Patient‐Reported Dysphagia and Swallowing‐Related Quality of Life up to 10 Years After (Chemo)radiotherapy for Head and Neck Cancer: A Cross‐Sectional Study

ABSTRACT

Background

This study sought to better understand patient‐reported dysphagia and swallowing‐related quality of life (swQOL) in patients 6 months to 10 years following definitive (chemo)radiotherapy for head and neck cancer (HNC).

Methods

In this cross‐sectional study, self‐reported dysphagia was assessed with the Eating Assessment Tool (EAT‐10) and swQOL with the MD Anderson Dysphagia Inventory (MDADI) at a single point in time. Sociodemographic and clinical data were also obtained.

Results

Completed questionnaires were received from 626 of 962 (65%) patients with various tumor sites, stages, and time since treatment (M = 58 months, range 6–128). Dysphagia (EAT‐10 ≥ 3) was reported by 444 patients (71%), and poor swQOL (MDADI < 60) by 138 (22%). Several sociodemographic, clinical, and treatment variables were significantly associated with dysphagia and poorer swQOL. Outcomes were not associated with time since treatment.

Conclusions

There remains a high prevalence of dysphagia and poor swQOL up to a decade posttreatment for HNC. This data can help guide survivorship care and lead mitigation strategies to reduce the persistent physical, emotional, and social burden of dysphagia.

1. Introduction

Efforts to mitigate long‐term or late radiation associated dysphagia (RAD) following head and neck cancer (HNC) treatment have received ongoing attention over the past few decades. Strategies such as more precisely targeted radiotherapy techniques, reducing radiation dose to the swallowing organs at risk (SWOARs), prophylactic exercises, treatment de‐escalation, and rationalization of type and timing of feeding tube use have been explored [1, 2, 3]. Yet clinicians continue to encounter patients suffering from RAD years after treatment. In some instances, dysphagia is chronic or long‐term—an unresolved sequelae of acute posttreatment toxicity caused by chronic inflammation, ulcerative changes, or persistent lymphoedema leading to fibrosis [1]. In other instances, dysphagia may be late onset, developing many years after a previously stable period and thought to be caused by fibrosis and atrophy due to progressive and prolific accumulation of collagen and fibrous tissue, microvascular damage, and overproduction of pro‐fibrotic growth factors β (TGF‐β1) [1, 4, 5]. The two phenomena can be difficult to distinguish without long‐term, prospective data, though a thorough clinical history might allow for relatively accurate differentiation. Irrespective of the underlying pathogenesis, RAD can lead to aspiration pneumonia, feeding tube dependence, malnutrition, poor health‐related quality of life (QOL), and anxiety and depression [6, 7, 8, 9, 10].

The landscape of HNC is changing; the increasing prevalence of human papillomavirus (HPV)‐related oropharyngeal cancer means that more patients are living longer following HNC treatment [3]. However, with improved survival rates—including for patients with long‐term RAD—comes escalating challenges such as increasing health care utilization and costs. Effective and cost‐efficient screening, detection, and management of these patients necessitate review of service delivery models to ensure the right care is being given to the right patient at the right time. This requires robust, reliable data to identify those at risk and to understand the impact of RAD on physical function, QOL, and wellbeing. Assessment of physical function requires clinical (ideally instrumental) measures, but to truly understand the impact of RAD on the health and wellbeing of HNC survivors, patient‐reported outcome measures (PROMs) are critical.

Data on the prevalence of late or long‐term RAD and its impact based on patient self‐report remains limited. Much of the existing literature only reports on studies that have gathered self‐report data up to 2 years posttreatment [11, 12, 13, 14]. Studies that extend beyond this are often limited by small sample sizes [15, 16, 17], restricted patient cohorts [5, 18, 19], or reliance on surrogate measures of dysphagia that may not be valid or may not capture the patient perspective [17, 20, 21]. Consequently, understanding of the prevalence and impact of long‐term RAD, particularly in patients routinely treated with modern techniques such as intensity‐modulated radiation therapy (IMRT), is inadequate [22]. High‐quality, large‐scale studies using validated PROMs are needed to quantify and understand the nature of long‐term RAD, identify at‐risk patients, inform survivorship care models, provide patient education, and guide future research into mitigation strategies.

The aim of this study was to examine patient‐reported dysphagia and swallowing‐related QOL (swQOL) in a large sample of HNC patients who were 6 months to 10 years postdefinitive (chemo)radiotherapy, to estimate the prevalence of and associations between patient‐reported dysphagia and swQOL, and to explore the association between patient‐reported problems and clinical characteristics, treatment‐related symptoms, and time posttreatment.

2. Methods and Materials

Institutional Human Research Ethics Committee approval was received prior to the commencement of the study (reference HREC/83995/PMCC) and consent was implied through completion and submission of the questionnaires.

2.1. Study Design and Participants

This was a cross‐sectional study, where data on patient‐reported swallowing problems were collected at a single point in time from patients who were treated at Peter MacCallum Cancer Centre with definitive (chemo)radiotherapy from 6 months to 10 years prior. Potential participants were identified through the Head and Neck Unit patient database and their medical record was screened for eligibility. Patients were invited to participate if they were: over 18 years of age at the time of diagnosis; diagnosed with any stage (I–IV) cancer of the nasopharynx, oropharynx, hypopharynx, or larynx; received intensity‐modulated radiotherapy (IMRT), with or without chemotherapy, with curative intent; and remained alive and well with no evidence of disease recurrence since initial treatment. Patients had to be able to signify consent through choosing to participate and read English, Chinese (simplified), or Vietnamese, according to the most recent notes in their medical record. Patients were excluded if they had a history of head or neck surgery (other than posttreatment neck dissection) or a documented medical condition that could impact swallowing function (e.g., stroke, neurological condition, tracheostomy).

