Introduction
Hearing loss is the third most prevalent chronic condition in older adults,Reference Low1 with a reported prevalence of up to 83 per cent.Reference Wu, Chin and Tong2 Studies have reported that hearing loss may cause poor communication,Reference Brink and Stones3–Reference Strawbridge, Wallhagen, Shema and Kaplan6 social withdrawal and isolation,Reference Fellinger, Holzinger, Dobner, Gerich, Lehner and Lenz7–Reference Weinstein and Ventry10 depression,Reference Chia, Wang, Rochtchina, Cumming, Newall and Mitchell8, Reference Cacciatore, Napoli, Abete, Marciano, Triassi and Rengo11–Reference Veras and Mattos13 dementia,Reference Veras and Mattos13, Reference Herbst and Humphrey14 decreased functional status,Reference Strawbridge, Wallhagen, Shema and Kaplan6 and other maladaptive behaviour.Reference Hallberg, Hallberg and Kramer4 Despite this, many elderly individuals are not assessed or treated for their hearing loss.Reference Jupiter and DiStasio15–17 The importance of this omission needs to be studied, especially in the light of our ageing population and the adverse physical and psychosocial impact that hearing loss may have on affected individuals.
Efforts to quantify the impact of hearing loss range from audiometric measurements, through clinical tests such as the whisper test and tuning fork tests, to self-reported quality of life (QoL) questionnaires. Quality of life is a broad, multidimensional concept best assessed by both objective and subjective measurements,Reference Tsuruoka, Masuda, Ukai, Sakakura, Harada and Majima5 and is a subject of great interest and significance. Few studies have investigated the effect of age-related hearing loss on QoL,17, Reference Dalton, Cruickshanks, Klein, Klein, Wiley and Nondahl18 especially in Asian populations. Quality of life measures may assist documentation of the burden of chronic conditions and enable evaluation of health changes and post-interventional effects, and hence assist the allocation of public health resources.
Our study used the Short Form 36 Health Survey, a well validated, generic, health-related QoL measure widely used across different disciplines. We also used the Hearing Handicap Inventory for the Elderly Screening Version questionnaire, a hearing-specific questionnaire.
Our study aimed: (1) to investigate the impact of untreated age-related hearing loss on older individuals' QoL, using both generic and disease-specific research instruments; and (2) to determine the usefulness of QoL questionnaires and audiometry for guiding future interventional strategies.
Materials and methods
Study population
This was a cross-sectional study conducted over two months. Eighty patients aged 50 years and over with untreated hearing impairment were recruited consecutively from the audiology clinic. Informed consent was obtained. Patients selected for the study were chosen by criteria that excluded: hearing aid use; hearing impairment secondary to otological disease, ototoxicity or noise; congenital or early onset hearing loss; and the presence of psychiatric disorders or dementia.
Audiometric assessment
Hearing examination involved pure tone air and bone conduction audiometric assessment, conducted in a soundproof room by a trained audiologist using a regularly calibrated audiometer. Masking was done where necessary. Hearing impairment was defined as the pure tone average (PTA) of air conduction hearing thresholds >25 dB HL for four frequencies (0.5 to 4.0 kHz) in the better ear. These criteria were chosen on the basis of their impact on the functional aspects of hearing.Reference Dalton, Cruickshanks, Klein, Klein, Wiley and Nondahl18 The American Speech Language Hearing Association classification of hearing loss was applied, i.e.: mild = 26–40 dB HL; moderate = 41–55 dB HL; moderately severe = 56–70 dB HL; and severe to profound = ≥71 dB HL.Reference Clark19 Twenty patients were recruited from each category of hearing loss.
Quality of life assessment
Two questionnaires were used: the Short Form 36 Health Survey and the Hearing Handicap Inventory for the Elderly Screening Version. Both were administered by one of the authors. Verbal translations of the questionnaires were used for non-English speaking subjects.
