Dysphagia-Related Quality of Life in Oculopharyngeal Muscular Dystrophy: Psychometric Properties of the SWAL-QOL Instrument

Abstract

Introduction

The Swallowing Quality of Life instrument (SWAL-QOL) is a patient-reported outcome (PRO) measure of swallowing-related quality of life (SR-QoL). Its psychometric properties in oculopharyngeal muscular dystrophy (OPMD) are not known.

Methods

We administered the SWAL-QOL to U.S. OPMD Registry participants. We described SR-QoL profiles and assessed reliability and validity.

Results

Mean composite score in 113 individuals with OPMD was 54.4 ± 20.7, indicating moderate impairment. Severe impairments were observed in eating duration, burden, and fatigue scales. Internal consistency reliability of all scales was found to be satisfactory, and 9 of 10 scales demonstrated adequate test-retest reliability. Data confirmed 86% of hypotheses, supporting construct validity. The SWAL-QOL limitations in OPMD include: floor/ceiling effects in 7 of 10 scales and low specificity of sleep, fatigue, and communication scales for dysphagia.

Discussion

SR-QoL is reduced in OPMD. Given several limitations of the SWAL-QOL, development of an improved dysphagia-specific QoL instrument for OPMD is warranted.

INTRODUCTION

Oculopharyngeal muscular dystrophy (OPMD) is a rare, usually autosomal dominant, late-onset muscle disease that causes ptosis, dysphagia, dysphonia, and proximal limb weakness.^1,2 Dysphagia in OPMD is a debilitating symptom that may lead to malnutrition and aspiration pneumonia.³

A recent Cochrane Review of dysphagia interventions in chronic, progressive muscle disease noted the lack of standardized methods for assessment of dysphagia in previous and ongoing trials of sporadic inclusion body myositis, myotonic dystrophy, and OPMD, and concluded that “standardized, validated, and reliable outcome measures are needed to assess dysphagia and any possible treatment effect.”⁴ Both physiologic measures and patient reported outcomes (PROs) have been employed in prior studies of OPMD.^5–7 PROs are increasingly used as endpoints in clinical trials.⁸ With recent initiation of the first clinical trials in OPMD,⁷ it has become imperative to identify a dysphagia PRO that produces reliable and valid data for use in assessing OPMD.

The SWAL-QOL is an existing PRO intended to be a disease-specific measure of quality of life (QoL) for patients with oropharyngeal dysphagia.⁹ It was given high ratings for its psychometric properties in a recent review of dysphagia QoL PROs.¹⁰ The SWAL-QOL was initially developed and its validity explored in a sample drawn primarily from Veterans Association medical centers.^9,11,12 Among the dysphagia etiologies in the initial validation sample, 13% were characterized as degenerative neurologic disease, and none were characterized as chronic progressive muscle disease. ¹² Instruments used in research or clinical practice should have sound measurement properties, including reliability and validity, in the population to which they are applied.¹³ The SWAL-QOL’s psychometric properties in a purely OPMD population have not been described.

The purpose of this study was to investigate psychometric attributes of the SWAL-QOL when applied to OPMD and thereby assess its utility as an OPMD research outcome measure. We further aimed to define SR-QoL impairment profiles in OPMD. We hypothesized that the SWAL-QOL would demonstrate good reliability and construct validity in OPMD and that SR-QoL would be severely reduced in OPMD.

MATERIALS AND METHODS

Subjects and Study Design

This was a prospective, cross-sectional study. We sampled participants in the U.S. OPMD Registry between October, 2012 and November, 2015. All participants were ≥ 18 years of age and completed the SWAL-QOL in English by mail administration. A subset of consecutive registry participants was concurrently administered the Neuro-QoL Lower Extremity Function Short Form (Neuro-QoL) as a PRO measure of mobility.¹⁴ Only individuals classified as definite OPMD by clinical or genetic criteria were included, as previously described.¹⁵ Because many items in the SWAL-QOL are worded with reference to “your swallowing problem,” we excluded individuals who denied dysphagia at the time of survey administration.

We evaluated test-retest reliability in a subset of 20 participants who were part of a broader OPMD outcome measures study at our center and completed 2 administrations in person separated by 2–8 weeks. This time period was selected to be long enough to prevent recall, but short enough to expect that no clinical change occurred. We allowed repeat visits up to 8 weeks apart, because some participants lived long distances from our center.

