Study Design

This cross-sectional non-randomized prospective study with controls was carried out according to the Declaration of Helsinki and previously approved by the local Institutional Review Board. Written informed consent was obtained for all the participants included in the study. Authors of the original English version of the DHI consented to the completion of the present study.

Four main phases were followed: I-DHI development (phase 1), which encompassed the translation of items for the DHI cross-cultural adaptation into Italian, I-DHI reliability analysis (phase 2), I-DHI validity analysis (phase 3), and I-DHI interpretability analysis (phase 4). The COnsensus-based Standards for the selection of health Measurement INstruments (COSMIN) [37] checklist was followed as a reference for the completion of the present study.

Participants

Participants were selected among the consecutive cohort of patients who were referred for known or suspected OD to two Phoniatrics clinics in Northen Italy between November 2019 and February 2020 and between June and November 2016. In order to examine clinical validity, healthy participants were randomly selected from the community. Inclusion criteria were (i) presence of OD deriving from any etiology, in patients; (ii) no history of OD, related diseases, or feeding tube placement, in healthy participants; (iii) age ≥ 18 years; and (iv) ability to independently read a written text. Both patients and healthy participants were excluded from the study if they presented a state of cognitive decline (i.e., Mini Mental State Examination [38] < 24).

Procedures

Instrumental, clinical, and demographic data were recorded and stored anonymously on the institutional computer’s hard drive, while backup copies were saved on external memory drives. Patients completed the I-DHI and underwent a FEES examination for the assessment of OD in the same session. FEES was performed using a XION EF-N flexible endoscope with a 3.4 mm diameter and a length of 320 mm (XION GmbH, Berlin, Germany) mounted on an EndoSTROBE camera (XION GmbH, Berlin, Germany). Two Phoniatricians with > 20 years of clinical and research experience in dysphagia and FEES conduction carried out the FEES examinations. FEES examinations were conducted using thin liquid (5–10–20 ml of blue-dyed water × 3 trials for each volume; International Dysphagia Diet Standardization Initiative [39]—IDDSI 0; Viscosity: < 50 mPa·s at 50 s−1 and 300 s−1), pureed food (5–10–20 ml of pudding × 3 trials for each volume; IDDSI 4; Viscosity: 2583.3 ± 10.41 mPa·s at 50 s−1 and 697.87 ± 7.84 mPa·s at 300 s−1), and regular food (half biscuit × 2 trials; IDDSI 7 Regular). In the present study, the viscosity levels of liquids and pureed food complied with the descriptors of the National Dysphagia Diet Task Force [40]. All the patients followed the same bolus administration order of liquids (3 × 5 ml, 3 × 10 ml, 3 × 20 ml), puree (3 × 5 ml, 3 × 10 ml, 3 × 20 ml), and solids. If either consistency or volume was considered unsafe to be administered or if severe swallowing efficacy impairment was observed, the FEES protocol was not completed with unsafe volumes/consistencies. Owing to safety reasons, the FEES protocol was interrupted if at least one of the following conditions occurred: (i) laryngospasm during the water test in the clinical assessment; (ii) severe impairment of the oral control of the bolus with pureed food which led to chocking; (iii) and severe impairment of the oral preparatory swallowing stage with solids which prevented the processing of the solid into a bolus. Based on FEES video recordings, OD severity was rated with the Dysphagia Outcome and Severity Scale (DOSS) [41]. The DOSS is a 7-point ordinal scale with scores 1–2 indicating severe and moderately severe OD with non-oral nutrition necessary, scores from 3 to 5 corresponding to moderate, mild moderate, and mild OD with a modified diet on a full per-oral nutrition, and scores 6–7 representing functional and normal swallowing abilities with a normal diet. In addition, swallowing safety was assessed using the Penetration-aspiration scale (PAS) [42, 43]. PAS is an 8-point ordinal scale, with score 1 indicating no penetration and aspiration, score 2 representing transient penetration with ejection, scores 3–5 laryngeal penetration without ejection and/or reaching the vocal folds, and scores 6–8 tracheal aspiration. In accord with recent research studies addressing PAS psychometric properties [44, 45], PAS scores 1–2 were considered to reflect normal swallowing function. Penetration was scored as present with PAS > 2 ≤ 5 [44, 45], while aspiration with PAS > 5 [42]. A PAS score was attributed to each administered bolus. The worst bolus (i.e., the bolus with highest PAS score) for each consistency tested and the worst PAS score among all consistencies was considered for analysis purposes. As appropriate for ordinal measures [46], inter-rater agreement was calculated using the linear weighted kappa coefficient for both DOSS and PAS scales.