Short Form 36 Health Survey
This questionnaire served as a generic measure of health-related QoL. It is a well validated, multipurpose survey comprising 36 questions on QoL. It generates an eight-domain profile of functional health and well-being scores, covering: physical role functioning, role physical, bodily pain, general health perception, vitality, social functioning, emotional role functioning and mental health. Each of these eight domains is graded on a 100-point scale, with a higher score indicating better health. These eight domains are summarised into two summary scores – the physical component score and mental component score – by assigning weights to the individual subscales, as described by the survey's authors. The Short Form 36 Health Survey has been shown to be suitable for use in elderly populations.Reference Lyons, Perry and Littlepage20
Hearing Handicap Inventory for the Elderly Screening Version questionnaire
This served as a disease-specific research tool. It was developed specifically for use within an elderly population, to screen for self-assessed hearing handicap.Reference Weinstein and Ventry10 Many studies have shown the significant audiometric correlates and test–retest reliability of this tool.Reference Jupiter and DiStasio15, Reference Jupiter and Palagonia21, Reference Bess, Lichtenstein, Logan and Burger22 It comprises 10 standardised questions, five of which relate to emotions and five to social situations. A response of ‘yes’ gains 4 points, ‘sometimes’ 2 points and ‘no’ 0 points. Scores for this questionnaire range from 0 (no handicap) to 40 (maximum handicap). A cut-off score of more than 8 has commonly been used to indicate the presence of at least mild hearing handicap.Reference Yueh, Shapiro, MacLean and Shekelle16
Statistical analysis
Data analysis was conducted using Stata version 10.2 software (Stata, College Station, Texas, USA). The level of significance was set at 5 per cent. For the Short Form 36 Health Survey individual domains, physical component score and mental component score, and the Hearing Handicap Inventory for the Elderly Screening Version questionnaire total score, we used the ordinary least squares linear regression model to examine factors associated with each outcome. We performed both univariate and multivariate analyses, with the latter done to examine adjusted mean difference between the hearing loss groups. Variables included in the multivariate analysis were based on clinical judgment, and included degree of hearing loss, medical conditions and gender. As 40 dB HL is often employed as the threshold for screening and audiometry,Reference Yueh, Shapiro, MacLean and Shekelle16 we used 40 dB HL as the criterion standard to adjudicate clinically significant hearing loss. Receiver operating characteristic curve analysis was applied to evaluate the sensitivity and specificity of different cut-off scores for the Hearing Handicap Inventory for the Elderly Screening Version questionnaire, in order to find the most appropriate cut-off score to indicate hearing loss of at least 40 dB HL. We also grouped Hearing Handicap Inventory for the Elderly Screening Version questionnaire scores as dichotomous variables (>8 versus ≤8) to indicate hearing handicap, and used the binary logistic regression model to examine factors associated with the outcome. Once again, both univariate and multivariate analyses were performed, with the latter including variables deemed to be clinically relevant.
Results and analysis
Study population characteristics
Subjects comprised 33 (41 per cent) men and 47 (59 per cent) women. Their median age was 69 years (interquartile range: 59–77). Fifty-six (70 per cent) were married, while the remaining 24 (30 per cent) were single, divorced or widowed. The majority (58.8 per cent) had obtained only primary or lower secondary education. Sixty (75 per cent) reported the presence of other, concurrent medical conditions.
The hearing loss observed was predominantly bilateral, symmetrical, high frequency loss. The median hearing loss for the right ear was 61.25 dB HL (interquartile range: 41.25–74.38) and for left ear 63.75 dB HL (interquartile range: 41.88–73.75). The median hearing loss of the better ear used in our study was 55 dB HL (interquartile range: 33.75–70). There was no significant difference in male:female ratio, comparing the different hearing loss groups (p = 0.90). Other medical conditions were equally prevalent amongst the genders (p = 0.69) and amongst the different hearing loss groups (p = 0.91).
