 cmH2O) [33]. However, two other studies reported no effects of HFNC (10–60 LPM; FiO2 0.21) on airway safety [40,41].

3.6.5.3. Respiratory-swallowing patterns.

Increased frequency of typical respiratory phase patterns surrounding the swallow (expiration after the swallow; SW-E) was reported during low-level CPAP (4 cmH2O) in one study [30]. One study reported no effects of HFNC (15–45 LPM; FiO2 0.21) on respiratory-swallowing phase pattern [35]. In contrast, an increase in atypical phase surrounding the swallow (inspiration after the swallow; SW-I) during Bilevel PAP (8 cmH2O/4 cmH2O) was reported in one study [30].

3.6.6. Results on effects of noninvasive respiratory support in individuals with respiratory diseases on measures of a) swallowing physiology, b) airway protection, and c) respiratory-swallowing patterns

3.6.6.1. Swallowing physiology.

In two observational studies, the use of HFNC (10–50 LPM; FiO2 30–100 % and 30–50 LPM; FiO2 35–99 %) along with patient-specific variables (e.g. cognitive status, physical abilities, and performance on an MBSS) influenced clinical decisions related to oral diet intake in patients with heterogeneous diagnoses and acute respiratory failure [28,32]. In another study, decreased delay (improvement) in the onset of swallowing initiation was reported in patients who received CPAP for obstructive sleep apnea [39].

3.6.6.2. Airway protection

One study reported no correlation between the effects of HFNC on airway protection in adults with hypoxic respiratory failure [32]. In contrast, a second study reported the occurrence of silent laryngeal penetration (PAS score 5) and aspiration (PAS score 8) using HFNC (30–50 LPM; FiO2 35–99 %) [28].

3.6.6.3. Respiratory-swallow patterns.

In six studies, an increased frequency of typical respiratory-swallowing phase patterns with noninvasive respiratory support was reported. The frequency of exhale-swallow-exhale increased with HFNC rates (50 LPM; FiO2 0.35) in post-extubation patients [34], during nasal NIV coupled with a patient-controlled off-switch (13 cmH2O/6 cmH2O) in patients with COPD exacerbation [38]. In addition, the frequency of expiration after the swallow (SW-E) increased with home mechanical ventilation in patients with severe restrictive lung disease [29], and with CPAP (variable pressures) in patients with stable COPD [31] and obstructive sleep apnea [36,37]. In contrast, investigators in one study reported an increase in the inspiratory phase after the swallow (SW-I) with the use of Bilevel PAP (8 cmH2O/4 cmH2O) in stable patients with COPD [31] (Table 3).

Table 3.

Effects on measures of swallowing, airway protection and respiratory-swallowing patterns.