We used clinical data collected at time of survey administration to define disease severity groups using several binary variables:

1. “treatment received”: defined as history of at least 1 therapeutic intervention for dysphagia (esophageal dilatation, cricopharyngeal myotomy, or botulinum toxin injection to the cricopharyngeus);

2. “prolonged mealtime”: defined as mealtime ≥ 30 minutes¹⁶, based on patient-reported duration of an average meal;

3. “dysphagia duration ≥ 10 yr”:– based on patient-reported onset of dysphagia;

4. “aspiration pneumonia”: defined as a patient report of at least 1 episode of aspiration pneumonia; and

5. “underweight”: defined as BMI < 18.5 kg/m², based on patient reported height and weight.

This study was approved by the University of New Mexico Human Research Protections Office. All individuals provided written informed consent to participate.

Surveys and Scoring

The SWAL-QOL (44 items) consists of 10 scales (30 items) and a 14-item Dysphagia Symptom Battery (DSB) intended to assess the severity of dysphagia symptoms. Patients are asked to respond to each item based on experiences during the past month. The 10 scales of the SWAL-QOL are: burden, eating duration, eating desire, food selection, fear, mental health, social functioning, communication, sleep, and fatigue. The first 8 of these scales are considered dysphagia-specific, while the last 2 are considered generic QoL scales, based on a principal components analysis by the instrument’s developers.¹² Responses to each SWAL-QOL item are provided on a 5-point Likert scale. Items within each QoL scale and the DSB are averaged, then linearly transformed to a score of 0–100, with lower scores indicating greater impairment.¹² We derived a Composite SWAL-QOL score (“Composite Score”) following the method of Plowman-Prine, et al.,¹⁷ that is an average of the 10 scale scores, and which excludes the DSB (which is not considered a scale by the SWAL-QOL developers).^12,18

The Neuro-QoL is a patient-reported measure of lower extremity dysfunction. It consists of 8 items with responses elicited on a 5-point Likert scale. Item responses are summed and converted to standardized scores using conversion tables. Higher scores indicate better function.¹⁴ Evidence for reliability and validity have been found in other disease populations, including multiple sclerosis and Parkinson disease.^19,20 In a previous study of OPMD, we found that the internal consistency reliability of the Neuro-QoL was excellent (Cronbach α=0.94), and that the Neuro-QoL showed preliminary evidence of construct validity in OPMD by correlation with duration of leg weakness and by differentiating between the group that used assistive devices for ambulation and the group that did not.¹⁵

Score distributions

We used descriptive statistics to report measures of central tendency, range of scores, and proportion of respondents with the minimum (floor) or maximum (ceiling) score. We described score distributions for each item (“item-level analysis”), and for each SWAL-QOL scale, the DSB, and the Composite Score (“scale-level analysis”). In the scale-level analysis, we tested for normality using the Shapiro-Wilk test. Floor/ceiling effects were considered to be present if >15% of participants achieved the minimum or maximum possible score.²¹ We classified scores using the following rubric: mild or no impairment in QoL (71–100), moderate impairment (50–70), and severe impairment (0–49), following a prior study.²²

Associations with personal and clinical characteristics

We evaluated the correlations between each SWAL-QOL scale and age and dysphagia duration.

Dysphagia symptom burden

We calculated the prevalence of “high burden” dysphagia symptoms from responses on the DSB. We defined a symptom as high burden if the participant reported the symptom present “almost always” or “often.”

Reliability

Internal consistency reliability was evaluated using the Cronbach alpha coefficient. A Cronbach alpha >0.70 was considered satisfactory.²³ Test-retest reliability for repeated administrations was measured by intraclass correlation coefficients (ICC) using a 2-way random effects model, with 95% confidence intervals (C.I.).²¹ ICCs >0.70 were considered acceptable.²³

Validity

Validation of a measure in a given context is an ongoing process. Current views of validity suggest use of a unitary concept of construct validity that entails incorporation of various sources of evidence to judge whether a measure produces data that are valid.^24,25 Sources of such evidence include, but are not limited to:²⁴ 1) factor structure; 2) relations to other variables (a source of evidence for which we here use the terms convergent and discriminant validity); and 3) consequences of scores (a source of evidence for which we here use the term known groups validity).

