Introduction
Hearing impairment (HI) is one of the most common disabilities of the ageing population affecting over 466 million people worldwide [
1]. One of the main debilitating features of HI is communication difficulties which affect personal relationships and leads to withdrawal from social situations [
2]. HI may be particularly disadvantageous for older adults who have not developed skills to cope with communication difficulties [
3], and as a consequence, HI in the elderly is associated with reduced quality of life, depression, functional decline, lowered self-esteem and social isolation [
4‐
9].
HI is common in people living with dementia [
10], and is one of nine potentially modifiable risk factors [
11]. However, the
mechanism linking HI to dementia remains to be elucidated. One theory is HI may indirectly increase the risk via psychosocial pathways [
12]; people with HI are more likely to feel lonely and socially isolated [
2], and social and emotional dysfunction are independently associated with the risk of dementia. Loneliness has shown to accelerate levels of cognitive impairment [
13] and risk of developing Alzheimer’s disease dementia [
14]. People with poor social networks are 60% more likely to develop dementia compared with those with good social networks [
15].
The International Classification of Functioning, Disability and Health, an international standard set by the World Health Organisation [
16], recognises that many factors are responsible for self-perception of disability. In the case of HI, for any given auditory threshold, there will be a large variability in the level of associated disability [
17]. Many factors must be considered, such as whether the individual is socially active and whether they use hearing aids and additional family support, as well as personal factors including age, background, and lived experiences. For this reason, it should not be assumed that a person with a mild HI would have a milder disability compared with someone with a more severe HI.
Evidence to support this theory comes from hearing aid studies. Management of HI by hearing aids improves quality of life [
18‐
20]. Also, albeit inconsistently, the use of hearing aids has been shown to improve short-term cognitive performance [
21], and attenuate the increased risk of cognitive decline associated with HI [
22].
Tools have been designed to measure hearing-related disability, but are not appropriate for this use for many reasons; they are either now outdated, no longer culturally or technologically relevant, have not been validated on people living with cognitive impairment, or were designed for a different purpose (i.e. as a measure of before and after for hearing aid rehabilitation). The aim of this study was to design and validate a short, easy to administer, culturally relevant questionnaire to measure the social and emotional impact of HI. This tool could then be used to investigate the indirect psychosocial pathway hypothesis linking HI to increased rates of cognitive decline and dementia. This may have a major impact on future public health as a case for more aggressive treatment of HI and social isolation, to reduce the burden and onset of cognitive impairment.
Methods
Development and validation of the Social and Emotional Impact of Hearing Impairment (SEI-HI) questionnaire will be described in four phases. For ease of understanding, demographic characteristics of participants involved in each phase are described in Table
1. Ethical approval was obtained from NRES Committee North East–Newcastle and North Tyneside (ref 170445, 15/NE/0152). All participants gave their informed written consent.
Table 1
Participants characteristics involved in the four phases
Number | 80 | 120 | 95 | 31 |
Age (SD) | 63.55 (11.85) | 57.82 (12.19) | 59.13 (12.67) | 67.94 (9.67) |
Male/female | 28/52 | 36/84 | 36/59 | 20/11 |
YOE (SD) | 14.38 (3.52) | 15.63 (3.04) | 15.47 (3.28) | 12.00 (2.35) |
NH/HI | 45/35 | 120/0 | 57/38 | 12/19 |
Phase 1: Content
Pure tone average (PTAv) thresholds in the better hearing ear were recorded for frequencies at 500, 1000, 2000 and 4000 Hz for each participant. They were then asked to complete the 25-item hearing handicap inventory for the elderly (HHIE) [
23], responding with ‘yes’, ‘sometimes’ or ‘no’ to each question and to expand on their responses verbally, which were noted as ‘free feedback’.
To establish the criteria that were the most commonly reported to be problematic, quantitative responses to each item on the HHIE were analysed and each item was given a rank (from 1 to 24) depending on scoring frequency. For each participant, items with positive responses, (i.e. a response of ‘yes’ or ‘sometimes’) were given a score of 1, which was tallied and the 10 highest scoring items were identified as the most common and pertinent scenarios. The consultation allowed further adaptations of the scenarios to suit the influence of HI on our sample and more generally to devise an assessment procedure suitable for a variety of individuals, with clinical and non-clinical samples. Together, this evidence informed the creation of the 14-item SEI-HI questionnaire (Online resource
1).
Phase 2: Scoring and outcomes
Free feedback from Phase 1 helped to develop the appropriate scoring rating for the questionnaire. Respondents from Phase 1 revealed the need for an ‘in between’ measure as in certain situations, responses were not as clear cut as ‘yes’ or ‘sometimes’. Participants stated: “It is not never, but not as much as sometimes” or “It’s more than sometimes but I wouldn’t say that it was ‘yes’ a definite issue for me.” For this reason, the 14 items were formatted using a five-point Likert scale. According to Millers law, the limit on the amount of information that can be held in our working memory at any one time is 7 items, plus or minus 2 [
24]; therefore, the five-point scale was selected as it would offer enough choice and still be manageable for the participants with varying levels of cognition.
One of five responses can be given for each question: 1 = never, 2 = occasionally, 3 = half the time, 4 = frequently, 5 = always. Responses from the 14 questions are summed to develop a raw overall score between 14 and 70, which is then converted into an overall percentage disability by simple calculation:
$$ SEI- HI\ Disability\ \left(\%\right)=\left(\left(\frac{score}{14}\right)-1\right)\times 25 $$
The higher the percentage, the more restricted a person feels. One hundred and twenty participants with normal hearing thresholds (Table
1) were asked to complete the SEI-HI questionnaire according to written instructions, following hearing screening to ensure normal hearing levels (classified as PTAv < 25 dB.)