Healthy individuals
Measure	Study	Noninvasive respiratory support	Results
Measures of swallowing
Modified Barium Swallow Study Impairment Profile (MBSImP)	Eng et al., 2019 [40]	HFNC 0, 20, 40, 60 LPM FiO2: 21 %	Higher (worse) MBSImP oral component scores at flow rates of 60 LPM compared to 0 LPM (p = 0.0024), 20 LPM (p = 0.0012), and 40 LPM (p = 0.0342). Tongue control during bolus hold and oral residue showed significant effects to flow rate; pharyngeal stripping wave and tongue base retraction were significant in older age group.
Temporal measures of swallowing	Sanuki et al., 2017 [35]	HFNC 0, 15, 30, 45 LPM FiO2: 21 %	Latency (in seconds) in swallowing initiation relative to pharyngeal bolus presentation was significantly and progressively shortened with increasing levels of HFNC from 15 LPM (p = 0.019), 30 LPM (p < 0.001), and 45 LPM (p < 0.001) compared to the control condition of 0 LPM.
	Nishino et al., 1989 [33]	CPAP 0, 5, 10, 15 cmH2O	Latency (in seconds) of swallowing initiation relative to pharyngeal bolus presentation was prolonged as CPAP increased in pressure.
Measures of airway protection
Duration of Laryngeal Vestibular Closure	Allen & Galek, 2021 [41]	HFNC 0, 10, 20, 30, 40, 50, 60 LPM FiO2: 21 %	Duration of laryngeal vestibular closure (dLVC), measured using Swallowtail Software, in healthy adults (<60 years) in response to varied airflow levels was assessed. A positive linear relationship was found between the duration of laryngeal vestibular closure in response to varying flow rates of airflow with thin liquids (20 mL) (p < 0.001). The dLVC increased by 0.002s as airflow increased by one unit, with greater variability in duration reported at higher levels of 50 LPM and 60 LPM.
Penetration-Aspiration measures	Arizono et al., 2021 [27]	HFNC 0, 10, 20, 30, 40, 50 LPM FiO2: 21 %	Subjective observations of clinical signs of airway invasion were reported with each airflow level. High airflow rates of 40 LPM were associated with increased occurrence of overt signs of airway invasion. Coughing and choking were reported to occur in 26 % (8/30) of healthy subjects (p < 0.5).
	Allen & Galek, 2021 [41]	HFNC 0, 10, 20, 30, 40, 50, 60 LPM FiO2: 21 %	Penetration-Aspiration Scale scores: not significantly associated with flow rate with occurrence of PAS 1, PAS 2, and PAS 4 reported in 99.2 % of total swallows.
	Eng et al., 2019 [40]	HFNC 0, 20, 40, 60 LPM FiO2: 21 %	Penetration-Aspiration Scale scores: not significantly associated with flow rate; occurrence of PAS scores of 2 in >80 % frequency.
Measures of respiratory-swallowing phase patterns
Respiratory phase surrounding swallowing	Hori et al., 2016 [30]	CPAP and Bilevel PAP (full face mask) Control condition CPAP level (4 cmH2O) Bilevel PAP (IPAP 8 cmH2O/EPAP 4 cmH2O).	The occurrence of swallowing followed by inhalation (S-I) was greater in participants receiving Bilevel PAP, compared to control and CPAP conditions, irrespective of age (p = 0.0557). Expiration after the swallow (S-E) increased with CPAP condition, irrespective of age.
	Sanuki et al., 2017 [35]	HFNC 0, 15, 30, 45 LPM FiO2: 21 %	The occurrence of swallow initiation relative to the phases of breathing was assessed in healthy participants under different HFNC flow conditions. No statistically significant differences were found in the phases of respiration in which swallowing occurred between any of the HFNC conditions (p = 0.409).
Individuals with respiratory diseases
Measures of swallowing
Modified Barium Swallow Study Impairment Profile (MBSImP)	Flores et al., 2019 [28]	HFNC HFNC Range: 30–50 LPM FiO2 range: 35 %–99 %	Descriptive data from retrospective chart reviews on PO diet pre and post MBSS; results of the MBSS were not provided. Before the MBSS, 9 of 10 patients were kept nil per os (NPO) based on bedside evaluation. After the MBSS, 9 of 10 were provided an oral diet with or without modifications, and 8 of 10 required liquid modifications (i.e., thickened liquids or no liquids). Six patients (75 %) tolerated the PO diet one week after initiation.
Fiberoptic Endoscopic Evaluation of Swallowing (FEES)	Leder et al., 2016 [32]	HFNC HFNC Range10-50 LPM HFNC average 22.6–30.5 LPM FiO2 range: 30–100 %	ICU patients on HFNC were screened using the Yale Swallow Protocol; if failed, instrumental swallowing evaluation using FEES to determine readiness for oral intake was completed. Descriptive data on swallowing physiology were not provided. Patients able to resume oral diet based on screening: 39/50 patients (78 %). Patients required instrumental evaluation using FEES and required diet modifications (i.e., thickened liquids): 5/39. Patients remained NPO due to severe respiratory issues and were not appropriate for swallowing evaluation or PO intake: 11/50 (22 %).