Self-reported quality of life and hearing handicap
The medians for the Short Form 36 Health Survey physical component score and mental component score were 42.60 (interquartile range: 30.23–49.70) and 49.42 (interquartile range: 33.68–53.43), respectively. Neither univariate nor multivariate analyses revealed any significant association between the severity of hearing loss on audiometry and the Short Form 36 Health Survey physical component score, mental component score or eight individual domain scores (Table I). However, female gender and the presence of other medical conditions were significantly associated with a lower physical component score (female gender: 7.31 units decrease (p = 0.005); presence of other medical conditions: 6.10 units decrease (p = 0.04)).
Table I Short Form 36 Health Survey: individual dimension scores and physical and mental component scores
Data represent multivariate adjusted medians (95 per cent confidence intervals). *26–40 dB HL; †41–55 dB HL; ‡56–70 dB HL; ** > 70 dB HL. §p ≤ 0.05. PCS = physical component score; MCS = mental component score; ref = reference; mod = moderate; mod sev = moderately severe; sev prof = severe to profound; med conds = medical conditions
A dose-graded correlation was observed between the severity of hearing loss on audiometry and the self-reported hearing impairment score from the Hearing Handicap Inventory for the Elderly Screening Version questionnaire. Using the category of mild hearing loss as a reference, we observed a 7.54 unit increase in Hearing Handicap Inventory for the Elderly Screening Version questionnaire score in moderate hearing loss subjects (p = 0.002), a 9.44 unit increase in moderately severe hearing loss subjects, and a 14 unit increase in severe to profound hearing loss subjects (p = <0.001) (Table II). In addition, similar significant positive correlations were noted between the severity of audiometrically determined hearing loss and the emotional and social subscores of the Hearing Handicap Inventory for the Elderly Screening Version questionnaire (Table II).
Table II Hearing Handicap Inventory for the Elderly Screening Version questionnaire scores: total score, emotional subscore, social subscore and odds ratio for total score >8
Data represent multivariate adjusted medians (95% confidence intervals (CIs)) unless otherwise specified. *26–40 dB HL; †41–55 dB HL; §56–70 dB HL; #>70 dB HL. ‡0.001< p <0.05; **p < 0.001. OR = odds ratio; ref = reference; mod = moderate; mod sev = moderately severe; sev prof = severe to profound; NC = not calculable; med conds = medical conditions
With regards to individual question responses to the Hearing Handicap Inventory for the Elderly Screening Version questionnaire, more than 50 per cent of subjects experienced ‘frustration on talking to family members’, ‘difficulty when someone speaks in a whisper’, ‘difficulty on listening to the television (TV) or radio’ and ‘hearing loss limits or hampers personal or social life’ (i.e. they reported ‘yes’ or ‘sometimes’ to these questions).
When a Hearing Handicap Inventory for the Elderly Screening Version questionnaire cut-off score of more than 8 was used to denote hearing handicap, 100 per cent of severe to profound hearing loss subjects were categorised as hearing-handicapped, compared with 65 per cent of those in the moderately severe category, 55 per cent of those in the moderate category, and 25 per cent of those in the mild category (p < 0.001). Individuals with moderate and moderately severe hearing loss were respectively 3.79 and 5.92 times as likely to report handicap, compared with those with mild hearing loss (Table II). This indicates that hearing handicap was already experienced by some individuals with mild hearing loss, and that the degree of self-reported handicap correlated to the degree of severity of audiologically determined hearing loss. Furthermore, we found significant correlations between the Hearing Handicap Inventory for the Elderly Screening Version questionnaire handicap severity level (no handicap: 0–8; mild to moderate handicap: 10–24; severe handicap: 26–40) and the audiometrically determined hearing loss severity (p < 0.001).
The sensitivity, specificity, percentage correctly classified and likelihood ratio at various Hearing Handicap Inventory for the Elderly Screening Version questionnaire cut-off scores are shown in Table III, using 40 dB HL as the criterion standard. A cut-off score of more than 8 yielded 72.8 per cent sensitivity and 71.4 per cent specificity in detecting clinically significant hearing loss of at least 40 dB (receiver operating characteristic = 0.83).
