Relationship Between Subjective Questionnaires and Videofluoroscopy of Dysphagia Evaluation: A Systematic Review and Meta-Analysis

ABSTRACT Background Early detection of dysphagia is important for preventing aspiration pneumonia. Although videofluoroscopy is currently the primary diagnostic tool for dysphagia, access to this tool may be limited because of radiation exposure risk, high cost, and other factors. Purpose In this study, a meta-analysis was used to determine the strength of the correlation between dysphagia detection outcomes obtained using subjective questionnaires and videofluoroscopy. Methods The PubMed and Embase databases were searched for original articles up to December 2022. Studies published in English that used cross-sectional designs to assess the correlation between subjective questionnaires and videofluoroscopy were considered eligible for inclusion. The search terms used included “dysphagia,” “questionnaire,” and “videofluoroscopy.” Two reviewers critically appraised and extracted the correlation coefficient r values. In addition, a random-effects meta-analysis was conducted. The Q statistic was used to assess the heterogeneity among the included studies. Publication bias was checked using the funnel plot and Egger's tests. Multilevel analysis was used to determine sensitivity to consider within-study correlations. In addition, subgroup analyses were conducted based on type of questionnaire, head and neck cancer, and English-speaking regions. Results The meta-analysis included five studies and 856 patients using the Eating Assessment Tool-10 and one study and 27 patients using the Sydney Swallow Questionnaire. The results of the random-effects meta-analysis showed a moderate relationship between the subjective questionnaires and videofluoroscopy (r = .35, 95% CI [0.20, 0.48]). Similar results were also obtained using multilevel analysis (r = .34, 95% CI [0.25, 0.42]). No publication bias was found for any of the studies (p = .88). In the subgroup analyses, a moderate relationship between Eating Assessment Tool-10 and videofluoroscopy (r = .31, 95% CI [0.19, 0.42]) and an ultrahigh relationship between Sydney Swallow Questionnaire and video-fluoroscopy (r = .74, 95% CI [0.50, 0.87]) were found. Furthermore, moderate associations were observed within each head and neck cancer and English-speaking regions subgroup. However, no significant differences were found between these two subgroups. Conclusions These results indicate the subjective questionnaires considered in this study share a moderate relationship with videofluoroscopy. Subjective questionnaires may be used as an auxiliary tool by nurses and homecare givers for the early assessment of dysphagia risk in patients.


Introduction
Dysphagia is a common problem in patients who are elderly, were involved in a cerebral vascular accident, have head and neck cancer (HNC), or have an age-related degenerative disease such as Alzheimer's disease (Sadakane-Sakuramoto et al., 2021;Thiyagalingam et al., 2021).Dysphagia is known to increase morbidity and mortality and cause severe problems, the most severe of which is aspiration pneumonia (Yoshimatsu et al., 2022).Early dysphagia detection may help nurses and healthcare givers to become aware of this problem in advance and refer affected patients for further evaluations and treatments (Feng et al., 2019;Simões et al., 2020).
Dysphagia evaluation includes objective and subjective assessments (Bartlett et al., 2022;Seo et al., 2021).Videofluoroscopy, a radiographic procedure, is one of the most widely used of these assessments (Huang et al., 2023).Penetration-Aspiration Scale (PAS) scores measured using videofluoroscopy can help detect airway invasion (Borders & Brates, 2020).PAS scores range from 1 to 8, which respectively correlate to minor (1) to severe (8) degrees of airway invasion (Borders & Brates, 2020).The severity of dysphagia symptoms is descriptive of the risk of bolus penetration into the airway (Alkhuwaiter et al., 2022).Subjective assessment tools for dysphagia use self-reported questionnaires such as the Eating Assessment Tool-10 (EAT-10) and Sydney Swallow Questionnaire (SSQ; Szczesniak et al., 2014;Zhang et al., 2023).Total scores for the EAT-10 and SSQ range, respectively, from 0 to 40 and from 0 to 1,700 (Möller et al., 2020;Szczesniak et al., 2014), with higher scores indicating more severe dysphagia symptoms.
Videofluoroscopy visualizes bolus flow and the degree of airway invasion directly (Lee et al., 2020).However, limitations of this technique include risk of radiation exposure and the high cost of related equipment (Stafler et al., 2022).Moreover, videofluoroscopy facilities are usually available only at medical centers, not at nursing homes.In addition, equipment manipulation and image interpretation require a trained radiologist.In contrast, subjective questionnaires are easily implemented by nurses or caregivers.Therefore, an improved understanding of the discrepancies between questionnaires and videofluoroscopy may help nurses and homecare givers select the appropriate subjective questionnaire to evaluate swallowing efficacy.
Although the relationship between subjective questionnaires and videofluoroscopy diagnoses has been investigated (Audag et al., 2019;Bartlett et al., 2022;Zhang et al., 2023), the reported correlation ranges were inconsistent.Therefore, in this study, a meta-analysis and a systematic review were conducted to rigorously assess the correlations between videofluoroscopy and two commonly used subjective questionnaires, EAT-10 and SSQ.