Phase 1: I-DHI Development

The original version of the DHI [11] was translated and culturally adapted into Italian, following the 5-stage process described by Beaton et al. [47]. Items of the original DHI questionnaire were independently translated into Italian by a bilingual expert panel composed of two speech and language pathologists and two Phoniatricians familiar with OD patients care with a conceptual translation being preferred over a literal one (stage I: forward translation). An Italian final consensus version was obtained after addressing idiomatic, semantic, and conceptual issues (stage II: synthesis). Three independent professional translators who had no knowledge of the questionnaire translated it back into English (stage III: back translation). Subsequently, conceptual and cultural equivalence of all reports were reviewed by the three translators and the expert panel and a final version of the I-DHI was produced (stage IV: expert committee review). The final version of the I-DHI was pilot-tested on 15 consenting patients with OD who were fluent in both Italian and English and approved, since no major suggestions emerged (stage V: pretesting).

The questionnaire aimed at investigating OD patients’ HRQOL, in terms of reported OD physical symptoms, consequences of OD on social and psychological functioning, and perceived dysphagia severity. Mirroring DHI original version, the I-DHI aimed to distinguish between different levels of HRQOL in adults with OD of etiological heterogeneity.

The final version of the I-DHI consisted of 25 items divided into three subscales (Appendix): (i) the Physical subscale refers to the person’s self-perception of physical discomfort due to OD (9 items), (ii) the Functional subscale reflects the impact of OD on the person’s daily activities (9 items), and (iii) the Emotional subscale represents the person’s affective response to OD (7 items). Participants were invited to state whether each item applied to them “never,” “some of the time,” or “all the time,” scoring 0, 2, and 4, respectively. A total I-DHI score was obtained summing the scores of the subscales (possible range 0–100), with higher values meaning worse HRQOL. On completion of the questionnaire, participants were requested to rate the perceived severity of their swallowing problem using a 1–7-point ordinal scale with 1 = “no problem,” 4 = “moderate swallowing problem,” and 7 = “severe problem.”

Phase 2: I-DHI Reliability

The I-DHI internal consistency (i.e., the degree of the interrelatedness among the items) and test–retest reliability (i.e., the proportion of the total variance in the measurements which is due to ‘true’ differences between patients) [37] were examined. Cronbach’s α was computed for each I-DHI subscale and total score to analyze internal consistency. In order to assess I-DHI test–retest reliability, a randomly selected subsample of 26 patients completed the I-DHI a second time, 2 weeks after the initial assessment. The period of 2 weeks consented a reduction in recall bias and guaranteed stability of patients’ health status between the two assessment sessions. During this period, patients did not undergo any medical, surgical, or behavioral treatment for swallowing. Participants selected for assessing test–retest reliability were compared to the main sample for age, gender, and diagnosis (categorized as neurological disorders, respiratory disorders, head and neck cancer, and other disorders). Demographic and clinical characteristics of this subsample are presented in Table 1.

Table 1 Demographic and clinical characteristics of study populationsPhase 3: I-DHI Validity

I-DHI validity was examined, that is, the degree to which the I-DHI measures the construct(s) it purports to measure (i.e., HRQOL of adults with OD). To this purpose, criterion validity (i.e., the degree to which the scores of an instrument are an adequate reflection of a gold standard) and construct validity (i.e., the degree to which the scores of an instrument are consistent with hypotheses) [37] were tested.

I-DHI Criterion Validity

In order to investigate I-DHI criterion validity, the Italian version of the Swallowing Questionnaire (I-SWAL-QoL) [14] was chosen as the gold standard owing to its satisfactory psychometric properties [9] and wide use in clinical practice and research studies. The I-SWAL-QoL consists of 44 items divided into 11 subscales, namely, the burden of eating difficulty, eating duration, eating desire, symptoms, food selection, communication, fear, mental health, social functioning, fatigue, and sleep. Each subscale encompasses a variable number of items, from 2 to 14. Each item is scored on a 5-point Likert scale from 1 to 5, with lower scores indicating worse HRQOL. In order to assess criterion validity, correlation analyses between I-DHI and I-SWAL-QoL subscales were performed. Inverse correlations were expected, as for I-DHI and I-SWAL-QoL a worse HRQOL is suggested by higher and lower scores, respectively. No previous study investigated DHI criterion validity, and thus no more precise hypotheses on correlational strength could be made. A randomly selected subsample of 62 patients with OD completed the SWAL-QoL for criterion validity purposes. The demographic and clinical characteristics of this subsample are presented in Table 1.

