Post‐extubation dysphagia and dysphonia amongst adults with COVID‐19 in the Republic of Ireland: A prospective multi‐site observational cohort study

Abstract Objectives This study aims to (i) investigate post‐extubation dysphagia and dysphonia amongst adults intubated with SARS‐COV‐2 (COVID‐19) and referred to speech and language therapy (SLT) in acute hospitals across the Republic of Ireland (ROI) between March and June 2020; (ii) identify variables predictive of post‐extubation oral intake status and dysphonia and (iii) establish SLT rehabilitation needs and services provided to this cohort. Design A multi‐site prospective observational cohort study. Participants One hundred adults with confirmed COVID‐19 who were intubated across eleven acute hospital sites in ROI and who were referred to SLT services between March and June 2020 inclusive. Main Outcome Measures Oral intake status, level of diet modification and perceptual voice quality. Results Based on initial SLT assessment, 90% required altered oral intake and 59% required tube feeding with 36% not allowed oral intake. Age (OR 1.064; 95% CI 1.018–1.112), proning (OR 3.671; 95% CI 1.128–11.943) and pre‐existing respiratory disease (OR 5.863; 95% CI 1.521–11.599) were predictors of oral intake status post‐extubation. Two‐thirds (66%) presented with dysphonia post‐extubation. Intubation injury (OR 10.471; 95% CI 1.060–103.466) and pre‐existing respiratory disease (OR 24.196; 95% CI 1.609–363.78) were predictors of post‐extubation voice quality. Thirty‐seven per cent required dysphagia intervention post‐extubation, whereas 20% needed intervention for voice. Dysphagia and dysphonia persisted in 27% and 37% cases, respectively, at hospital discharge. Discussion Post‐extubation dysphagia and dysphonia were prevalent amongst adults with COVID‐19 across the ROI. Predictors included iatrogenic factors and underlying respiratory disease. Prompt evaluation and intervention is needed to minimise complications and inform rehabilitation planning.


| BACKG ROU N D
The SARS-CoV-2 virus (termed COVID-19) is a novel respiratory virus, which has led to an international pandemic. COVID-19 has resulted in an unprecedented number of critically ill adults, which has overwhelmed intensive care unit (ICU) services worldwide.
Endotracheal intubation and mechanical ventilation have been a central management procedure for critically ill patients with  in ICU settings.
Post-extubation dysphagia (PED) and dysphonia are common in critical care patients, 1,2 and recent research has highlighted PED and dysphonia within the COVID-19 population. 3,4 In a single-centre observational cohort study, 79% of adults referred to SLT over a twomonth period during COVID-19 had been intubated and all patients who had persistent dysphagia at discharge had been intubated. 3 In another single-centre study, 50% (N = 204) adults admitted into the intensive care unit with COVID-19 infection were referred to SLT for a swallow assessment. 4 Of these patients, 33% required diet modification and 67% were not allowed oral intake. 4 Iatrogenic causes of dysphagia include prolonged intubation 5 and intubation injury including laryngeal oedema, granulations, ulceration and vocal cord immobility. 6 In a recent study of twenty patients with COVID-19 infection who underwent laryngeal endoscopy, the most common laryngeal complications were voice-related complaints, breathing and swallowing. 6 All participants who underwent laryngoscopy in this study presented with abnormal findings, and the most common diagnoses were vocal cord immobility, posterior glottic stenosis and subglottic stenosis. 6 The majority of these patients had been intubated with an average duration of three weeks during their inpatient admission. Moreover, all patients who had been proned during intubation presented with glottal pathology. 6 Tracheostomy insertion can lead to aspiration risk and difficulties managing secretions. 7 In a recent prospective study involving forty-one adults with COVID-19 infection who had a tracheostomy inserted post-extubation, 19% had severe pathology on laryngeal examination. 7 Of note, the vast majority of participants had the tracheostomy inserted beyond fifteen days of oral intubation. 7 1.018-1.112), proning (OR 3.671; 95% CI 1.128-11.943) and pre-existing respiratory disease (OR 5.863; 95% CI 1.521-11.599) were predictors of oral intake status postextubation. Two-thirds (66%) presented with dysphonia post-extubation. Intubation injury (OR 10.471; 95% CI 1.060-103.466) and pre-existing respiratory disease (OR 24.196; 95% CI 1.609-363.78) were predictors of post-extubation voice quality.
Thirty-seven per cent required dysphagia intervention post-extubation, whereas 20% needed intervention for voice. Dysphagia and dysphonia persisted in 27% and 37% cases, respectively, at hospital discharge.
Discussion: Post-extubation dysphagia and dysphonia were prevalent amongst adults with COVID-19 across the ROI. Predictors included iatrogenic factors and underlying respiratory disease. Prompt evaluation and intervention is needed to minimise complications and inform rehabilitation planning.