Design
The searching, screening, and reporting of evidence for this meta-analysis and systematic review were based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (Selçuk, 2019).

Search Strategy
The structure of this meta-analysis also followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines (Page et al., 2021).A PubMed search for Medical Subject Headings terms corresponding to "dysphagia," "subjective questionnaire," and "videofluoroscopy" identified the Medical Subject Headings term "deglutition disorder" for dysphagia only, which was also included as a search term in this study.To expand the search to include the use of subjective questionnaires, we searched three dysphagia-related journals (Dysphagia, Journal of Neurogastroenterology and Motility, and Neurogastroenterology and Motility) using the keyword "questionnaire," which returned the EAT-10 and SSQ.Therefore, the keywords "dysphagia," "deglutition disorder," "questionnaire," "subjective questionnaire," videofluoroscopy," "eating assessment tool-10," "EAT-10," "Sydney Swallow Questionnaire," and "SSQ" were used to search the PubMed and Embase databases with an upper cutoff date of December 2022.

Study Selection
After using EndNote X7 reference management software (Clarivate, Philadelphia, PA, USA) to find and remove duplicate articles, the identified articles were screened in two phases.In the first phase, titles and abstracts were assessed for relevance, including the presence of search terms related to dysphagia, subjective questionnaires about dysphagia, and videofluoroscopy.In the second phase, the full text of the articles were evaluated using the inclusion and exclusion criteria.Two independent reviewers double-coded all articles, and any disagreements were resolved through discussion until a consensus was reached.

Inclusion and Exclusion Criteria
The inclusion criteria were studies that (a) investigate the correlation between a subjective questionnaire (EAT-10 or SSQ) and videofluoroscopy, (b) provide Pearson or Spearman correlation coefficients, and (c) were written in English.Exclusion criteria were studies that (a) were published as a review, editorial, protocol, case report, letter to the editor, or qualitative research paper; (b) were published in a language other than English; (c) were duplicate publications; (d) did not provide usable data; and (e) were not a full manuscript (e.g., a poster).

Critical Evaluation of Quality
At least two assessors performed independent assessments of the quality of each included study using the Quality in Prognosis Studies tool (Hayden et al., 2013).For each study, the six domains of the Quality in Prognosis Studies tool, namely "participation," "attrition," "prognostic factor measurement," "confounding measurement and account," "outcome measurement," and "analysis and reporting," were assessed and categorized as posing a low, moderate, or high risk of bias.