Factor structure

We performed confirmatory factor analysis (CFA) to test the hypothesis that the underlying putative structure of the SWAL-QOL, i.e., that 30 items in the instrument represent 10 independent factors (“scales”), was consistent with the data in our sample of OPMD patients. CFA was conducted using the maximum likelihood extraction method. We used 3 measures of overall goodness of fit, namely, the χ² test, comparative fit index (CFI),²⁶ and root-mean-square error of approximation (RMSEA).²⁷ Given that our sample size was rather small for factor analysis, we included the CFI because it is considered an index that performs well even with small sample sizes.²⁶ The following were considered indicators of acceptable model fit: χ² P-value>0.05, CFI >0.90, and RMSEA <0.08.

We performed a second CFA to determine whether the 10 scale scores can be averaged into a single score (the “Composite Score”).

Relations to other variables

Convergent validity

In the original SWAL-QOL validation study, the DSB was used to test the clinical validity of the 10 SWAL-QOL scales by evaluating whether dysphagia symptom severity was associated with scale scores.¹² For convergent validity, we hypothesized that the DSB has a strong, positive correlation (r≥0.50) with each dysphagia–specific scale. Because OPMD causes lower limb weakness and the fatigue scale includes an item “feel weak?” we hypothesized a strong correlation between the fatigue scale and the Neuro-QoL.

Discriminant validity

We hypothesized weak to moderate positive correlations (i.e., r<0.5) between all scales (except fatigue) and the Neuro-QoL. We hypothesized weak positive correlations between the DSB and generic QoL scales.

Consequences of scores

Known groups validity

We compared Composite Scores between groups expected to differ in dysphagia severity. We hypothesized that the treatment received, prolonged mealtime, dysphagia duration ≥ 10 yr, aspiration pneumonia, and underweight groups would have lower scores than the treatment-naïve, normal mealtime, dysphagia duration < 10 yr, non-aspiration pneumonia, and non-underweight groups.

We tested 24 a priori hypotheses assessing correlations between SWAL-QOL scores and other measures and assessing differences between 5 known groups. See Supplementary Table S1 (available online) for a list of all hypotheses. Following Terwee et al., we considered acceptance of at least 75% of hypotheses as demonstrating support for construct validity.²¹ If sample sizes within groups for a given “known groups” hypothesis did not provide 80% power on the t-test to detect a group difference of at least 10 on the Composite Score (which represents a medium effect size; Cohen d ~0.50),²⁸ we excluded that hypothesis when determining the percentage of hypotheses that were confirmed.

Statistical analyses

We used the Pearson product-moment correlation coefficient r to measure correlations. Correlations were considered weak if |r| <0.3, moderate if 0.3≤|r| <0.5, strong if 0.5≤|r| <0.7, and very strong if |r| ≥0.7. We used the 2-sample t-test to compare group means. All tests were 2-tailed, and the level of significance was set at α=0.05. Analyses were performed using STATA (Stata-Corp LP, College Station, TXs) and SPSS (SPSS, Chicago, IL).

RESULTS

Patient characteristics

One-hundred sixteen registry participants met the inclusion/exclusion criteria. Of these, 113 (97%) completed SWAL-QOL surveys. Participants resided in 25 U.S. states, including 39% in New Mexico. Table 1 shows the personal and clinical characteristics of the sample. Ninety-six (85%) completed the Neuro-QoL. Twenty of 113 participants (mean age 62.4±12 years, 9 men) completed test-retest assessment (with mean time between SWAL-QOL administrations of 34±12 days). No participant underwent a dysphagia therapeutic intervention between repeated administrations.

Table 1.

Personal and clinical characteristics of participants (n=113)

	Mean (SD) or N (%)
Age, yr	65.8 (8.0)
Age at disease onset (ptosis or dysphagia), yr	50.3 (8.0)
Men	55 (48.7)
Genetically diagnosed	71 (62.8)
Clinically diagnosed	42 (37.2)
Dysphagia duration, yr	12.4 (8.1)
Gastrostomy	0 (0)
Assistive device for ambulation	47 (41.6)
Body mass index (kg/m2)	25.0 (4.5)

SD=standard deviation

N=number of participants

Score distributions

Item-level analysis

Results are shown in Supplementary Table S2 (available online). The rate of missing responses was very low (0.4%, 20 of 4972 total items in returned surveys). The full range of scores (1–5) was observed for all items except 1 (“food sticking in your throat”, for which the range was 1–4). Almost all items (40/44) demonstrated floor/ceiling effects.