	Caparroz et al., 2019 [39]	CPAP with a nasal mask automatic pressure mode	Baseline evaluation using FEES to determine: 1) premature spillage, 2) velopharyngeal dysfunction, 3) laryngeal penetration, 4) tracheal aspiration, and 5) presence of residue after 3 swallows. Premature spillage is operationally defined as the bolus falling over the base of the tongue or lower before whiteout in FEES. Patients presented with dysphagia on FEES with a primary deficit of premature spillage: 18/70 (27.3 %). Patients were prescribed CPAP therapy for 3 months and underwent FEES following completion of CPAP therapy: 11/18. Patients demonstrated improvement of dysphagia, specifically in premature spillage scores, with the improvement in OSA: 9/11.
Temporal measures of swallowing
	Terzi et al., 2014 [38]	A standard mechanical ventilator delivering NIV and NIV equipped with off-switch (13 cmH2O/6 cmH2O)a FiO2 range: 30–60 %	Swallowing was assessed under two conditions: spontaneous breathing compared to conventional NIV, and spontaneous breathing compared to NIV equipped with an off-switch button that participants activated during swallowing. Swallowing duration decreased under NIV conditions with a statistical significance of (p = 0.016) irrespective of bolus sizes (5 mL or 10 mL).
Measures of airway protection
Penetration-Aspiration measures	Flores et al., 2019 [28]	HFNC HFNC Range: 30–50 LPM FiO2 range: 35–99 %	Using penetration-aspiration scales (PAS), 50 % (5/10) patients demonstrated silent laryngeal penetration or aspiration during the videofluoroscopic swallowing study. No statistical significance in the effects of HFNC due to the small sample size.
	Leder et al., 2016 [32]	HFNC HFNC Range: 10–50 LPM Average flow: 22.6–30.5 LP FiO2 range: 30–100 %	Using FEES observations, the presence or absence of aspiration during FEES exam was reported as a measure of oral intake readiness. 39/50 patients were deemed appropriate for PO intake based on bedside performance or FEES results.
	Caparroz et al., 2019 [39] (Garguilo et al., 2016; Rattanajiajaroen & Kongpolprom, 2021; Terzi et al., 2014; Hor et al., 2016; 2019; Sato et al., 2011, 2021) [29-31,34,36-38].	CPAP Unspecified-with a nasal mask automatic pressure mode Subjective observations i	Using FEES observations, the presence or absence of laryngeal penetration or tracheal aspiration were reported from FEES exam. Subjective observations of clinical signs of aspiration were reported
Measures of respiratory-swallowing phase patterns
	Garguilo et al., 2016 [29]	NIV equipped with patient-controlled off-switch Unspecified pressure.	Swallowing of two bolus types (thin liquid and pudding) was assessed under two conditions: spontaneous breathing and with NIV (patient0controlled off switch to allow patients to withhold ventilation as desired). In spontaneous breathing, the percentage of swallows followed by inspiration (SW-I) was above 40 %. With NIV use, significant decrease of SW-I below 21 % (p < 0.0001). Swallows occurred in mid-expiratory phase in all conditions.
	Terzi et al., 2014 [38]	Conventional NIV and NIV equipped with patient-controlled off-switch (13 cmH2O/6 cmH2O)a	In conventional NIV compared to spontaneous breathing condition: The percentage of swallows followed by inspiration pattern (S-I) showed a significant decrease in the NIV condition (p = 0.0003). Swallowing-induced NIV auto triggering, patient/ventilator asynchrony and resulted in inhalation post the swallow. In NIV with off-switch compared to spontaneous breathing: no swallowing-induced ventilator-triggering, but prolonged expiration occurred. The main pattern reported is exhale-swallow-exhale in 88.7 % ± 19.4 % of the swallows, and the remaining pattern was inhale-swallow-exhale.
	Hori et al., 2019 [31]	CPAP and Bilevel PAP (full face mask) Control condition CPAP level (4 cmH2O) Bilevel PAP (8 cmH2O/4 cmH2O)	CPAP significantly decreased the frequency of swallows followed by inspiration (S-I) and increased the frequency of swallows followed by expiration (S-E). CPAP improved timing of swallowing during the respiratory cycle (from early to intermediate phase). Bilevel PAP and control conditions increased the frequency of swallowing followed by inspiration (S-I) (p < 0.01). Durations of the pause in CPAP and Bilevel PAP conditions were consistently shorter compared to control condition (p < 0.01).
	Rattanajoajaroen et al., 2021 [34]	HFNC 50 LPM and 5 LPM FiO2: 35 % Temperature 34 °C	High airflow (50 LPM) resulted in a higher percentage of exhale-swallow-exhale pattern (74 %) and a lower percentage of an inhale-swallow-inhale pattern (14.3 %) with a statistical significance of (p = 0.048).
	Sato et al., 2011 [36]	CPAP Unspecified	During CPAP therapy: Swallowing followed by expiration (S-E) increased from 34.3 ± 14.9% to 41.8 ± 28.8 % compared to baseline. Swallows followed by inspiration decreased from 44.5 ± 15 %to 11.8 ± 12.8% compared to baseline.
	Sato et al., 2021 [37]	CPAP Unspecified Titration performed and proper pressure were measured	Before CPAP therapy, patients demonstrated inhale-swallow-inhale (I-S-I) pattern in 60%–70 % of swallows. During CPAP therapy, swallows occurred immediately after expiration (E-S) in 92–94 % and was followed by expiration (S-E) in 81–87 % of the swallows.