Data Extraction and Transformation
Two reviewers performed data extraction from each study independently and listed the variables as follows: author, publication year, Pearson's or Spearman's correlation coefficients, p values, country where the research was performed, and sample size.Because variance in sample correlation coefficients can strongly affect pooled results (Borenstein et al., 2009), the correlation coefficients were transformed to the Fisher's z scale before analysis.After the meta-analysis, these synthesized values were transformed back into their original correlation units using Equation 2.
Equation 1: Fisher's z scale value = 0:5 Â ln 1þr 1−r , where r is the reported correlation coefficients from the studies Equation 2: , where z is the Fisher's z scale value.
The Journal of Nursing Research

Statistical Analysis
Heterogeneity was evaluated using the I-squared index (I 2 ) and examined using the Cochran's Q test.Publication bias was assessed using the funnel plot and Egger's tests.Both fixed-and random-effects models were fitted in this metaanalysis.Values of I 2 < 50% and p > .1 were considered as indicating homogeneous between-study variance and appropriate for use with the fixed-effects model.For values of I 2 ≥ 50% and p ≤ .1, a random-effects model was used to account for the heterogenous effects (Higgins et al., 2003;Huedo-Medina et al., 2006).In addition, subgroup analyses were conducted to assess possible factors of heterogeneity between studies and to further discern the correlations among studies (Leeflang et al., 2008).Furthermore, a multilevel model was used to fit the correlated data within a single study.All of the analyses were conducted using R statistical software (v4.0.3;R Core Team, 2020).

Article Search
The initial keyword search identified 10,065 potentially relevant studies.After screening, 57 were shortlisted as potentially qualified and evaluated using the inclusion criteria.Ultimately, six studies were qualified and included in the meta-analysis.The article search process and reasons for exclusion are shown in -Figure 1.

Critical Evaluation of Quality
The included studies were assessed for quality assessment, and the related risk levels were calculated based on study participation, study attrition, measurement, study confounding, statistical analysis, and reporting (Table 1).All of the included studies were assessed as having low risk values for measurement and statistical analysis, indicating their good reliability.

Characteristics of the Included Studies
The six included studies covered 883 patients and assessed the relationship between the two questionnaire surveys and videofluoroscopy-measured PAS scores.Information on study region, number of patients, raw correlation coefficient, type of subjective questionnaire, and different patient groups (i.e., HNC, stroke, neurological disorder, Parkinson's disease, and dysphagia symptoms) is shown in Table 2.

Meta-Analysis Results
Cumulative sample sizes and pooled correlation with 95% confidence interval (CI) are shown in Table 3.The six articles (n = 883 patients) reported on the correlations between the questionnaire and videofluoroscopy results.The results of the

Figure 1
Search of the Literature Flowchart funnel plot and Egger's tests did not indicate publication bias (Figure 2; p = .88).High heterogeneity (I 2 = 51%, p = .02)was observed among the six articles.The result of the random-effects meta-analysis showed a pooled correlation of .35(95% CI [0.20, 0.48]).In addition, two of the included studies (Arrese et al., 2017;Bofill-Soler et al., 2021) were separated into three parts, with each part treated as a single datum for analysis.As this approach may potentially overestimate the effects of a single study, a multilevel model that incorporated a covariance matrix within the individual studies was used.The results of the multilevel model (pooled correlation = .34,95% CI [0.25, 0.42]) were similar to those shown in Table 3. Subgroup analyses were performed on patient subgroups, namely, HNC status, English speaking status, and questionnaire used (EAT-10 or SSQ).The pooled correlations were .35(95% CI [0.15, 0.52]; Table 4) and .36(95% CI [0.10, 0.57]; Table 4) for patients with and without HNC, respectively, indicating no significant difference ( p = .95).For patients who did and did not speak English, the pooled correlations were .28(95% CI [0.09, 0.45]; Table 5) and .44 (95% CI [0.23, 0.62]; Table 5), respectively, which also indicated no significant difference ( p = .24).Five of the included studies examined the relationship between EAT-10 and videofluoroscopy.The pooled correlation between EAT-10 and videofluoroscopy from these studies was .31(95% CI [0.19, 0.42]; Table 6).Only one of the included studies reported on the correlation between SSQ and videofluoroscopy (.74, 95% CI [0.50, 0.87]; Table 6).A significant difference between these two correlations was identified ( p < .01).