Scale-level analysis

Four scales, burden, eating duration, eating desire, and social functioning, were not normally distributed. The scales with the lowest mean scores were eating duration, burden, and fatigue which showed severe impairments (Table 2). The mean Composite Score was 54.4±20.7, indicating moderate impairment in SR-QoL. The majority of scales (7 of 10) demonstrated floor/ceiling effects. The most prominent ceiling effect was in the eating desire scale, which had the highest mean score and in which 51% of participants scored at the maximum (Table 2).

Table 2.

Distribution of scores and floor/ceiling effects for the SWAL-QOL scales, DSB, and Composite Score.

	Mean (SD)	Score at floor N (%)	Score at ceiling N (%)
Scale
Burden	42.0 (31.2)	20 (17.7)	3 (2.7)
Eating duration	31.0 (29.3)	35 (31.0)	5 (4.4)
Eating desire	83.3 (23.7)	2 (1.8)	58 (51.3)
Food selection	60.8 (29.0)	6 (5.3)	18 (15.9)
Fear	55.3 (25.0)	3 (2.7)	2 (1.8)
Mental health	52.3 (30.4)	7 (6.2)	9 (8.0)
Social	62.9 (31.2)	7 (6.2)	25 (22.1)
Communication	53.7 (32.1)	11 (9.7)	19 (16.8)
Sleep	58.0 (32.0)	11 (9.7)	22 (19.5)
Fatigue	45.1 (27.4)	9 (8.0)	4 (3.5)
DSB	51.8 (18.6)	1 (0.9)	0 (0.0)
Composite Score	54.4 (20.7)	0(0.0)	1(0.9)

SD=standard deviation

N=number of participants

DSB=Dysphagia Symptom Battery

Associations with personal and clinical characteristics

Table 3 shows correlations of scales with age and dysphagia duration. Age had a weak negative correlation with the eating duration scale (lower eating duration score indicates longer time to complete meals), and a weak positive correlation with sleep. Age did not correlate significantly with any other scales. Dysphagia duration had weak negative correlations with burden, eating duration, fear, mental health, and communication scales.

Table 3.

Correlations of SWAL-QOL with personal and clinical characteristics (n=113)

	Age		Dysphagia duration
	r	P-value	r	P-value
Scale
Burden	−0.09	0.36	−0.29	<0.01
Eating duration	−0.27	<0.01	−0.30	<0.01
Eating desire	0.03	0.76	−0.09	0.36
Food selection	−0.07	0.46	−0.14	0.13
Fear	−0.02	0.85	−0.20	0.03
Mental health	−0.04	0.65	−0.21	0.02
Social functioning	−0.07	0.44	−0.13	0.15
Communication	−0.18	0.05	−0.20	0.03
Sleep	0.20	0.03	0.00	0.98
Fatigue	0.07	0.49	−0.11	0.25

r=Pearson correlation coefficient

Dysphagia symptom burden

More than half of respondents reported the following symptoms almost always or often: “having to clear your throat,” “having thick saliva or phlegm,” “food sticking in your throat”, and cough (Figure 1). The least frequent symptoms were: “food or liquid coming out your nose,” “food or liquid dribbling out your mouth,” “drooling,” and “problems chewing.” These infrequent symptoms showed very high ceiling effects (ranging from 34–48%) in the item-level analysis (Table S2). The prevalence of pharyngeal symptoms was greater than that of oral symptoms.

FIGURE 1.

Prevalence of “high burden” dysphagia symptoms in the sample.

Reliability

Internal consistency reliability was excellent for all SWAL-QOL scales and the Composite Score, with Cronbach alpha coefficients ranging from 0.80–0.96 (Table 4). The Cronbach alpha for the whole SWAL-QOL instrument (all 44 items) was 0.97. Test-retest reliability was generally good, with ICCs >0.70 for all scales except 1, social functioning, which had a low ICC of 0.46 (Table 4).

Table 4.

Internal consistency and test-retest reliability.