2.2. Data Collection

Potential participants were sent an invitation to participate and a copy of the questionnaires. Questionnaires included the Eating Assessment Tool (EAT‐10) [23], the M.D. Anderson Dysphagia Inventory (MDADI) [24], and five items from the National Cancer Institute (NCI) Patient Reported Outcomes version of the Common Terminology Criteria for Adverse Events (PRO‐CTCAE) [25, 26] measurement system. Study documents were sent electronically if an email address was listed or otherwise via post. In addition to the invitation letter and questionnaires, a list of helpful resources was also sent to all potential participants—these resources included local support groups, information sheets, and reputable websites relevant to side effects following HNC treatment. The invitation letter and list of resources were translated into Chinese (simplified) and Vietnamese to enhance the representative diversity of patients from culturally and linguistically diverse (CALD) backgrounds, given over 80% of patients with nasopharyngeal cancer are of Asian background [27] and these two languages are the most common Asian languages represented at our institution. A modified version of Dillman's total design method [28] was used to maximize response rates, with repeat post or email materials sent as a first reminder and phone calls made as a second reminder.

2.2.1. PROMs

2.2.1.1. Dysphagia

The EAT‐10 is a 10‐item tool that uses Likert‐type responses to provide a measure of dysphagia severity [23]. Items focus on the experience of various swallowing problems and have five response levels ranging from “0” (no problem) to “4” (severe problem). Responses are summed to create a total score ranging from 0 to 40. Higher scores reflect more severe swallowing problems, and a cut‐off score of ≥ 3 is used as an indicator of abnormal swallowing [23]. The tool has demonstrated internal consistency, test–retest reproducibility, and criterion‐based validity [23]. The EAT‐10 has been translated into several different language versions, including Chinese (Mandarin) and Vietnamese, published by the Nestle Nutrition Institute [29].

2.2.1.2. swQOL

The MDADI is a 20‐item tool that uses Likert‐type responses to provide a measure of swQOL [24]. Items focus on perceived limitations in swallowing and associated impact on daily activities and QOL and have five response levels ranging from “1” (strongly agree) to “5” (strongly disagree), although two items are scored in reverse. A global assessment of QOL is based on responses to the item “My swallowing limits my day‐to‐day activities.” Emotional, physical, and functional subscale scores are calculated by summing responses to relevant items (six, eight, and five items, respectively). Subscale scores are summed to create a composite score ranging from 20 to 100. Higher scores reflect better swQOL. In current clinical trials, a composite score < 60 represents “poor” swQOL, scores between 60 and 79 are considered “adequate,” and scores ≥ 80 are considered “optimal” [30]. The MDADI has been translated into Chinese (Mandarin) and validated on patients with HNC [31]. The MDADI has not been translated into Vietnamese, so Vietnamese‐speaking participants had the English version sent and were instructed to only complete this if an English‐speaking relative or friend could assist.

2.2.1.3. Persistent Treatment Toxicities

The NCI PRO‐CTCAE items provide a measure of self‐reported symptom toxicity [26]. The recall period is “the last 7 days” and items selected for this study have five response levels ranging from “none” to “very severe.” Responses are scored from 0 to 4. Toxicities of interest were those that may still be present following HNC treatment and would be likely to impact a patient's perception of their swallowing. They included the severity “at its worst” for: dry mouth, mouth/throat sores, taste changes, dental problems, and jaw movement. Dental problems and jaw movement were developed from the free‐text option using the wording: “In the last 7 days, what was the severity of your teeth or denture problems at its worst?” and “In the last 7 days, what was the severity of your difficulty opening or moving your jaw at its worst?”

2.2.2. Demographic and Clinical Variables

Demographic variables collected from the medical record included: gender, date of birth, residential postcode (to determine remoteness using the Accessibility/Remoteness Index of Australia and socioeconomic disadvantage using the Index of Relative Socio‐economic Disadvantage [IRSD] quintiles) [32, 33], smoking status, and alcohol history at the time of diagnosis. Clinical variables included: tumor site, tumor (T1–T4) and nodal classification (N0–N3), overall stage (I–IV) according to the American Joint Committee on Cancer (AJCC) Staging of HNC 7th edition (for those diagnosed before 2017) or 8th edition (for those diagnosed from January 1, 2017), p16 status (oropharyngeal primary only), date of treatment completion, radiotherapy dose and fractionation schedule, chemotherapy agent and regimen, radiation to the neck (unilateral/bilateral), and mean radiation dose to the larynx, pharyngeal constrictors (as a whole), and parotid(s). Study data were entered into and managed using the REDCap instance hosted by Peter MacCallum Cancer Centre [34].

2.3. Statistical Analysis

Data was exported from REDCap, then imported into R (reference index version 6.1.3) for scoring and analysis. Responses to self‐report measures were scored according to author guidelines and standard recodes applied. Specifically, EAT‐10 and MDADI scores were recoded to discrete variables comprising two categories: “case” (dysphagia and poor function, respectively) or “noncase,” using published cut‐off scores (i.e., ≥ 3 and < 60, respectively) [23, 24]. For the EAT‐10, responses were also coded using a cut‐off of ≥ 4 as suggested by Bofill‐Soler et al. (2021) [35] to specifically represent dysphagia after radiotherapy for HNC [35]. Descriptive statistics were used to summarize patient and clinical characteristics for survey responders and nonresponders, and to summarize responses to self‐report measures (EAT‐10, MDADI, NCI‐PRO‐CTCAE items), as well as the prevalence of dysphagia as assessed by these measures. Confidence intervals for relevant proportions were calculated using the Wilson method. The joint frequency distribution of classifications based on EAT‐10 and MDADI scores was also explored.