^avalues were obtained from personal communication with the corresponding author

4. Discussion

The results of the 15 studies included in this systematic review are equivocal regarding the impact of noninvasive respiratory support on swallowing, airway protection, and respiratory swallowing patterns. Conflicting findings are likely due to variability in study designs, wide variation in methods used for swallow and respiratory signal acquisition, analysis, reporting, underlying pathology (or no pathology) in study subjects, and small sample sizes. As such, the impact of noninvasive respiratory support on swallowing, airway protection, and respiratory swallowing patterns cannot be confidently assessed based on the current evidence. It is important to note that relevant articles have been published between the time of our analysis and submission and these were not included in this review [44-47].

4.1. Variation in methods for observing swallowing and airway protection

Studies included in this review utilized variable methods to assess swallowing physiology. In five of 15, investigators used instrumental methods to objectively assess swallowing physiology. In three studies investigators used videofluoroscopic swallowing study (VFSS) [44] and in two studies, flexible endoscopic evaluation of swallowing (FEES) was used [32,39]. These imaging methods allow for some direct or indirect visualization of swallowing movements and bolus flow and provide specific information regarding functional anatomy, physiology, and airway safety [48-50]. In most of the studies (9 of 15) investigators used EMG and piezoelectric sensors for obtaining swallowing data, methods that use electrical signals as surrogates for muscle activity, and laryngeal movements to assess swallowing [29-31,33-38]. Additionally, swallowing screening assessments were used in one study including the 30 mL water swallow test (WST) and repetitive saliva swallow test (RSST) [27]. Although both methods are used clinically to determine the risk and presence of clinical signs suggestive of aspiration, diagnostic information related to swallowing physiology cannot be obtained.

Similarly, measures of airway protection varied across studies. In five studies, validated PASscale for rating airway safety from VFSS [51] [28,40,41] or FEES [32,39] were used. However, in the remaining studies, clinical signs suggestive of aspiration were reported [27,29,31,34,36-38], thereby limiting the conclusions drawn regarding the impact of noninvasive respiratory support on airway safety.

4.2. Variation in bolus administration protocol and participant positions

Bolus administration methods differed among studies included in this review. Only two studies [28,40] used the standardized and validated protocol of MBSImP which includes 12 swallowing tasks with various bolus consistencies, volumes, and presentation methods [42]. Similarly, participant positions during testing varied across studies. In two studies, investigators assessed swallowing in patients with obstructive sleep apnea while they received CPAP during sleep. In four other studies, awake participants were supine during swallowing tasks. The supine position is not a natural body position optimal for eating or swallowing and may contribute to changes in the timing and pressures of swallowing movements [52].