Phase 3: Psychometric validation
To verify psychometric validity and reliability of the SEI-HI questionnaire, face validity, internal consistency, concurrent validity, reliability and the role of experimenter bias of the SEI-HI questionnaire were explored in 95 participants with mixed hearing thresholds (Table
1).
Face validity
Face validity was assessed by a subsample of 10 participants, chosen as they all had experience of working with or supporting people with HI. Five participants had HI ranging from mild to severe, and five had normal hearing. Participants were asked (1) whether they felt the instructions on the questionnaire were clear and easy to understand; (2) if all of the questions were clear and easy to understand; and (3) whether the questions were relevant to the HI population.
Internal consistency
Internal consistency and reliability of the scale was measured using Cronbach’s alpha coefficient. A coefficient of .7 or .8 is generally regarded as having high internal consistency.
Concurrent validity
Concurrent validity was examined using Spearman’s Rho correlation coefficient to observe similarities between outcomes on the SEI-HI questionnaire and two other questionnaires, the HHIE and Self-Assessment of Communication (SAC) [
25].
Test-retest reliability and experimenter bias
All participants were asked whether they would be available for repeat testing, and a convenience subsample comprising of the first 35 was selected for retesting over a 4–8-week period. In this subsample, there were 15 males and 20 females with mean age of 57.06 (SD = 13.01).
An intra-class correlation coefficient was used to examine the degree of correlation and agreement between the scores at the different time points (T1 and T2). Intra-class correlation coefficient estimates and their 95% confidence intervals were calculated based on a single-rating, absolute-agreement, 2-way mixed-effects model. Subsequently, a related-samples Wilcoxon Signed-Rank Test was undertaken to compare differences between the two time points. Fifteen (43%) of the retest participants were followed up by a different examiner to control for experimenter bias and 95% confidence intervals were inspected.
Phase 4: Feasibility
To ensure validity in a sample of people with cognitive impairment, participants with varying levels of cognitive impairment were asked to complete the SEI-HI questionnaire, with support from the experimenter. There were 18 patients with mild cognitive impairment and 13 with dementia (AD (n = 7); dementia with Lewy bodies (n = 2); frontotemporal dementia (n = 3); and corticobasal degeneration (n = 1)).
Discussion
The SEI-HI questionnaire has demonstrated a high level of reliability and validity as a measure of current psychosocial impact of hearing-related disability in older adults.
Internal consistency was demonstrated by strong Cronbach α scores. Due to the spectrum of disability associated with HI, despite the high α, all items were included to ensure the breadth of relevant questions to maximise clinical potential.
Due to individual differences in many factors including lifestyle, attitudes, comorbid health conditions, and available support networks, it is logical to assent that not any two people with the same levels of HI will be affected in the same way [
17,
26]. Because of this, there is no gold standard criterion for objectifying subjective responses to HI, and therefore we were unable to measure criterion validity. However, in cases where this is not suitable, measuring construct validity is adequate [
26,
27].
The test-retest reliability of the SEI-HI questionnaire is very satisfactory at ICC = 0.905. As on average a 6-week timescale had passed, it can be assumed with reasonable certainty that participants would not remember their previous scores and thus the coefficient has not been inflated as a result of the retesting procedure. Any minor changes could be reflective of changes in circumstance or attitudes towards the disability or irrelevant temporal factors, such as mood, which may cause a fluctuation in scores over time [
28].
Altogether, this lends support for the use of the SEI-HI questionnaire, not only as a cross-sectional instrument to measure current subjective hearing disability, but to be also used for longitudinal purposes. Due to the strong correlation, small standard error and no evidence of experimenter bias, it can be expected that changes over time are as a result of intervention rather than experimental error [
29]. Continuing research remains to evaluate the sensitivity to change of the SEI-HI questionnaire, for a more valuable longitudinal measure, which could then monitor the effects of audiological rehabilitation on the impact of HI to dementia.
Although the development of the SEI-HI questionnaire is based upon the scenarios reported in the HHIE, it aims to measure slightly different aspects than the HHIE and SAC, and it is promising to see the significant correlation between scores on the SEI-HI questionnaire and these measures, supporting the specificity of the SEI-HI questionnaire. This high specificity means that it does not allow for cross-condition comparisons. For example, the socio-emotional difficulties measured are due to common problems in relation to HI, and not suitable for measuring quality of life affected by other conditions or disabilities. Similarly, although the SEI-HI questionnaire has shown high validity and reliability in our sample, it may have limited uses in other cultures and thus may not be generalisable to non-English speaking countries. However, using the principles upon which the scale was designed and validated would allow for translation and validation into other languages. Participants recruited for Phase 4, with MCI to mild dementia, were able to complete the SEI-HI questionnaire with support from the experimenter. This has not been designed as an informant questionnaire, due to the sensitive and subjective nature of hearing disability. In cases with a more severe cognitive impairment, it will be left to the clinicians’ or researchers’ judgement to include carer-responses to aid answering these questions.
Given the clinical importance of investigating this association, it is essential to have a specific, valid and reliable questionnaire to compute current levels of hearing functioning. To the best of our knowledge, the SEI-HI questionnaire is the first validated instrument to measure current levels of subjective hearing disability in recent years. To conclude, this study has shown that the SEI-HI questionnaire is a favourable and relevant instrument to assess current levels of subjective hearing disability regardless of hearing threshold. It can be used with confidence to control for subjective levels of disability in people with varying levels of HI, to assess further the risk of cognitive decline. The use of the SEI-HI questionnaire would help to determine whether the social and emotional impacts of HI have more of an influence on the risk for dementia, in addition to hearing thresholds alone.
Compliance with ethical standards
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.