Univariable and multivariable linear regression models were used to examine associations between self‐reported swallowing problems and patient/clinical characteristics and persistent treatment toxicities. Patient characteristics included age (centered at 60 years and re‐scaled, so coefficients provide an estimate for each 5‐year increase); gender; residence (major city, regional/remote); smoking history (never smoker, ever smoker [current or ex‐smoker]); and alcohol use (never heavy drinker, ever heavy drinker [current or ex‐heavy alcohol use]). Clinical characteristics included tumor site (nasopharynx, oropharynx, hypopharynx, larynx); T classification (T1, T2, T3, T4); total radiation dose (< 70 Gy, 70 Gy); chemotherapy (yes, no); time since completion of treatment (months); and neck radiation (none, unilateral, bilateral). Persistent treatment toxicities included dry mouth, mouth/throat sores, taste changes, dental issues, and jaw movement. Overall disease stage was not included in the final analysis as participants' date of diagnosis spanned the transition from the 7th to the 8th edition of the AJCC Staging of HNC. This transition marked significant changes to the staging of p16‐positive disease, so T classification was considered more appropriate. These characteristics were chosen based on clinical expertise and previous literature, and were included in both the univariable and multivariable models, with the exception of factors excluded from the multivariable models due to collinearity (e.g., multiple variables were considered a surrogate for radiotherapy dose) and missing data [36].

Exploratory analyses were performed using separate multivariable linear regression models to examine the association between EAT‐10 and MDADI composite scores with mean radiation dose to critical swallowing structures (larynx, pharyngeal constrictor, ipsilateral parotids, and contralateral parotids), controlling for time since treatment.

3. Results

Figure 1 shows the participant flow from screening through to questionnaire completion and submission.

FIGURE 1.

Participant flow diagram.

Questionnaires were sent to 962 eligible patients and responses submitted by 626 (65.1% response rate). Three MDADI responses were missing due to language. Demographic and clinical characteristics for all eligible patients (i.e., respondents and nonrespondents) are shown in Table 1. Questionnaires were received and returned by patients with a wide range of tumor sites and sizes, representing a typical HNC cohort treated within a large tertiary head and neck oncology unit. There was relatively even representation in demographic, tumor, and treatment characteristics between the respondents and nonrespondents, and time since treatment did not differ, with a mean time since treatment of 58.4 months for respondents and 56.1 months for nonrespondents. Diversity in terms of remoteness (of residence) and socioeconomic disadvantage was representative of the distribution seen at our center. Representative diversity was adequate for CALD patients but not equal across respondents and nonrespondents, despite having study material translated into Chinese and Vietnamese, with 37% of non‐English speaking patients responding (n = 22 out of 59) compared to 67% of English‐speaking patients (n = 604 out of 903).

TABLE 1.

Characteristics of respondents (n = 626) and nonrespondents (n = 336).

	Respondents	Nonrespondents
Age at treatment completion
Mean (SD)	60.6 (9.1)	59.2 (11.4)
Median (IQR)	60.5 (55, 67)	60 (52, 67.3)
Range	30, 90	28, 92

	n	%	n	%
Sex
Male	523	83.5	281	83.6
Female	103	16.5	55	16.4
Language
English	604	96.5	299	89.0
Not English	22	3.5	37	11.0
Remoteness
Major city	455	72.7	270	80.4
Inner regional	135	21.6	52	15.5
Outer regional	35	5.6	14	4.2
Remote	1	0.2	0	0
Socioeconomic disadvantage (IRSD quintiles)
1 (Most disadvantaged)	66	10.5	40	11.9
2	98	15.7	61	18.2
3	118	18.8	86	25.6
4	143	22.8	66	19.6
5 (Least disadvantaged)	199	31.8	82	24.4
Missing	2	0.3	1	0.3
Smoking status
Never smoker	224	35.8	116	34.5
Ex‐smoker	310	49.5	151	44.9
Current smoker	90	14.4	65	19.3
Missing	2	0.3	4	1.2
Alcohol use
Non/social drinker	503	80.4	255	75.9
Ex‐heavy drinker a	54	8.6	27	8.0
Current heavy drinker a	67	10.7	51	15.2
Missing	2	0.3	3	0.9
Tumor site
Nasopharynx	61	9.7	60	17.9
Oropharynx	438	70	190	56.5
Larynx	107	17.1	74	22.1
Hypopharynx	20	3.2	12	3.6
T classification
1	201	32.1	114	33.9
2	207	33.1	113	33.6
3	139	22.2	78	23.2
4	79	12.6	31	9.2
N classification
0	146	23.3	91	27.1
1	211	33.7	98	29.2
2	237	37.9	120	35.7
3	32	5.1	27	8.0
p16 status b
Positive	403	92.0	167	87.9
Negative	29	6.6	19	10.0
Missing	6	1.4	4	2.1
Radiotherapy dose
63−68 Gy	84	13.4	52	15.5
70 Gy	542	86.6	284	84.5
Chemotherapy
Yes	510	81.5	257	76.5
No	116	18.5	79	23.5
ECOG
0	545	87.1	265	78.9
> 0	79	12.6	70	20.8
Missing	2	0.3	1	0.3
Time since treatment completion (months)
Mean (SD)	58.4 (33.5)		56.1 (33.9)
Median (IQR)	54 (29, 84.8)		50 (27.8, 85)
Range	6, 128		4, 125
Only collected for respondents c
Neck radiotherapy laterality	n	%
Unilateral	77	12.3
Bilateral	484	77.3
Nil	65	10.4
Mean radiotherapy dose to larynx (n = 310)
Mean (SD)	34.2 (11.6)
Median (IQR)	32.5 (27.3, 40)
Range	9.7, 71
Mean radiotherapy dose to pharyngeal constrictors (n = 430)
Mean (SD)	48.0 (9.1)
Median (IQR)	48.9 (44, 53.9)
Range	14.8, 70.1
Mean radiotherapy dose to ipsilateral parotid d (n = 621)
Mean (SD)	34.1 (16.3)
Median (IQR)	35.3 (25, 46.6)
Range	0, 68.6
Mean radiotherapy dose to contralateral parotid e (n = 602)
Mean (SD)	20.5 (10.0)
Median (IQR)	24.4 (16.1, 26.0)
Range	0, 52.4

Abbreviations: IQR, interquartile range; IRSD, index of relative socioeconomic disadvantage; SD, standard deviation.