4.3. Variation in methods for obtaining and analyzing respiratory-swallow phase patterns

Variable methods for obtaining and analyzing respiratory signals limit the comparisons of the impact of noninvasive respiratory support on respiratory-swallowing phase patterns across studies. Respiratory Inductance Plethysmography (RIP), the reference standard method for obtaining respiratory signals [53] was used in four studies [29,33,35,38]. Methods utilized in the remaining studies varied and included piezoelectric sensors [30,31], an electrocardiogram with EMG [34], polysomnography coupled with nasal airflow, and surface EMG [36,37].

4.4. Type of noninvasive respiratory support and settings

Variable forms and intensities of noninvasive respiratory support were used across studies. In healthy subjects, settings of HFNC were consistent across studies (range 10–60 LPM, 0.21 FiO2) [27,35,40,41]. However, PAP levels differed in two studies (4cmH2O vs 5-10cmH2O) [30,33]. Inconsistencies and variable settings of noninvasive respiratory support were observed across studies of participants with respiratory diseases. We reason this can be due to the heterogeneity of respiratory diagnoses included in the studies and the level of support required to meet patients’ respiratory needs.

4.5. Hypotheses on the impact of noninvasive respiratory support on swallowing, airway protection, and respiratory-swallowing patterns

The lack of consistency in the methods used to assess swallowing and respiratory signals limits direct comparisons of results and conclusions about the effects of noninvasive respiratory support on swallowing physiology and respiratory-swallowing phase patterns. Moreover, each type of noninvasive respiratory support differs in its effects on the upper and lower airways. Additionally, high flow nasal oxygen may have composite complex effects on the respiratory system due to the multiple mechanistic effects including positive airway pressure generation, dead space washout, and the Venturi effect on oxygen delivery. As such, we suspect that the effects on the swallow mechanisms would be variable as well. Finally, understanding the disease states and any possible underlying swallowing impairment of patients is vitally important when assessing the impact of these respiratory devices. For example, HFNC in healthy subjects can lead to discomfort and may influence the impact on swallowing [54], whereas HFNC is favored, precisely because of comfort and increased compliance, over NIV in patients with COPD post-extubation [55]. Further, different noninvasive respiratory support devices will likely exert different effects on swallowing function and its coordination with respiration. Thus, we propose two directional hypotheses of possible mechanisms contributing to either positive or negative effects of noninvasive respiratory support on swallowing physiology, airway protection, and swallowing coordination with respiration.

Hypothesis 1.

The effects of noninvasive respiratory support on the activity of upper airway muscles and on pulmonary mechanics may result in optimized respiratory-swallowing patterns and improved airway closure. The biologic plausibility of this hypothesis is based on two observations. First, positive airway pressure has been associated with physiological benefits including unloading respiratory muscles [56], and lowering respiratory rate via increased resistance to expiratory flow which induces longer expiratory time [57]. Second, positive airway pressure influences the activity of the upper airway muscles including the glottal constrictor (thyroarytenoid) and dilator (cricothyroid) in awake lambs and healthy subjects [22,23]. Increasing levels of positive airway pressure results in narrowing of the glottis and a reduction in tidal volume [58]. Together, these mechanisms may increase airway closure and the likelihood that swallowing occurs during the expiratory cycle, a pattern that has been shown to have biomechanical advantages to swallowing and airway protection [59,60].

Hypothesis 2.

Factors such as patient-device asynchrony or discomfort associated with noninvasive respiratory support may result in negative effects on swallowing physiology and respiratory-swallowing patterns. The biologic plausibility of this hypothesis is based on two observations. First, the ability to synchronize breathing with the respiratory devices is essential for adequate ventilation. Asynchronous interactions often result in adverse effects, such as increased work of breathing, increased respiratory rate, and inappropriate triggering of assisted breaths [38]. In addition, during bilevel PAP, swallowing produces a slight negative upper-airway pressure which triggers inspiratory support [30,31,38]. These episodes of inspiratory support can promote suboptimal respiratory-swallowing patterns with swallowing followed by inspirations enhancing the risk for aspiration [38]. Second, respiratory discomfort that can be associated with positive pressure may induce anxiety, stress, and pain which can affect the ventilatory pattern and result in an increased respiratory rate [56]. We reason that together, these factors negatively impact respiratory-swallowing patterns resulting in increased risks of airway invasion and reduced swallowing efficiency.