	Number of items	Cronbach alpha	Retest reliability (ICC) (95% CI)
SWAL-QOL scale
Burden	2	0.90	0.89 (0.76, 0.96)
Eating duration	2	0.82	0.89 (0.74, 0.95)
Eating desire	3	0.80	0.72 (0.42, 0.89)
Food selection	2	0.88	0.74 (0.43, 0.89)
Fear	4	0.85	0.74 (0.45, 0.89)
Mental health	5	0.95	0.77 (0.50, 0.90)
Social functioning	5	0.95	0.46 (0.03, 0.75)
Communication	2	0.94	0.88 (0.71, 0.95)
Sleep	2	0.89	0.78 (0.52, 0.90)
Fatigue	3	0.91	0.84 (0.64, 0.93)
Composite Score	30	0.96	0.84 (0.61, 0.94)

ICC= Intraclass correlation coefficient

CI=confidence interval

Validity

Factor structure

CFA showed that although the significance level (P<0.0001) of the χ² test was less than the cutoff of 0.05, our other fit indices were acceptable. The CFI value was 0.94, and the RMSEA estimate was 0.07 (95% C.I.: 0.06, 0.08). These results indicate that our data are consistent with the putative 10-factor structure of the SWAL-QOL.

CFA for the Composite Score demonstrated that the 10 scales loaded on 1 factor, with correlations between each scale and a single factor ranging from 0.44–0.88. These results show that a 1 latent variable model is adequate and support our use of a single Composite Score derived by averaging the 10 scale scores.

Relations to other variables

Convergent and discriminant validity

Convergent validity was demonstrated by the strong correlations between the DSB and nearly all dysphagia-specific scales (Table 5). Dysphagia symptom severity was most strongly correlated with the fear scale. There was a strong correlation between the Neuro-QoL and the fatigue scale. For discriminant validity, a weak correlation was found between dysphagia symptom severity and the sleep scale. Weak to moderate correlations were found between the Neuro-QoL and 7 of the 8 dysphagia-specific scales (Table 5).

Table 5.

Correlations of SWAL-QOL with DSB and Neuro-QoL

SWAL-QOL scale	DSB (n=113)		Neuro-QoL (n=96)
	r	P-value	r	P-value
Burden	0.58	<0.0001	0.34	<0.001
Eating duration	0.57	<0.0001	0.16	0.11
Eating desire	0.45	<0.0001	0.26	0.01
Food selection	0.64	<0.0001	0.41	<0.0001
Fear	0.70	<0.0001	0.26	0.01
Mental health	0.62	<0.0001	0.33	<0.001
Social functioning	0.56	<0.0001	0.37	<0.001
Communication	0.57	<0.0001	0.38	<0.001
Sleep	0.27	<0.01	0.28	<0.01
Fatigue	0.33	<0.001	0.58	<0.0001

r=Pearson correlation coefficient

DSB= Dysphagia Symptom Battery

Consequences of scores

Known groups validity

We observed lower mean Composite Scores in all 5 groups expected to have more severe dysphagia based on external criteria, but only 3 of the group differences were statistically significant (Table 6). The sample sizes for the aspiration pneumonia and underweight groups, however, were small (n=27 and n=5, respectively), and we thus had low power to detect statistically significant group differences.

Table 6.

Known groups validity. Comparison of Composite Scores for dysphagia severity groups.

	Mean (SD)	P-value	Cohen d
Treatment received
Yes (n=37)	47.8 (20.1)	0.02	0.49
No (n=76)	57.7 (20.4)
Prolonged mealtime*
Yes (n=78)	51.0 (21.0)	0.01	0.59
No (n=32)	62.9 (18.3)
Dysphagia duration ≥ 10 yr
Yes (n=65)	50.4 (19.6)	0.02	0.47
No (n=48)	59.9 (21.2)
History of aspiration pneumonia†
Yes (n=27)	49.5 (22.5)	0.11	0.37
No (n=75)	57.0 (19.7)
Underweight
Yes (n=5)	39.4 (32.0)	0.10	0.77
No (n=108)	55.1 (20.0)

SD=standard deviation

^*Three participants had missing responses.

^†Eleven participants had missing responses.

Overall, the data confirmed 86% (19/22) of our hypotheses, supporting the construct validity of the SWAL-QOL in OPMD (Table S1).

DISCUSSION

In this large cross-sectional sample of OPMD participants, we found that SR-QoL is moderately reduced, and that burden, eating duration and fatigue were the most severely affected domains of SR-QoL. Prolonged meal duration has previously been recognized as a factor influencing QoL in OPMD patients, because it contributes to social withdrawal and decreased enjoyment of meals.²⁹ Mean composite SWAL-QOL score in OPMD participants was 54.4±20.7, which is worse than the mean score (75.9±19.4) of 70 patients with amyotrophic lateral sclerosis (ALS). ³⁰

These results expand on our previous work documenting reduced general health-related QoL (HRQoL) in OPMD.¹⁵ In that study, dysphagia severity had stronger associations with mental than with physical components of HRQoL. The current data support these prior findings by demonstrating that dysphagia severity is strongly associated with the fear, mental health, and social functioning scales of the SWAL-QOL. The roles eating and drinking play in social functioning,³¹ and embarrassment or anxiety related to coughing, choking, or aspiration events²⁹ likely contributed to our findings of a strong association between these mental and social domains of the SWAL-QOL and patient-reported dysphagia symptom severity.

