Dyspnea Index: An upper airway obstruction instrument; translation and validation in Swedish

Abstract Objective Upper airway dyspnoea is a challenging condition in which assessing the discomfort experienced by the patient is essential. There are three patient‐reported outcome (PRO) instruments developed particularly for this patient group, none of which is available in Swedish. The aim of this study was to translate the Dyspnea Index (DI) into Swedish and validate the instrument for use in the Swedish‐speaking population by investigating its basic psychometric properties. Design A prospective instrument validation study. Setting Tertiary referral centre. Participants Fifty‐three (n = 53) patients with upper airway dyspnoea and 19 healthy controls. Main outcome measures The questionnaire was translated into Swedish (swDI) with a forward‐backward method. Reliability, repeatability, responsiveness and construct validity were assessed by asking the subjects to complete the swDI, a visual analog scale (VAS) at exertion and at rest and the Voice Handicap Index (VHI). Results The swDI showed excellent internal consistency (Cronbach's α: 0.85) and repeatability (interclass correlation coefficient: 0.87 and Pearson's r: .89) in the patient group. No ceiling effect was observed (maximum score achieved was 39; 85% of the patients scored ≤ 36). SwDI scores moderately correlated with VAS at exertion (r: .59) and at rest (r: .42), yet poorly with the VHI (r: .36). The effect size (ES) was 3.8. Conclusions The swDI is a valid, robust and reliable questionnaire for self‐assessment in Swedish‐speaking patients with upper airway obstruction. A future anchor‐based longitudinal study is needed to assess the smallest detectable change (SDC) and minimal important change (MIC) that were not estimated in our study.


| INTRODUC TI ON
Upper airway dyspnoea is difficult to evaluate by means of any functional examination without considering the severity of the symptoms reported directly by the patient. The extent of discomfort may affect the type and priority of the intervention both in acute and recurring conditions. There are several ways to assess dyspnoea clinically (eg flexible video laryngoscopy and computed tomography) as well as functionally (spirometry). [1][2][3] The severity of the symptoms due to obstruction of the upper airway may vary. A structural obstructive lesion (eg laryngotracheal stenosis, tumours, and vocal fold ankylosis) normally presents with constant breathing problems at exertion but also at rest in severe cases. 1,2 Functional or other non-structural disorders (eg paradoxical vocal fold movement (PVFM) and laryngospasm) occasionally occur in asymptomatic patients upon exposure to a possible triggering factor. 2,4 Neurological conditions may appear in a persistent (nerve paralysis or injuries) or intermittent (laryngeal dystonia) manner. 2,3 There are no distinct anatomical margins as far as the lower part of the upper airway is concerned; however, the extrathoracic part of the trachea is commonly included. [5][6][7] In a systematic review, Noud 9,10 The Dyspnea Index (DI) was found to be the only validated questionnaire uniquely developed for and used in various types of upper airway dyspnoea. 1,4,11,12 Considering that none of the aforementioned instruments is available in Swedish, we were prompted to translate and validate the DI.
The DI is a questionnaire initially developed by Gartner-Schmidt et al to quantify the severity of symptoms, particularly in adults with upper airway dyspnoea, 4 which was later validated for adolescents with PVFM. 11 The original English version of DI, shown in Appendix 1, is a Likert scale instrument including 10 items. The respondent is asked to evaluate the frequency of each statement using a 5-point interval scale (0 = never, 1 = almost never, 2 = sometimes, 3 = almost always, 4 = always). The total score ranges from 0 to 40. A higher cumulative sum score is associated with more severe dyspnoea. 4 The purpose of the present study was to translate the DI into Swedish and validate the instrument for use in the Swedish-speaking population by investigating its basic psychometric properties such as reliability in terms of internal consistency, reproducibility, construct validity and responsiveness.

| Ethical considerations
This study was performed in accordance with the Declaration of Helsinki. This human study was approved by Ethics Review Board in Uppsala. All parents, guardians or next of kin provided written informed consent for the minors to participate in this study. All adult participants provided written informed consent to participate in this study.

| Translation
The English version of the DI was translated into Swedish by using the forward-backward method as described below. An authorised translator and two experienced otorhinolaryngologists, native Swedish speakers fluent in English, independently translated the questionnaire to Swedish. The reconciliation of the forward translations on a common pilot version was made by the first and the last author of this manuscript. This pilot version was then translated back to English by another authorised translator who did not participate in the initial forward translation and was not aware of the original version. The back translation was then compared with the original DI. For those items for which the back-translated and the original DI did not match, the choice of words was discussed by the aforementioned authors until an agreement was reached and a final version was reconciled. A pilot group of physicians with various backgrounds and experience was created. The group answered the questionnaire after the backward translation was applied, and their comments were taken into consideration to produce the final Swedish version of the Dyspnea Index (swDI), as seen in Appendix 2.

| Validation
All patients with upper airway obstruction were recruited at the All patients who were planned to undergo an intervention were asked to complete the Swedish version of the Voice Handicap Index (VHI) before and after treatment. The VHI is a robust, extensively used, validated instrument in Swedish for the self-assessment of voice problems. 13 Our intention was to investigate whether VHI score correlates to DI score since phonation and breathing are two different functions diversely affected by conditions in this airway subsite.
Behavioural therapy was offered exclusively to those diagnosed with PVFM by speech-language pathologists with special expertise in this particular condition. The remaining patients were treated surgically with conventional endoscopic procedures under general anaesthesia (CO 2 laser cordectomy, balloon dilatation, excision with cold instruments) by experienced airway surgeons in our department.
Subjects who were incapable of making informed, intelligent and voluntary decisions, under the age of 14, or non-fluent Swedish speakers were excluded from the study.
Our intention was to evaluate swDI psychometric properties in accordance with the COnsensus-based Standards for the selection of the health Measurement INstruments (COSMIN) initiative guidelines. 14

| Reliability
The reliability of the swDI was assessed with Cronbach's reliability coefficient α 15 calculating the interitem internal consistency among the patient group before and after intervention, the control group separately and all study subjects. Eventual floor or ceiling effects were examined. The interclass correlation coefficient (ICC) with individual modelled as a random effect, Pearson's and Spearman's correlation coefficient (r) 15 were used to evaluate the agreement between two repeated measurements facilitating comparisons between the present study and other existing validations 4,11 using the same method.