Discussion
The results of this study indicate subjective questionnaires, for example, EAT-10 and SSQ, are capable of detecting aspiration risk in patients with HNC, stroke, and neurological disorders at a moderate level of correlation with PAS scores

The Journal of Nursing Research
Chih-Jun LAI et al.
( .35,95% CI [0.20,0.48]).This moderate correlation was also reflected in the results of the subgroup analyses on patients with and without HNC and those who did and did not speak English.This was the first meta-analysis study to show a correlation between the results of subjective questionnaires and objective videofluoroscopy, with the results highlighting a moderate relationship between PAS scores and subjective questionnaires.Although PAS scores indicate the degree of airway invasion, they do not fully reflect the degree of change in swallowing physiology, which may explain the moderate correlation value found.This study closes an important knowledge gap in this field by providing a well-founded comparison between using questionnaires and using videofluoroscopy.Although the latter is an important objective tool for diagnosing dysphagia, the cost of related equipment makes it impractical for most general healthcare providers, nursing homes, and caregivers not located in medical centers to use.The results indicate that questionnaires, which are highly accessible, noninvasive, and inexpensive, may be used to assess dysphagia severity status to a degree of accuracy that is moderately correlated with the results obtainable using fluoroscopy.In addition, fewer than half of patients with    dysphagia report swallowing symptoms to a healthcare giver or nurse without being directly and systematically asked (Ansari et al., 2020;Cosentino et al., 2022;Ginocchio et al., 2022, Yang et al., 2015;Yee et al., 2020).Referring high-risk patients to medical centers where they can receive further treatment and evaluation after obtaining questionnaire results represents a more effective utilization of medical resources (Ishak et al., 2021).
The relationship between subjective questionnaires designed to assess swallowing disorder severity and the videofluoroscopybased PAS score was the focus of this study, with the primary goal being to understand the severity of dysphagia symptoms and the PAS score, which represent the physiological manifestation of airway invasion (Patterson & Lawton, 2023;Printza et al., 2021).
Subgroup analyses based on HNC status and knowledge of English generated correlation results similar to the overall analysis.HNC status was targeted for subgroup analysis because HNC is widely recognized as a leading cause of dysphagia (Frowen, 2019), and considerable research attention has been focused on patients with HNC.The results in this study indicate that the correlation between subjective questionnaire outcomes and videofluoroscopic findings was not affected by HNC status.In addition, although the questionnaires used in the included studies (EAT-10 and SSQ) were originally developed in English (Belafsky et al., 2008;Wallace et al., 2000), ability to speak English did not affect the correlation between questionnaire scores and videofluoroscopic results.This is likely because of the concise nature of the questionnaires, which feature a small number of questions that are clear, acceptable, and easy to answer.

Strengths and Limitations
The self-report questionnaires used in the included studies showed moderate and acceptable correlations with the PAS scores measured using videofluoroscopy.However, the questionnaires were subject to several limitations.First, the questionnaires cannot fully replace videofluoroscopy and should only be used to provide a preliminary assessment of dysphagia severity.More detailed examinations of the mechanisms behind dysphagia diagnoses should be conducted using videofluoroscopy.Second, the analysis in this study contained only a single study that did not use the EAT-10 questionnaire.The authors hope to investigate the relationship between the SSQ and videofluoroscopy further in the future.

Conclusions
This systematic review and meta-analysis assessed the correlations between the outcomes of subjective questionnaires (EAT-10 and SSQ) and objective videofluoroscopy.We found a moderately strong relationship between the two assessment approaches.Nurses and homecare givers may reference our results to apply acceptable swallowing screening questionnaires to the early detection of at-risk patients with dysphagia.

Figure 2 Funnel
Figure 2 Funnel Plot and Egger's Tests

Table 2
Characteristics of Included Studies Note.EAT = Eating Assessment Tool; HNC = head and neck cancer; s/p = status post; SSQ = Sydney Swallow Questionnaire.

Table 1
Risk of Bias Summary

Table 3
Correlations Between Subjective Questionnaires and Videofluoroscopy

Table 4
Correlation Between Subjective Questionnaires (Patients With vs. Without Head and Neck Cancers) and Videofluoroscopy

Table 6
Correlation Between Subjective Questionnaires (Eating Assessment Tool-10 vs. Sydney Swallow Questionnaire) and Videofluoroscopy