Certain patient-reported dysphagia symptoms were highly prevalent in OPMD, including frequent throat clearing, thick saliva or phlegm, food sticking in the throat, and cough. Moreover, 1 symptom was endorsed by every respondent in the sample: food getting stuck in the throat. This symptom corroborates the inefficient swallowing profiles noted in OPMD patients due to insufficient lingual and pharyngeal pressure to efficiently propel the bolus from the oral cavity to the esophagus, leading to residue or stasis of material in the vallecular and/or pyriform sinuses.³² We have found in our clinical experience that patients adapt to food sticking in the throat by drinking fluids with meals. This strategy facilitates passage of the bolus; however, the technique becomes less effective as dysphagia progresses and more effort is required in the setting of reduced functional physiologic reserve.

There are currently no consensus guidelines on clinical management of OPMD. As McHorney et al. noted, symptoms elicited by the SWAL-QOL may be responsive to clinical interventions.¹² Based on our profile of the dysphagia symptom burden in OPMD, interventions that thin saliva or phlegm could be considered in the clinical management of OPMD. The pattern of dysphagia symptoms shows that pharyngeal symptoms (e.g., throat clearing, coughing, and food sticking in the throat) are more prevalent and burdensome than oral symptoms (e.g., drooling, problems chewing, and food dribbling). This is consistent with physiologic studies that have documented very weak pharyngeal contractions and delayed pharyngeal emptying in OPMD.^2,33,5

We found strengths in certain psychometric attributes of the SWAL-QOL. All SWAL-QOL scales demonstrated satisfactory internal consistency reliability, and most SWAL-QOL scales demonstrated adequate test-retest reliability in assessing OPMD. Similar results were found in a psychometric study of the SWAL-QOL in patients with oropharyngeal cancer.³⁴ The only scale in our study that showed less than satisfactory test-retest reliability was social functioning (ICC 0.64). In the original SWAL-QOL validation study, social functioning had an ICC of 0.89 with 2-week retesting.¹² Because the social impact of dysphagia in OPMD may be related to instances of embarrassing public choking²⁹, it is possible that the mean time between survey administrations of approximately 1 month in our study allowed for such events to take place nearer one administration or the other and thus impacted our estimate of the short-term stability of this scale.

The SWAL-QOL showed evidence of construct validity in OPMD. The theorized 10-factor structure of the SWAL-QOL is consistent with our data. The data supported 86% of our hypotheses, meeting the criterion of Terwee et al.²¹ Dysphagia symptom severity demonstrated moderate to strong correlations with all scales except sleep and fatigue, suggesting that most of the SWAL-QOL scales measure what they intend to measure, i.e. dysphagia-specific QoL. Moreover, the Composite Score discriminated between known groups expected to differ in dysphagia severity (Table 6).

We identified several weaknesses of the SWAL-QOL when applied to an OPMD patient population. The first weakness is that 7 of 10 scales showed floor/ceiling effects (Table 2). The SWAL-QOL was constructed with the intent that scales have no floor/ceiling effects.⁹ In certain cases, floor/ceiling effects are an undesirable psychometric attribute, because they imply that an instrument cannot discriminate at the lowest or highest ends of a scale. Such effects reduce measurement precision. In other cases, floor/ceiling effects could reflect real clinical phenomena (i.e., true scores are at the minimum or maximum of a scale). We believe both situations are represented in our study. Because none of the participants in our sample had undergone gastrostomy for nutritional support, it is possible that our sample contained relatively healthier OPMD subjects (without the greatest severity of dysphagia and dysphonia) and therefore showed ceiling effects for some scales, such as food selection and communication. Moreover, the large ceiling effect in the eating desire scale likely reflects the fact that appetite for food is preserved in OPMD patients despite the progressive loss of swallowing function. In contrast, a study of the SWAL-QOL in ALS patients found moderate impairments in eating desire.³⁵ The ceiling effect we observed in the social functioning scale, however, is likely a psychometric weakness rather than a reflection of a real clinical phenomenon; it is known that social participation is impaired in OPMD patients.^15,36 Two scales (burden and eating duration) showed floor effects. The floor effects in these scales likely indicate a psychometric weakness rather than real clinical phenomena since we would expect the SWAL-QOL to discriminate degrees of impairment in burden and eating duration among subjects who scored low in these domains, given that none of the subjects had progressed to feeding tubes, and only 5 were underweight.