| Construct validity
Known-group validation was performed by means of an independent samples t test and Mann-Whitney test between the patient group and the control group. Additionally, the subjects in the patient group were asked to subjectively rate their breathing problems at rest and at exertion on a visual analog scale (VAS) in parallel with completing the swDI. The VAS was a 100-mm long, straight, horizontal line where 0 mm represented "no breathing problems at all" and 100 mm represented "worst possible breathing problem." Furthermore, all subjects receiving a treatment were asked to fill in the Swedish version of the VHI. Subsequently, correlations between the swDI and VAS and the swDI and VHI were calculated with Pearson's and Spearman's correlation coefficient in an attempt to evaluate swDI's convergent and discriminant validity, respectively. 15

| Responsiveness
The patient group was requested to fill in the questionnaire before and approximately 2 months following a surgical intervention, or a completed speech-behavioural therapy carried out by a speech therapist depending on the diagnosis. A paired t test calculating the effect sizes (ES) and a Wilcoxon signed-ranks test was then performed to estimate swDI responsiveness.

| RE SULTS
The study population consisted of 72 subjects (Table 1) Table 2. The subjects involved in each part of the validation process are shown in Table 3.

| Reliability
The analysis of internal consistency showed a Cronbach's α of  and Spearman's ρ of 0.95 (P < .001).

| Construct validity
The 53 participants from the patient group (Mean: 28.9, SD: 6.7, SEM: 0.9) scored significantly higher swDI score compared with the  The ES was then calculated to be 3.8 (Figure 2). The related-samples
A summary of the swDI's psychometric properties and the recommended cut-off values according to the COSMIN taxonomy 14 is shown in Table 4.

| Synopsis of key findings
The results of this study demonstrate that the swDI shows good psychometric properties and may be used in clinical praxis as well as in research as a complement to clinical evaluation in the assessment and follow-up of Swedish-speaking patients with upper airway dyspnoea. Our endeavour was to deliver a valid, robust and reliable questionnaire to be used for self-assessment of breathing problems caused by upper airway obstruction in a Swedish-speaking population.

| Strengths of the study
A specific consensus regarding the process of translating a PRO instrument has not been reached. [16][17][18] It is generally recommended to acquire at least 2 versions in the target language from a varied profile of forward translators, with a subsequent panel assessment to reach an agreement of a preliminary adaptation. 15,17 Considering the importance of the language used in defining the context of the translated document, we chose to engage two authorised translators ensuring a professional linguistic approach. Additionally, two experienced specialists in airway problems, the authors excluded, were consulted to maintain the intended clinical prospect in the target language. All physicians participating in this process agreed that there were no culturebound disease implications that would modify the context of the original version. Some items, in particular items 5 and 9 required a lengthy discussion during the translation process, as "stress" is a word with an occasionally negative meaning in Swedish apart from its nuance in American English.

TA B L E 3 Subjects involved in each part of the validation process
However, the use of backward translation, although commonly disputed, 16,19 was used with the purpose of controlling changes altering the original meaning. Furthermore, receiving input from the target population and making appropriate modifications is considered equally important at that phase. [15][16][17]19,20 No major comments were made from the pilot group testing of our final version of the swDI.

| Comparisons with other studies
Subjects interpret the severity or frequency of a symptom when providing their own judgment in PROs differently. Thus, there is always the risk of classification bias. 21  suggest a minimum of 100 subjects to ensure stability of the covariance matrix, whereas Cappelleri et al 24 recommend at least 300 for a thorough individual item evaluation, pointing out that weakly correlated constructs may require more subjects for precise estimates.
Upper airway obstruction is caused by a heterogeneous palette of rare diseases. 1 Örebro University Hospital services a local population of approximately 300,000 and shares the tertiary specialist care expertise of approximately 2 million people with another hospital in the same region. Thus, there is a limited capacity to recruit a sufficient number of study subjects to analyse swDI's psychometric properties with modern statistical methods such as factor analysis or item response theory. Engaging more study subjects would certainly substantiate our findings. Although comparable with previous DI validation studies, 4,11 the lean study sample combined with the heterogeneity of the study subjects' condition and the lack of a sample size calculation based on a power analysis, is undoubtedly considered as a primary limitation of the present study.
Moreover, a discrepancy could be observed between the subjects involved in the different phases of the validation process as seen in Table 3. A second evaluation with DI before intervention is missing in more than half of the patient subgroup due to the fact that these patients were planned for an intervention by other physicians not participating in the study; hence, the only opportunity to fill in the questionnaire was on the day of intervention. Furthermore, one subject in the patient group missed to grade the experienced discomfort with the VAS scale.
As shown in Table 4, our findings are in conformity with COSMIN criteria for good measurement properties. 14

| Clinical applicability of the study
We consider that the swDI is filling the lack of an instrument for Swedish-speaking patients with upper airway dyspnoea and will cer- The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

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