^a Based on Australian guidelines of regularly drinking more than 10 standard drinks a week or four standard drinks on any day [37].

^b p16 status reported for patients with oropharynx primary only; n = 438 responders, 190 nonresponders.

^c Radiotherapy details were collected for respondents only to limit the collection of redundant data.

^d Mean dose to parotid that received the greater dose for midline primaries.

^e Mean dose to parotid that received the lowest dose for midline primaries.

3.1. Prevalence of Dysphagia

Self‐reported dysphagia (EAT‐10 total score ≥ 3) was identified in 444 patients (70.9%, 95% CI: 67.3%–74.3%). The mean EAT‐10 total score was 10.4 (SD = 10.1). Using the higher cut‐off (total score ≥ 4), dysphagia was identified in 410 patients (65.5%, 95% CI: 61.7–69.1).

Descriptive statistics for responses to individual EAT‐10 items are presented in Table S1. High severity ratings (responses of “3” or “4”) were greatest for the following items: “When I swallow food sticks in my throat” (18.5% and 13.3%, respectively), “Swallowing solids takes extra effort” (16.8% and 12.0%, respectively), and “The pleasure of eating is affected by my swallowing” (10.1% and 11.7%, respectively).

3.2. Prevalence of Poor swQOL

MDADI composite scores indicated “poor” swQOL in 138 patients (22.2%, 95% CI: 19.1%–25.6%), “adequate” swQOL in 201 (32.3%, 95% CI: 28.7%–36.0%), and “optimal” swQOL in 284 (45.6%, 95% CI: 41.7%–49.5%). In the subset of patients with oropharynx cancer (n = 438), scores indicating “poor,” “adequate,” and “optimal” swQOL were observed in 100 (22.9%), 145 (33.2%), and 192 (41.1%) patients, respectively. Descriptive statistics for the composite score, global assessment, and subscales are reported in Table 2.

TABLE 2.

Descriptive statistics for MDADI scores ^a (n = 623).

	Global	Functional	Emotional	Physical	Composite
Mean (SD)	78.0 (24)	80.3 (19.0)	76.5 (19.8)	71.5 (20.5)	75.4 (18.8)
Median (IQR)	80 (60, 100)	84 (72, 100)	80 (63.3, 93.3)	70 (55, 90)	76.8 (63.2, 92.6)
Range	20, 100	20, 100	20, 100	20, 100	26.3, 100

Abbreviations: IQR, interquartile range; SD, standard deviation.

^a Higher scores represent better swallowing‐related quality of life.

3.3. Prevalence of Persistent Treatment Toxicities

Descriptive statistics for the PRO‐CTCAE items are presented in Table S2. Xerostomia was the most prevalent and severe toxicity reported, with 34.0%, 13.6%, and 6.1% experiencing a “moderate,” “severe,” or “very severe” dry mouth over the previous 7 days, respectively. Only 18.2% reported no dry mouth symptoms. Mouth or throat sores were the least prevalent symptom, with 67.1% of respondents reporting this was not a problem at all over the previous 7 days, consistent with the resolution of acute toxicities at this later time posttreatment.

3.4. Subgroup Analysis

EAT‐10 and MDADI scores by tumor site and stage of disease are presented in Table 3. For patients with an oropharynx or larynx tumor, as T‐classification increased, those patients reported more severe dysphagia (higher EAT‐10 scores), higher prevalence of dysphagia (higher rates of EAT‐10 scores ≥ 3), and poorer swQOL (lower mean MDADI composite scores). This trend was not evident for those with nasopharynx tumors, and numbers were too small for hypopharynx tumors to identify any trends. For patients with “early” (T1‐2, N0‐2b) HPV‐associated oropharyngeal cancer (n = 255), the mean EAT‐10 score was 8.4 out of 40 (SD = 8.7), the proportion of patients with an EAT‐10 score ≥ 3 was 67.5% (n = 172), and the mean MDADI composite score was 79.4 (SD = 17.7), just below the cut‐off of ≥ 80 to represent “optimal” swQOL.

TABLE 3.

EAT‐10 and MDADI scores by tumor site and stage of disease.

	T1	T2	T3	T4
Nasopharynx
n	25	12	11	13
EAT‐10 score a (mean [SD])	8.5 (8.8)	13.3 (11.6)	9.8 (9.6)	10.9 (10.9)
EAT‐10 dysphagia (≥ 3) (n [%])	15 (60.0)	11 (91.7)	8 (72.7)	9 (69.2)
MDADI score b (mean [SD])	75.4 (14.1)	71.1 (20.0)	75.8 (16.5)	77.1 (16.9)
Oropharynx
n	127	162	84	65
EAT‐10 score (mean [SD])	7.4 (8.2)	9.1 (8.9)	14.5 (10.6)	16.5 (11.2)
EAT‐10 dysphagia (≥ 3) (n [%])	77 (60.6)	117 (72.2)	74 (88.1)	55 (84.6)
MDADI score (mean [SD])	81.7 (17.1)	77.7 (17.5)	68.1 (18.4)	64.0 (17.9)
Larynx
n	45	28	34	0
EAT‐10 score (mean [SD])	4.4 (7.0)	7.6 (9.9)	14.2 (11.8)
EAT‐10 dysphagia (≥ 3) (n [%])	19 (42.2)	16 (57.1)	26 (76.5)
MDADI score (mean [SD])	86.3 (15.3)	80.1 (18.0)	67.6 (21.6)
Hypopharynx
n	4	5	10	1
EAT‐10 score (mean [SD])	11.8 (9.8)	7.4 (6.1)	19.6 (12.6)
EAT‐10 dysphagia (≥ 3) (n [%])	4 (100.0)	3 (60.0)	9 (90.0)
MDADI score (mean [SD])	72.9 (17.3)	74.1 (19.7)	61.8 (24.4)

^a EAT‐10: higher scores represent more severe dysphagia.