Table III Sensitivity, specificity and correct classification rate for various Hearing Handicap Inventory for the Elderly Screening Version questionnaire cut-off scores*
Receiver operating characteristic area = 0.8329; standard error = 0.057; 95% confidence interval = 0.72–0.95.
* Using 40 dB HL as the standard criterion for hearing loss. Sens = sensitivity; spec = specificity; class = classification; LR + = positive likelihood ratio; LR − =negative likelihood ratio
Discussion
Our study found that the Short Form 36 Health Survey, a generic QoL measure, lacked sensitivity and specificity in assessing the impact of hearing loss on QoL. The recent study by Helvik et al. Reference Helvik, Jacobsen and Hallberg23 reached similar conclusions; however, these authors utilised a different generic QoL instrument – the Psychological General Well-Being inventory. Other studies have also shown that generic QoL measures lack precisionReference Naess24, Reference Nordenfelt25 and sensitivity to specific effects of hearing loss.Reference Bess26 In contrast, however, several other studies have reported a significant decrease in the Short Form 36 Health Survey physical component score, mental component score and individual domain scores with increasing hearing loss severity.Reference Brink and Stones3, Reference Chia, Wang, Rochtchina, Cumming, Newall and Mitchell8, Reference Dalton, Cruickshanks, Klein, Klein, Wiley and Nondahl18, Reference Ringdahl and Grimby27, Reference Bess, Lichtenstein, Logan, Burger and Nelson28 In a study conducted by Newman et al.,Reference Newman, Jacobson, Hug and Sandridge29 a significant correlation was present even in subjects with marginal hearing loss.
Studies have shown that women have lower psychosocial well-being than men.Reference Helvik, Jacobsen and Hallberg23, Reference Moum, Sørensen, Tambs and Holmen30 No conclusive evidence had been given to account for this observation. In the current study, female gender and the presence of other medical conditions were found to be associated with a significantly lower Short Form 36 Health Survey physical component score, even after controlling for other possible confounders in the multivariate analysis. A possible explanation for this is that health problems such as orthopaedic conditions and cardiovascular disease, which often limit physical activity, commonly afflict the elderly, and may result in a decline in physical well-being. Women may have lower physical capacity and endurance, contributing to poorer physical component scores.
However, it must be noted that comparison between different studies is difficult due to differences in population sampling, methodology, hearing loss classification and the QoL instrument employed. Moreover, affected individuals have varying perceptions of hearing loss, and this may be influenced by general life circumstances and events. In addition, one should not ignore the synergistic effect of multiple comorbidity on QoL scores,Reference Brink and Stones3 and individuals' varying ability to cope with their hearing loss. Some studiesReference Brink and Stones3, Reference Yueh, Shapiro, MacLean and Shekelle16, Reference Campbell, Crews, Moriarty, Zack and Blackman31 have noted a decrease in Short Form 36 Health Survey physical component score with increasing severity of hearing loss; this may be attributable to the general health decline in the elderly (but may not have been adjusted for in these studies' analyses). Hence, there is an indirect relationship between the degree of hearing loss and the general well-being of affected elderly individuals.