4.6. Limitations

This review followed a systematic approach to the literature search with specific search terms; however, it is possible that some studies have been missed. Additionally, there were differences in study designs, including but not limited to the variability in study measurements, bolus types, the heterogeneous nature of the patients and subjects examined, and the differences in the type of noninvasive respiratory support tested, all of which lead to conflicting findings. This heterogeneity prevented the possibility of conducting a meta-analysis and precluded the ability to draw meaningful conclusions about the effect of noninvasive respiratory support or its impact on swallowing safety during oral intake when working with patients in clinical practice. Lastly, given the variability and the limited findings, a scoping review or alternative review methodologies may be more appropriate approaches to map the available evidence and identify knowledge gaps.

5. Conclusions

This systematic review reports there exists a wide variability across studies reporting the effects of noninvasive respiratory support on measures of swallowing physiology, airway safety, and respiratory-swallowing patterns. Findings suggest mixed results about the impact of noninvasive respiratory support on measures of swallowing physiology and respiratory-swallowing patterns in healthy individuals and in patients with various respiratory diseases. Pilot research studies and subsequent randomized clinical trials that use valid, reproducible, and reliable data acquisition and measurement methods of swallowing physiology are needed to determine the impact of noninvasive respiratory support modalities and settings on airway protection and bolus clearance mechanisms during swallowing. This information is crucial to inform decisions about the safety of initiating or maintaining oral intake for patients receiving these respiratory treatments.

Appendices.

Table A.1.

Summary of cut-off noninvasive respiratory support levels impacting swallowing and respiratory-swallow patterns reported in the studies

Study	Noninvasive respiratory support type & levels	Cut-off LPM or cmH2O levels & reported impact
Healthy individuals
Arizono et al., 2021 [27]	HFNC 0, 10, 20, 30, 40, 50 LPM FiO2 at 0.21	Flow rates ≥40 LPM Effects: choking, increased effort, decreased number of swallows
Allen & Galek, 2020 [41]	HFNC 0, 10, 20, 30, 40, 50, 60 LPM FiO2 of 0.21	Flow rate of 50 LPM Effects: For every unit increase of airflow, duration of LVC increased by 0.002s PAS unchanged Increased difficulty (effort) swallowing at 50–60 LPM
Eng et al., 2019 [40]	HFNC 0, 20, 40, 60 LPM FiO2: 0.21	Flow rate of 60 LPM Effects: Worse MBSImP scores: oral bolus hold, oral residue, pharyngeal stripping wave, BOT retraction
Sanuki et al., 2017 [35]	HFNC 0, 15, 30, 45 LPM FiO2: 0.21 Temperature 37 °C	Flow rate of 15 LPM Effects: latency of swallow imitation shortened as airflow increased Unchanged number of swallows and respiratory-swallow patterns
Hori et al., 2016 [30]	CPAP and Bilevel PAP (full face mask) Control condition CPAP level (4 cmH2O) Bilevel PAP (IPAP 8 cmH2O/EPAP 4 cmH2O).	CPAP level (4 cmH2O) and Bilevel PAP (IPAP 8 cmH2O/EPAP 4 cmH2O). Effects: SW-I frequency increased with Bilevel PAP SW-E frequency increased with CPAP
Nishino et al., 1989 [33]	CPAP 0, 5, 10, 15 cmH2O	CPAP 15 cmH2O Effects: progressive increase in the latency of the swallow and decreased number of swallows with increased in CPAP pressure
Individuals with respiratory diseases
Rattanajoajaroen et al., 2021 [34]	HFNC 50 LPM and 5 LPM FiO2: 0.35 Temperature 34 °C	Flow rate of 50 LPM Effects: the percentage of E-Swallow-E pattern increased, and I-Swallow-I decreased.
Leder et al., 2016 [32]	HFNC Range:10–50 LPM Average flow: 22.6–30.5 LPM FiO2 range: 30–100 %	Unspecified cut-off NPO patients in the study were on airflow levels of a mean of 30.5 LPM
Flores et al., 2019 [28]	HFNC Range: 30–50 LPM FiO2 range: 35–99 %	Unspecified cut-off Effects: silent laryngeal penetration and aspiration noted, but no association could be drawn due to small sample size
Hori et al., 2019 [31]	CPAP and Bilevel PAP (full face mask) Control condition CPAP level (4 cmH2O) Bilevel PAP (8 cmH2O/4 cmH2O)	CPAP level (4 cmH2O) Effects: increased swallowing followed by expiration (SW-E) Bilevel PAP (8 cmH2o/4 cmH2O) Effects: increased swallowing followed by inspiration (SW-I)
Terzi et al., 2014 [38]	A standard mechanical ventilator delivering NIV (nasal mask) (13 cmH2O/6 cmH2O)* FiO2 range: 30–60 %	Effects: The percentage of swallow followed by inspiration decreased with NIV
Garguilo et al., 2016 [29]	NIV Life support ventilator (volume adjusted targeted ventilation/no EPAP)	Unspecified level Effects: The frequency of swallowing followed by expiration (SW-E) increased Improved dyspnea during swallowing
Sato et al., 2011 [36]	CPAP Unspecified	Unspecified level Effects: frequency of swallows followed by inspiration decreased
Caparroz et al., 2019 [39]	CPAP Unspecified-with a nasal mask automatic pressure mode	Unspecified level Effects: Improved timing of the pharyngeal swallow and residue scores on FEES
Sato et al., 2021 [37]	CPAP Unspecified Titration performed and proper pressure were measured	Unspecified level Effects: frequency of swallows followed by expiration increased.