I-DHI Construct Validity

In order to investigate I-DHI construct validity, the following hypotheses were made: (i) direct moderate (ρ = .50–.70) to strong (ρ > .70) correlations were predicted among I-DHI subscales, Total score and self-rated OD severity; (ii) significant inverse correlations were expected between I-DHI and OD severity on FEES scored using the DOSS; (iii) based on Shapira-Galitz et al. findings [34], weak correlations were anticipated between I-DHI and swallowing safety measured with PAS, while significantly higher I-DHI Functional subscale scores were expected for patients with penetration (PAS = 3–5) and/or aspiration (PAS = 6–8) compared to patients with safe swallow; (iv) significantly higher I-DHI subscales and Total scores were anticipated among patients with more severe self-rated OD for all pairwise comparisons (patients were divided into categories based on the I-DHI self-rated OD severity score: 1 = normal swallow, 2–3 = mild swallowing problem, 4–5 = moderate swallowing problem, and 6–7 = severe swallowing problem); (v) as a measure of I-DHI clinical validity, patients with OD were hypothesized to report significantly higher scores for all the I-DHI subscales and the Total score compared to healthy participants.

Phase 4: I-DHI Interpretability

Interpretability is defined as the attribution of qualitative meaning to an instrument’s quantitative scores or change in scores. It is regarded as an important measurement characteristic, although not encompassed among psychometric properties [37]. In order to describe the features of the score distribution, the number of items, the possible and observed score ranges, the observed number of distinct scores, and the median (Interquartile Range—IQR) for each subscale were presented. Floor and ceiling effects were considered present if at least 15% of participants reported the lowest or the highest possible scores, respectively [48]. Norm values (median; IQR) were presented to establish a baseline distribution for I-DHI scores. For both patients and healthy participants, differences among gender and age groups (18–39, 40–64, ≥ 65 years) were investigated.

Statistical Analysis

The present study adopted the quality criteria for measurement properties as defined by Terwee et al. [48]. Data were reported as median (IQR) or as absolute (relative) frequencies. Inter-rater agreement for the FEES outcomes was calculated using the linear weighted kappa coefficient. Kappa values were interpreted as poor (0), slight (0.00–0.20), fair (0.21–0.40), moderate (0.41–0.60), substantial (0.61–0.80), and almost perfect (0.81–1) agreement [46]. The normality of the variable distributions was assessed with the Kolmogorov–Smirnov test. Non-parametric tests were conducted since ordinal variables and non-normally distributed continuous variables were analyzed. Internal consistency was considered satisfactory if Cronbach’s α values were between 0.70 and 0.90. As suggested by Terwee et al. [48], test–retest reliability was assessed using Intraclass Correlation Coefficient (ICC) two-way random effects model, single measures, absolute agreement for the I-DHI subscales and Total score, and with weighted Cohen’s k for the self-rated OD severity subscale. ICCagreement and the weighted Cohen’s k values > 0.70 indicated acceptable test–retest reliability. Correlation analyses were performed using Spearman’s correlation coefficient. Correlations were considered mild for values between 0.30 and 0.50, moderate for values between 0.50 and 0.70, and strong for values > 0.70 [49]. For the assessment of criterion validity, correlational strength was considered satisfactory for values > 0.70 [48]. Comparisons between patients and healthy participants and among groups of patients were assessed with the U Mann Whitney (two groups comparisons) or the Kruskal–Wallis tests (multiple groups comparisons). Multiple comparisons were corrected with Bonferroni’s procedure. For all the analyses, a p-value smaller than 0.05 was considered significant. All the statistical procedures were carried out with IBM SPSS Statistics 26.0® package for Mac (SPSS Inc., Chicago, IL). Missing values were excluded from the analyses so that for each considered variable, only the available data were analyzed (pairwise deletion).