^b MDADI composite score: higher scores represent better swallowing‐related quality of life.

3.5. Joint Frequency Distribution Analysis

The relationship between patient‐reported dysphagia (EAT‐10 total scores) and swQOL (MDADI composite scores) was explored through cross‐tabulation, and a scatterplot of EAT‐10 and MDADI scores is shown in Figure 2. All patients (excepting one outlier) who reported no dysphagia (EAT‐10 score < 3) reported “adequate” to “optimal” swQOL (MDADI composite score > 60). However, for those who did report dysphagia (EAT‐10 score ≥ 3), only 31% reported “poor” swQOL (upper left quadrant); 41% reported “adequate” swQOL, and 28% reported “optimal” swQOL (both upper right quadrant).

FIGURE 2.

Scatterplot of EAT‐10 total scores (n = 626), with a horizontal line showing the cut‐off of ≥ 3 to represent dysphagia, and MDADI composite scores, with a vertical line showing the cut‐off of < 60 to represent poor swallowing‐related QOL (n = 623).

3.6. Linear Regression Analysis

On univariable analysis, both dysphagia and swQOL were associated with place of residence (worse dysphagia and swQOL reported for those living in regional/remote areas than those in the metropolitan city), smoking status (worse dysphagia and swQOL reported for current/ex‐smokers than for never smokers), alcohol history (MDADI only; worse swQOL for those with a history of heavy alcohol intake), site of disease (worse dysphagia and swQOL reported for those with hypopharyngeal tumors), T‐classification (worse dysphagia and swQOL reported with each increase in T‐classification), total radiation dose (worse dysphagia and swQOL reported for those who received 70Gy compared to those whose total dose was < 70 Gy), chemotherapy (worse dysphagia and swQOL reported for those who received chemotherapy), and neck radiation (worse dysphagia and swQOL reported for those who received bilateral neck radiation compared to those who received no or unilateral neck radiation) (all p < 0.05, see Table 4). Dysphagia and swQOL were not associated with age at diagnosis, sex, alcohol history (EAT‐10 only), or time since treatment. On multivariable analysis, residence (for MDADI/swQOL only), smoking status, site of disease, T‐classification, and neck radiation all remained significant (all p < 0.05), but residence (for EAT‐10), chemotherapy, and alcohol history were no longer significant (see Table 4).

TABLE 4.

Univariable and multivariable regression results for EAT‐10 total and MDADI composite scores.