K E Y W O R D S
COVID-19, dysphagia, dysphonia, intubation, post-extubation, speech and language therapy, swallowing, voice

Key Points
1. Post-extubation dysphagia and dysphonia are multifactorial and can lead to prolonged ICU stay, prolonged tube feeding, aspiration pneumonia and increased morbidity and mortality.
2. In this multi-site prospective cohort study across eleven acute hospitals, 90% of adults required an altered oral diet post-extubation and 36% were not allowed oral intake based on SLT evaluation. Sixty-six per cent presented with post-extubation dysphonia.
3. Age, proning and pre-existing respiratory disease were predictors of post-extubation oral intake status, whereas intubation injury and pre-existing respiratory disease were predictors of post-extubation dysphonia. 4. Over a third (37%) required dysphagia intervention postextubation, whereas 20% needed intervention for voice.

Dysphagia and dysphonia persisted in 27% and 37%
cases, respectively, at hospital discharge, indicating that speech and language therapists should be included in outpatient multidisciplinary COVID clinics in the community.
Over half of the participant group presented with dysphonia whereas 30% reported dysphagia, although 83% were on a normal diet. 7 Other potential factors contributing to PED are delirium, 8 proning, 6,9 disuse atrophy and critical illness neuropathy or myopathy during ICU stay 10 and neurological manifestations of COVID- 19. 11 Central and peripheral nervous system complications of COVID-19 include stroke, encephalitis and Guillain-Barre syndrome. These can damage the neurological swallow network, contributing to dysphagia amongst COVID-19 survivors. 11 PED is associated with worse outcomes in ICU including aspiration pneumonia, prolonged tube feeding, delayed initiation of oral intake, prolonged hospitalisation and increased morbidity and mortality. 12,13 Dysphonia is another recognised complication of intubation reported amongst adults with COVID-19.

| Study design
This multi-site prospective observational cohort study is reported according to the STROBE guidelines for observational cohort studies. 19 Ethical approval for this study was obtained from National Research Ethics Committee (NREC) (20-NREC-COV-051).

| Settings
In this multi-centre observational cohort study, speech and language therapists from eleven acute hospitals across Ireland participated.

| Participants
All adults admitted into a participating acute hospital in the ROI with COVID-19 and referred to SLT were included. Inclusion criteria were

| Independent variables
Demographic data included age, gender, pre-admission medical comorbidities and premorbid swallow status. During hospital admission, data on neurological manifestations of COVID-19 were captured by speech and language therapists from hospital records.
The patient's most recent chest X-ray at time of initial SLT assessment was rated by speech and language therapists based on medical entries into clinical notes using a validated five-point ordinal scoring system provided in dataset dictionary. 20