The second weakness of the SWAL-QOL is that it was not developed using a sample of OPMD patients. Therefore, an important threat to its validity when applied to OPMD is construct underrepresentation (inadequate sampling of the content domain),²⁴ meaning that there may be specific domains of SR-QoL pertinent to OPMD that are not addressed by the SWAL-QOL.

The third weakness of the SWAL-QOL is that several of its scales cannot be construed as dysphagia-specific in OPMD. Two scales are actually generic QoL scales (sleep and fatigue), and the items in these scales make no reference to dysphagia (e.g., “Feel weak?”; “Have trouble falling asleep?”). Moreover, the weak correlation between the sleep scale and dysphagia symptom severity (r=0.27) suggests that the sleep scale may not be highly pertinent to the construct of dysphagia-related QoL in OPMD. We also doubt that the fatigue scale is dysphagia-specific in OPMD. Given that OPMD causes limb myopathy and impairs mobility,³⁷ the construct of fatigue in OPMD will likely be influenced by limb weakness. This is supported by the stronger correlation between the Neuro-QoL and the fatigue scale (ρ=0.58) than between dysphagia severity and the fatigue scale (r=0.33). A recent study found that fatigue severity in OPMD is related to functional mobility impairments.³⁶ Another SWAL-QOL scale that is likely not dysphagia-specific in OPMD is the communication scale that concerns speech and intelligibility, rather than dysphagia per se. Because dysphonia is a known feature of OPMD,³⁸ the communication scale will likely reflect communication consequences of dysphonia and cannot be considered dysphagia-specific. The questionable dysphagia specificity of the sleep, fatigue, and communication scales in OPMD is also supported by the finding of no correlation between dysphagia duration and the sleep and fatigue scales, and only a weak correlation between dysphagia duration and the communication scale (Table 3). In line with our view, other researchers using the SWAL-QOL have questioned the dysphagia-specificity of the sleep, fatigue, and communication scales and have excluded these scales when calculating a total SWAL-QOL score.^34,39

When interpreting these findings, several limitations should be considered. First, we did not compare SWAL-QOL scores with a physiologic measure of dysphagia, but rather utilized the symptom battery to index dysphagia status similar to previous reports.¹⁷ A study that evaluated the association between measures of bolus flow on videofluoroscopic swallow studies and the SWAL-QOL in a group of dysphagic patients with heterogeneous diseases found only modest correlations, suggesting that physiologic measures and PROs “provide distinct yet complementary information about oropharyngeal dysphagia.”¹⁸ Second, we could not assess responsiveness of the SWAL-QOL to change with disease status over time, because our study was cross-sectional. Third, although registry participants resided in 25 states across the U.S., it is not known to what extent our sample represents the worldwide OPMD population. Finally, although we used CFA to support our use of a derived Composite SWAL-QOL Score for group comparisons, in practice, various “total” scores for the SWAL-QOL have been generated and used in prior studies.^40,39 Because studies have used different methods for calculating a global score, we urge caution when comparing results across studies.

In summary, SR-QoL is reduced in OPMD, with a higher burden of pharyngeal stage symptoms. Although we found evidence for reliability and validity of the SWAL-QOL in assessing OPMD, the SWAL-QOL has several weaknesses when applied to this patient population, including floor/ceiling effects of some scales that may impact its discriminative power, and the sleep, fatigue, and communication scales that do not specifically reflect dysphagia-related QoL in OPMD. We recommend that an improved dysphagia PRO be developed for swallowing-related outcomes in OPMD.

Abbreviations

ALS

amyotrophic lateral sclerosis

CFA

confirmatory factor analysis

CI

confidence interval

DSB

Dysphagia Symptom Battery

HRQoL

health-related quality of life

ICC

intraclass correlation coefficient

Neuro-QoL

Neuro-Qol Lower Extremity Function Short Form

OPMD

oculopharyngeal muscular dystrophy

PRO

patient-reported outcome measure

QoL

quality of life

SR-QoL

swallowing-related quality of life

SWAL-QOL

Swallowing Quality of Life instrument