	EAT‐10 (higher scores worse)					MDADI (higher scores better)
	Mean (SD)	Univariable		Multivariable		Mean (SD)	Univariable		Multivariable
		Estimate (95% CI)	p	Estimate (95% CI)	p		Estimate (95% CI)	p	estimate (95% CI)	p
Age a	—	0.3 (−0.1, 0.7)	0.18	0.2 (−0.2, 0.5)	0.38	—	−0.4 (−1.2, 0.4)	0.35	−0.02 (−0.7, 0.7)	0.95
Sex
Female	11.9 (10.8)	ref	0.09	ref	0.24	73.6 (18.1)	ref	0.30	ref	0.84
Male	10.1 (10.0)	−1.8 (−4.0, 0.3)		1.0 (−2.6, 0.6)		75.7 (19.0)	2.1 (−1.9, 6.1)		−0.3 (−3.4, 2.8)
Residence
Major city	9.7 (9.9)	ref	0.007	ref	0.052	76.8 (18.3)	ref	0.002	ref	0.04
Regional/remote	12.2 (10.6)	2.5 (0.7, 4.2)		1.3 (−0.01, 2.7)		71.6 (19.6)	−5.2 (−8.5, −1.9)		−2.7 (−5.3, −0.1)
Smoking status at diagnosis
Never smoker	8.9 (9.2)	ref	0.02	ref	0.01	79.6 (17.0)	ref	0.0002	ref	< 0.005
Ever smoker	11.3 (10.6)	2.4 (0.8, 4.1)		1.7 (0.3, 3.0)		73.0 (19.4)	−6.5 (−9.6, −3.5)		−5.1 (−7.7, −2.6)
Alcohol history at diagnosis
Never heavy drinker	10.0 (10.0)	ref	0.15	ref	0.97	76.3 (18.6)	ref	0.02	ref	0.79
Ever heavy drinker	11.9 (10.8)	1.8 (−0.2, 3.8)		−0.03 (−1.6, 1.5)		71.7 (19.3)	−4.6 (−8.3, −0.9)		−0.4 (−3.4, 2.6)
Tumor site
Oropharynx	10.7 (10.0)	ref	0.03	ref	0.02	75.0 (18.7)	ref	0.055	ref	0.019
Hypopharynx	15.0 (1.3)	4.3 (−0.3, 8.8)		4.4 (1.0, 7.9)		67.1 (21.2)	−7.9 (−16.4, 0.5)		−7.9 (−14.6, −1.3)
Larynx	8.3 (10.3)	−2.4 (−4.5, −0.2)		2.9 (0.3, 5.5)		78.8 (19.7)	3.8 (−0.2, 7.7)		−6.2 (−11.3, −1.2)
Nasopharynx	10.2 (9.9)	−0.5 (−3.3, 2.2)		0.06 (−2.2, 2.1)		74.9 (16.1)	−0.1 (−5.2, 5.0)		−1.0 (−5.2, 3.1)
T classification
T1	7.0 (8.1)	ref	< 0.005	ref	< 0.005	81.8 (16.6)	ref	< 0.005	ref	< 0.005
T2	9.1 (9.2)	2.1 (0.2, 4.0)		0.6 (−0.9, 2.1)		77.5 (17.7)	−4.3 (−7.7, −0.8)		−1.3 (−4.2, 1.5)
T3	14.4 (11.0)	7.5 (5.4, 9.5)		4.2 (2.3, 6.1)		68.2 (19.5)	−13.6 (−17.5, −9.7)		−7.7 (−11.3, −4.1)
T4	15.6 (11.2)	8.7 (6.2, 11.2)		3.8 (1.7, 5.9)		65.9 (18.2)	−15.9 (−20.6, −11.2)		−7.5 (−11.6, −3.5)
Radiotherapy dose
< 70 Gy	5.9 (7.7)	ref	< 0.005	—		83.5 (16.2)	ref	< 0.005	—
70 Gy	11.1 (10.3)	5.2 (2.9, 7.5)		—		74.1 (18.9)	−9.4 (−13.7, −5.1)		—
Chemotherapy
No	7.3 (8.7)	ref	0.0003	ref	0.26	80.2 (17.2)	ref	0.002	ref	0.10
Yes	11.1 (10.3)	3.8 (1.8, 5.8)		−1.3 (−3.5, 1.0)		74.3 (19.0)	−5.9 (−9.7, 2.1)		3.6 (−0.7, 7.9)
Time since treatment
Per month increase	—	−0.01 (−0.04, 0.01)	0.26	−0.0003 (−0.02, 0.02)	0.97	—	0.02 (−0.03, 0.06)	0.41	−0.01 (−0.04, 0.03)	0.66
Neck radiotherapy
None	4.9 (7.5)	ref	< 0.005	ref	0.009	85.2 (15.7)	ref	< 0.005	ref	< 0.005
Unilateral	5.2 (6.9)	0.3 (−2.9, 3.5)		2.0 (−1.9, 5.9)		84.8 (15.9)	−0.5 (−6.5, 5.5)		−6.4 (−14.0, 1.1)
Bilateral	12.0 (10.4)	7.0 (4.5, 9.5)		4.3 (0.6, 7.9)		72.5 (18.7)	−12.7 (−17.4, −8.0)		−10.3 (−17.2, −3.3)
Additional symptoms b
Dry mouth	—	4.8 (4.2, 5.4)	< 0.005	2.2 (1.5, 2.8)	< 0.005	—	−8.2 (−9.3, −7.0)	< 0.005	−3.3 (−4.6, −2.1)	< 0.005
Mouth or throat sores	—	5.3 (4.5, 6.1)	< 0.005	1.9 (1.1, 2.6)	< 0.005	—	−8.7 (−10.2, −7.2)	< 0.005	−2.1 (−3.6, −0.6)	< 0.005
Problems with taste	—	4.7 (4.0, 5.3)	< 0.005	2.0 (1.3, 2.7)	< 0.005	—	−8.3 (−9.5, −7.2)	< 0.005	−3.9 (−5.3, −2.6)	< 0.005
Teeth/denture problems	—	3.7 (3.0, 4.3)	< 0.005	0.3 (−0.4, 1.0)	0.37	—	−7.2 (−8.3, −6.0)	< 0.005	−1.6 (−2.9, −0.4)	0.01
Difficulty moving jaw	—	4.5 (3.7, 5.2)	< 0.005	1.5 (0.8, 2.3)	< 0.005	—	−8.3 (−9.5, −7.2)	< 0.005	−2.4 (−3.9, −0.9)	< 0.005

Note: Bold text indicates p < 0.05.

^a Age: per 5‐year increase, centered at 60 years.

^b 1–5 scale included as continuous, per one unit score increase.

3.7. Exploratory Analysis

Exploratory analysis identified significant associations between patient‐reported swallowing problems (both EAT‐10 and MDADI composite scores) and mean radiation dose to all four structures examined—pharyngeal constrictors, larynx, ipsilateral parotids, and contralateral parotids (all p < 0.005; Table S3), controlling for time since treatment. The strongest associations were identified for the mean radiation dose to the pharyngeal constrictors (OR 1.8, 95% CI: 1.3, 2.3) and (OR −3.2, 95% CI: −4.1, −2.3) for EAT‐10 and MDADI, respectively, and mean radiation dose to the contralateral parotids (OR 1.5, 95% CI: 1.1, 1.9) and (OR −2.8, 95% CI: −3.5, −2.1) for EAT‐10 and MDADI composite scores, respectively.

4. Discussion

This is one of the largest studies investigating both patient‐reported dysphagia and swQOL in a diverse sample of survivors up to 10 years after (chemo)radiotherapy for HNC. Our findings provide comprehensive data on prevalence, with 71% of patients reporting dysphagia and 22% reporting “poor” swQOL. Our study employed a robust follow‐up methodology, achieving a response rate of 65%, and there were no significant demographic or clinical differences between responders and nonresponders. The representativeness of our sample is further enhanced by its diversity in terms of remoteness and socioeconomic status, and, while there was a lower response rate from patients with CALD backgrounds, to our knowledge this is the first study in this patient population that has endeavored to maximize diversity by recruiting patients from a variety of language backgrounds using translated materials.