| Swallow and voice outcomes
Swallowing and voice outcomes were influenced by curtailed access during the COVID-19 pandemic to instrumental assessments typically used in intensive care settings such as fibreoptic endoscopic evaluation of swallowing (FEES). Presence and severity of PED was measured by speech and language therapists using the Functional Oral Intake Scale (FOIS). 22 The FOIS is a validated 7-point ordinal rating scale with high inter-rater reliability. 22 A second proxy dysphagia measure used by speech and language therapists was food and fluid consistencies required for dysphagia management, using the International Dysphagia Diet Standardisation Initiative (IDDSI). 23 Voice quality was evaluated by speech and language therapists using the overall Grade (G) score from the GRBAS scale. 24 The scale has established high rater reliability and is widely used in clinical research. 25

| Data sources/management
One nominated speech and language therapist from each hospital site was responsible for data entry at each location. A dataset and dataset dictionary were emailed to named speech and language therapists from each participating site. Speech and language therapists were instructed to populate the dataset prospectively and return the anonymised data to the first authors for analysis.

| Bias
To minimise observer bias, all clinicians used outcome measures routinely used in clinical practice with established rater reliability. Clear rules and procedures were in place for data collection, and data were clearly defined in a data dictionary provided to all settings. Merged data were anonymised to researchers.

| Study size
Patients who meet the eligibility criteria over the three-month data collection period were included in the study. The study size was determined by the prevalence of cases and in particular those needing respiratory support. Statistical advice was obtained regarding recruitment numbers and statistical power for the sample.

| Statistical analysis
Descriptive statistics were reported using medians and interquartile range (IQR) for continuous data. Categorical variables were presented as frequency (percentage). Variables were tested for normality using the Shapiro-Wilk test. To establish associations between dependent and independent variables, Spearman's rho correlations were conducted. To determine the trajectory of dysphagia and dysphonia from initial SLT assessment to SLT discharge, medians of ordinal dependent variables at both time points were compared using two-tailed Wilcoxon signed rank tests.
To determine independent predictors of oral intake status at time of initial SLT assessment, a binary logistic regression was used. For voice quality, an ordinal logistic regression was completed with the overall (G) four-point ordinal GRBAS rating as the dependent variable. Six independent variables were selected for the voice regression model (intubation injury, proning, maximum cuff pressure, duration of intubation, number of comorbidities and history of respiratory disease).
For both regression models, independent variables were selected based on evidence from previous research and a visual review of the data. Where a significant association was identified between independent variables (eg duration of intubation and presence of tracheostomy), only one was selected for a model. Mean imputation was made for one independent variable (maximum cuff pressure) as it was missing 24/100 cases. Model fits were confirmed using likelihood ratio chi-squared tests. A two-sided α of less than 0.05 was considered statistically significant. Statistical analyses were completed using the SPSS (v26) software.

| Participants
Data from 100 adults with PCR test confirmed COVID-19 infection who were intubated across eleven acute hospitals in ROI and referred to SLT between March and June 2020 inclusive were included in analysis. The 100 adults (69% male) had a mean age of 62 years (age range 17-88 years). Further demographic details are in Table 1.  Table 2. Missing data for some of these variables were due to limited access to data across clinical settings.

| Presence, severity and trajectory of swallowing and voice outcomes post-extubation
Median time between extubation and initial SLT evaluation was 4 days (IQR 2-11 days). Ninety per cent (n = 90) of patients presented with dysphagia (FOIS Level 1-6) at initial SLT assessment with 36% not allowed oral intake (FOIS Level 1) ( Table 3). Median FOIS score at initial SLT assessment was 2.5 (SD 2.139; range 1-7) (n = 100). IDDSI fluid and food consistency findings are detailed in Table 3. A significant negative correlation was observed between oral intake status at initial SLT assessment as rated by the FOIS and ICU length of stay (ICULOS) (r = −.227; p = .028) and also between oral intake status and hospital LOS, indicating that the lower the FOIS score, the longer the LOS (r = −.363; p = .000).
Median voice quality rating also altered significantly from initial SLT assessment (GRBAS score 1) to SLT discharge (GRBAS score 0) (z = −5.619; p = .000). Details regarding alteration in swallow and voice outcomes are in Table 3.