A.2. Search key words:

The search strategy using predetermined keywords conducted in the selected databases was: “Swallow* OR Dysphagia OR deglutition” AND “ventilation” OR “nasal cannula” OR “High flow nasal cannula” OR “HFNC” OR “artificial breathing” OR “BiPAP” OR “Bilevel Positive Airway Pressure AND adult* or participant* or subject* or individual* or people or person*. The search chart string used in Scopus was: (ALL (swallow* OR dysphagia OR deglutition) AND ALL (“ventilation” OR “nasal cannula” OR “High flow nasal cannula” OR “HFNC” OR “artificial breathing” OR “BiPAP” OR “bilevel AND positive AND airway AND pressure”) AND ALL (adult* OR participant* OR subject* OR individual* OR people OR person.

Table A.3.

Brief code used for full-text screening

“NIV”, “HFNC”, “CPAP”, “BiPAP” mentioned	Participants: patient vs healthy	Swallow valuation instrumental or bedside	Respiratory-swallow evaluation	Outcome measures used
Yes (1)	Patients (1)	Swallow evaluation present (1)	R-S evaluation present (1)	Swallow physiology, swallowing behaviors, respiratory-swallowing coordination, respiratory phase surrounding the swallow present (1)
No (0)	Healthy (0)	Swallow evaluation absent (0)	R-S evaluation absent (0)	Swallow physiology, swallowing behaviors, respiratory-swallowing coordination, respiratory phase surrounding the swallow absent (0)

A.4. Items used for full data extraction

Using eligible studies, a full data extraction codebook using these predetermined outcome measures was completed. The following items were used to extract data from the studies.

1. Information related to author(s) names, publication year, publication journal, and country in which the study was conducted

2. Participant demographics (number of participants, mean age and standard deviation, sex distribution of the sample, and inclusion/exclusion criteria)

3. Diagnoses of the sample participants (healthy vs respiratory conditions)

4. Presence of dysphagia or swallowing difficulty

5. Descriptive information related to noninvasive respiratory support (type of device, pressure used, airflow level, FiO2, and temperature)

6. Outcome measures used in the study (swallowing and respiratory measures)