Our prevalence estimates for both dysphagia and swQOL are similar to those found in previous studies of swallowing outcomes in long‐term HNC survivors. In the only other study that measured patient‐reported dysphagia (rather than swQOL) with a validated tool (the Sydney Swallow Questionnaire), dysphagia was identified in 59% of 83 patients who were 6 months to 8 years posttreatment for a range of HNC sites and stages [15]. Findings from studies that have investigated swQOL using the MDADI are summarized below (Table 5). Lower rates of “poor” and “adequate” function reported by Aggarwal et al. (2019) [5] likely reflect the study sample (i.e., patients treated for primarily small oropharynx cancers).

TABLE 5.

Studies investigating long‐term swallowing problems using the MDADI; prevalence of “poor” or “poor to adequate” function.

Author	No. of patients	Tumor site	Time posttreatment	Prevalence of MDADI score < 60 a	Prevalence of MDADI score < 80 b
Patterson [22]	42	OP, NP, HP, Lx, UKP	6 years	Not reported	71%
Dixon [18]	201	OP	2–5 years	32%	70%
Aggarwal [5]	889	OP	1–16 years	14%	44.8%
Orlandi [19]	148	OP	6 months–6 years	17.6%	Not reported
Current study	626	OP, NP, HP, Lx	6 months–10 years	22%	54.5%

Abbreviations: HP, hypopharynx; Lx, larynx; NP, nasopharynx; OP, oropharynx; UKP, unknown primary.

^a MDADI scores < 60 represent “poor” swQOL.

^b MDADI scores < 80 represent “less than optimal” swQOL (i.e., “poor” to “adequate”).

More recently, Huynh (2024) [20] reported that 31% of 239 long‐term HNC survivors, treated with either definitive (63%) or postoperative (37%) (chemo)radiotherapy between 5 and 13 years prior, had dysphagia. In this study, dysphagia was determined using a study‐specific cut‐off of ≥ 25 (out of 100) on the four‐item swallowing scale from the EORTC Head and Neck module (EORTC QLQ‐H&N35). Other researchers have tried to determine the prevalence of long‐term dysphagia using International Classification of Diseases (ICD) codes or documentation from medical records. Not surprisingly, when only diagnosable medical conditions are considered, rather than patient‐reported symptoms, prevalence rates are lower, ranging from 26% to 45% [14, 17, 38]. The high prevalence of patient‐reported problems reported in the literature irrespective of measurement tool used—including in our large sample—indicates that dysphagia and its impact on QOL is a common, long‐term concern for HNC survivors.

One of the most notable findings in our study is the lack of association between the time since treatment completion and reported swallowing problems (both dysphagia and swQOL), a trend also found by Szczesniak et al. (2014) in their study of self‐reported dysphagia in patients 6 months to 8 years posttreatment [15]. In contrast, Aggarwal et al. (2019) [5] reported worse MDADI scores in patients who were longer posttreatment, while Orlandi et al. (2019) [19] reported improved MDADI scores the further posttreatment patients were—both studies were limited to patients treated for oropharyngeal cancer only and results were only significant on univariable analysis. Due to the cross‐sectional nature of these earlier studies and our current study, we cannot determine what proportion of our patients experienced “late onset” dysphagia [39], versus those who experience persistent, chronic problems [40]. Irrespective, patients continue to report issues up to a decade following treatment, supporting the development of survivorship clinics, innovative mechanisms of long‐term support, and improved swallowing surveillance with strategies for early identification and intervention when issues are identified [41, 42].

Interpreting the potential impact of time since treatment is further complicated by the known discrepancy between patient‐reported problems and observed swallowing dysfunction [43, 44, 45], including a small number of studies suggesting that this discrepancy may increase over time. In previous work from our institution, 24% of patients demonstrated clinically significant deterioration in function on videofluoroscopic swallowing studies (VFSS) in the 4–5 years following treatment, despite stable or improved self‐reported eating activity [16] and other authors have shown significant correlations between swallowing on VFSS and the EAT‐10 before and soon after HNC treatment, but not 1 year posttreatment [46]. When considering patient self‐report alone, we found that dysphagia based on EAT‐10 scores was more prevalent than poor swQOL based on MDADI scores for patients many years after HNC treatment. This contrast between a patient's dysphagia symptoms and their ratings of its impact on emotional wellbeing or day‐to‐day function suggests two important action points. First, there is an adaptation that HNC patients demonstrate—70% of our sample reported “adequate” or “optimal” swQOL on the MDADI despite reporting dysphagia on the EAT‐10—so investigation of approaches that might harness this apparent “response shift” [47] is warranted. Second, swallowing is a complex, multifactorial construct, so studies should attempt to include the complementary perspectives of instrumental measures, clinician‐rated assessments, and PROMs where possible, acknowledging that these different elements all have different degrees of specificity and patient relevance [43].

It is important to acknowledge that prevalence estimates are determined by recommended cut‐off scores. The EAT‐10 cut‐off score of ≥ 3 to represent dysphagia (or “abnormal” swallowing) was generated from normative data rather than utilizing an approach such as receiver operating characteristic (ROC) analysis, so this may not be appropriate as a screening tool in the HNC population. Recommendations for increasing the EAT‐10 cut‐off to 4 to represent aspiration risk in HNC patients 3 months postchemoradiation have been made [35], though this cut‐off did not substantially change our prevalence rate for dysphagia, where estimates reduced from 71% to 66%. When people are many years posttreatment, eating and drinking are impacted by multiple chronic or late toxicities such as fibrosis, xerostomia, lymphoedema, dental issues, and taste changes, so further research using instrumental measures to better understand the psychometric properties of the EAT‐10 in this population is needed.