| SLT intervention needs and services provided
Over a third (n = 37/100) of patients required dysphagia intervention post-extubation (Table 6). In 70% (26/37) of these cases, dysphagia intervention was implemented, although 19% (7/37) had it provided in adapted form due to infection risk related to the pandemic. In 11% (4/37) of cases, dysphagia intervention was indicated but could not be provided at point of service delivery due to the pandemic service con-  (Table 6). In 30% (6/20) cases, voice intervention was indicated but could not be implemented also due to pandemic service constraints.

| DISCUSS ION
In this study, 90% of patients intubated as part of COVID-19 management across the ROI who were referred to SLT presented with new onset PED based on oral intake status. Over half (59%) required tube feeding based on SLT assessment, and over a third were not allowed oral intake post-extubation. This high rate of PED compares with recent research. 3,4 Post-extubation oral intake status was associated with length of ICU stay and hospital stay duration in this study, with reduced oral intake associated with longer duration of ICU and hospital admissions.
There was a threefold increase in impact on oral intake status with proning in this study. Lower cranial nerve paralysis and oropharyngeal oedema have previously been linked to proning, and cranial nerves IX to XII are hypothesised to be affected by proning. 6,26 Pre-existing respiratory disease was also identified as a positive predictor of PED in this study. Adults with respiratory disease may already have altered respiratory swallow coordination, which could be exacerbated post-extubation. There was approximately a 6% increase in the relative odds of oral intake status change per year of age in this study. Older people may have a pre-existing presbyphagia, which pre-disposes them to PED. Furthermore, frailty and sarcopaenia may also be prevalent amongst older people, which could contribute towards PED. In contrast to previous research, 5 duration of intubation was not predictive of oral intake status in this study.
This may be due to the fact that patients with tracheostomy were not excluded in this study, as researchers aimed to capture all adults with COVID-19 post-extubation. Additionally, prolonged intubation duration with COVID-19 may explain contrasting findings to previous PED research.
There was a tenfold increase in impact on voice quality for those with intubation injury, which aligns with previous research. 2: Tube dependent with minimal attempts at food or liquid. 14 0 3: Tube dependent with consistent oral intake of food or liquid. 9 0 4: Total oral diet of a single consistency 2 0 5: Total oral diet with multiple consistencies, but requiring special preparation or compensations.

10
6: Total oral diet with multiple consistencies without special preparation, but with specific food limitations.  Post-extubation dysphonia and dysphagia research is needed from future pandemic waves to establish the impact of evolving intensive care management and mutating virus variants on voice and swallowing outcomes. Post-discharge time points to capture longer term voice and swallowing difficulties would guide multidisciplinary service delivery in the community.

| CON CLUS IONS
This study highlights the prevalence of post-extubation dysphagia and dysphonia amongst adults intubated with COVID-19. Awareness of the predictors of altered swallowing and voice quality postextubation will promote early in-depth evaluation and monitoring during hospital stay. Prompt dysphagia and dysphonia evaluation and management is needed to minimise clinical and quality of life complications.

ACK N OWLED G EM ENTS
Thanks to participants across clinical settings for agreeing to contribute data for this study. Thanks also to all the speech and language therapists across data collection sites who assisted with local data collection for the purposes of this research.

D I S C LO S U R E S TAT E M E N T
Authors have no disclosures to report.

CO N FLI C T S O F I NTE R E S T
None.

AUTH O R CO NTR I B UTI O N
J. Regan and M. Walshe designed the study, applied for ethical approval, analysed the data and wrote the paper. All other authors contributed to the study design, acquired and transferred data for analysis and contributed to data analysis and interpretation. All authors gave approval for paper to be submitted for publication.

DATA AVA I L A B I L I T Y S TAT E M E N T
Authors are unable to share data due to ethical approval restrictions.

E TH I C S S TATEM ENT
Ethical approval for this study was obtained from National Research Ethics Committee (NREC) (20-NREC-COV-051).