This study identified risk factors associated with worse patient‐reported swallowing problems in long‐term survivors. A small number of previous studies have also identified these factors as being relevant, including tumor site and size [5, 16, 38, 48], smoking history [5], location of residence [16], and chemotherapy [5, 38]. To our knowledge, this is the first time that treatment laterality to the neck has been associated with long‐term patient‐reported swallowing problems. While treatment laterality was not associated with clinician‐ and patient‐reported outcomes in a sample of patients with low‐risk HPV‐associated oropharyngeal cancer 2 years posttreatment [49], our results may reflect the broader patient sample, ranging from N0 to N3 disease and including those who received no neck radiation. While the rationalization of neck treatment in a heterogeneous group is largely out of the clinician's control, the need for bilateral treatment should be taken into account when identifying those at higher risk of dysphagia and those who may benefit from more long‐term monitoring or support programs. The impact of smoking history on swallowing outcomes has received limited attention in the literature; however, toxicities such as osteoradionecrosis and laryngeal oedema may be more prevalent in persistent smokers, potentially increasing their risk of dysphagia [13]. Smoking cessation programs should be considered an integral part of survivorship care, not just for overall survival and locoregional control [50] but also for its potential impact on swallowing and QOL outcomes. The impact of age on long‐term swallowing function remains somewhat contentious despite presbyphagia being a known phenomenon in the normal population [51]. Similar to Aggarwal et al. (2019) [5], we did not find any association between age and swallowing problems; however, age has been identified as a significant factor in other long‐term studies [16, 20, 38]. Similarly, gender was not significant in our analyses, although other studies have identified females as reporting worse long‐term dysphagia [20] and swQOL [19] than males. Further, confirmation of factors associated with worse swallowing outcomes is particularly relevant for clinical practice, as they can inform targeted interventions, screening programs, and models of care. Notably, factors such as radiation dose to the SWOARs have been repeatedly shown to exacerbate long‐term dysphagia [16, 20, 48] and should be prioritized when designing personalized treatment and rehabilitation strategies for HNC survivors.

Dysphagia after radiotherapy for HNC is multifactorial, with potential contributors including fibrosis, muscle atrophy, lymphoedema, stricture, trismus, poor dentition, and persistent toxicities such as chronic xerostomia. Lower cranial neuropathy can occur as a late effect in around 4% of patients after treatment for oropharyngeal cancer [5] and between 4% and 14% of patients treated for nasopharyngeal [52, 53] cancer, potentially affecting swallowing efficiency, safety, and swallowing‐related QOL [5]. The higher severity of problems swallowing solids compared to fluids reported on the EAT‐10 in this study is not surprising, given the impact of radiation‐related soft tissue injury on pharyngeal swallow physiology [54]. Chronic xerostomia further impacts the effective transit of solids through the pharynx, and is strongly correlated with dysphagia both in the current study and in others [43]. We found that additional factors such as taste are also significantly associated with the severity of self‐reported dysphagia and swQOL. Undoubtedly, all these factors combine to influence the prevalence of swallowing problems in our sample and further investigation to better understand their role in long‐term swallowing outcomes along with ongoing efforts to mitigate these long‐term effects is warranted.

Despite significant efforts made to minimize nonresponse and maximize representative diversity, study findings may not be generalizable to all nonsurgical HNC patients. Approximately one in three eligible participants chose not to participate or did not receive the study invitation as their contact details were out‐of‐date. The sociodemographic and clinical characteristics of respondents and nonrespondents were very similar; however, we cannot rule out the possibility that the prevalence of dysphagia and “poor” swQOL were either over‐ or underestimated. This study was also undertaken at a single institution, which may limit generalizability to centers or regions with different treatment protocols or patient demographics. Nevertheless, at 65%, the response rate was relatively high and the sample representative of the Australian HNC population, as such the findings are likely relevant to similar healthcare settings and the diverse population of HNC survivors included. Due to the cross‐sectional design, we were not able to determine whether self‐reported dysphagia and “poor” swQOL represent chronic conditions present since treatment completion or whether they reflect the development of late‐onset problems. Further research, ideally using a longitudinal approach and in a multicenter setting, is needed to confirm and expand on our findings. Finally, data on baseline swallowing function and speech pathology interventions during and/or after radiotherapy would have been useful to include in our regression models. While this data would not have impacted reported prevalence, both baseline swallow function [16, 18, 48] and speech pathology interventions [41] have been shown to influence long‐term swallowing outcomes, so determining their associations with swallowing in conjunction with the other factors investigated would have been valuable.

5. Conclusion

This study highlights the significant burden of dysphagia and “poor” swQOL in HNC survivors following (chemo)radiotherapy, which is present irrespective of the time since treatment completion. As the population of HNC survivors continues to increase, there is an ongoing need for tailored strategies that reduce the risk of long‐term complications, while also developing screening programs and models of care that support patients throughout the survivorship phase when unavoidable complications persist or emerge in the years (or even decades) following treatment. Our findings suggest that these interventions should be multifaceted, addressing not only physical impairments but also the broader impact of dysphagia on patients' emotional and social well‐being. Combining patient‐reported outcomes with instrumental measures of swallowing function is essential in refining our understanding of this complex issue, both in future research and when guiding clinical practice.

Ethics Statement

The study was approved by the Peter MacCallum Cancer Center ethics committee prior to commencement (HREC approval: HREC/83995/PMCC).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Descriptive statistics for EAT‐10 items (n = 626).

Table S2: Descriptive statistics for PRO‐CTCAE items (n = 626).

Table S3: Regression results for EAT‐10 total and MDADI composite scores and mean radiation dose to swallowing “organs at risk.”

Frowen J. J., Drosdowsky A., Fua T. F., McDowell L. J., and Gough K., “Patient‐Reported Dysphagia and Swallowing‐Related Quality of Life up to 10 Years After (Chemo)radiotherapy for Head and Neck Cancer: A Cross‐Sectional Study,” Head & Neck 48, no. 3 (2026): 733–746, 10.1002/hed.70065.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.