On the other hand, our results for the Hearing Handicap Inventory for the Elderly Screening Version questionnaire showed a highly significant correlation between severity of hearing impairment and the degree of hearing handicap reported by subjects. This is consistent with other studies' findings.Reference Jupiter and DiStasio15, Reference Yueh, Shapiro, MacLean and Shekelle16, Reference Dalton, Cruickshanks, Klein, Klein, Wiley and Nondahl18 In age-related hearing loss, loss of high frequency hearing results in poor speech perception, as the perception of consonants (which give clarity to speech) is affected.Reference Chia, Wang, Rochtchina, Cumming, Newall and Mitchell8, Reference Veras and Mattos13, Reference Helvik, Jacobsen and Hallberg23 In our series, such poor speech perception may have accounted for subjects' communication difficulties, which may in turn have had a negative impact on their emotional and social well-being (Table II). Subjects reported problems when listening to TV and radio (61.2 per cent) and listening to whispered speech (75 per cent), and they experienced frustration when talking to family members (58.7 per cent). Similarly, another study reported 50 per cent impairment in these areas.Reference Weinstein32
• Hearing loss commonly affects elderly individuals' quality of life (QoL), yet is often neglected
• This study investigated the impact of untreated age-related hearing loss on elderly subjects' QoL, and also assessed the usefulness of QoL questionnaires as screening tools for the presence of significant hearing loss
• The Short Form 36 Health Survey and the Hearing Handicap Inventory for the Elderly Screening Version questionnaire were used to assess older subjects with untreated hearing impairment
• The former questionnaire, a generic measure, lacked specificity and sensitivity; however, the latter questionnaire detected significant hearing impairment, suggesting its possible use for hearing loss screening
It can therefore be inferred that untreated hearing loss does result in a significant decline in QoL. However, the perception of a hearing handicap (i.e. a response of ‘yes’ or ‘sometimes’) was reported by less than half (41.2 per cent) of subjects. This is consistent with Weinstein findings (46 per cent).Reference Weinstein32 Explanations for this include subjects' varying perceptions and definitions of hearing handicap, denial of hearing loss in some, and the degree of social and family support received. It is also important to note that hearing loss adversely affects not only the hearing-impaired individual but also the QoL of those around them. The hearing-impaired individual's frustrations, and some of their adaptive behaviours (e.g. increasing the volume of TV and radio) may disturb those around them. Thus, it may be informative to investigate the impact of hearing loss on the QoL of affected individuals' family members, in order to obtain a more objective picture of the overall impact of hearing loss.
Using 40 dB HL as the standard criterion for clinically significant hearing loss, it was evident that, as the Hearing Handicap Inventory for the Elderly Screening Version questionnaire cut-off score increased, this tool's specificity increased while its sensitivity fell (Table III). A cut-off score of more than 8 provided a good compromise between sensitivity and specificity (giving a sensitivity of 72.88 per cent and a specificity of 71.43 per cent), with 72.50 per cent of subjects correctly classified. Other studies have assessed this cut-off score, and found sensitivity and specificity rates ranging from 0.63 to 0.80 per cent.Reference Ciurlia-Guy, Cashman and Lewsen33–Reference Hands39 Correlation coefficient r = 0.33 (p < 0.003).
We found an imperfect relationship between the degree of hearing handicap (as indicated by the Hearing Handicap Inventory for the Elderly Screening Version questionnaire) and the physiological hearing loss (as determined audiometrically). Different individuals may experience different degrees of handicap from the same level of hearing loss. Hence, the psychosocial impact of hearing loss cannot be predicted from audiometric findings alone.Reference Hallberg, Hallberg and Kramer4 The Hearing Handicap Inventory for the Elderly Screening Version questionnaire thus serves as a good screening tool for functional hearing impairment among older adults. This questionnaire may be less sensitive in detecting early hearing loss; however, it identifies individuals who experience handicap from their hearing loss and are thus more likely to agree to, and benefit from, interventional measures (e.g. formal audiological evaluation, referral for hearing aids, and support groups). The Hearing Handicap Inventory for the Elderly Screening Version questionnaire is also easy to administer, and provides a convenient, cheap and practical means of screening for hearing loss, especially among bed-bound and physically debilitated elderly individuals, who may find it difficult to travel to an audiological centre for formal evaluation. This questionnaire can also be used for follow up, to assess QoL improvement post-intervention.
Many studies have emphasised the importance of self-help groups, medical intervention and early rehabilitation in ensuring maximum benefit for hearing-impaired elderly individuals.Reference Weinstein32, Reference Garstecki and Erler40, Reference Kriscos41
Conclusion
The disease burden of untreated age-related hearing loss may have been neglected due to its invisibility, gradual onset, perceived insignificance compared with other life-threatening conditions, and lack of data on QoL impact. In this study, we have shown that untreated hearing loss in the elderly resulted in a significant decline in QoL, which was reflected in Hearing Handicap Inventory for the Elderly Screening Version questionnaire scores. The Short Form 36 Health Survey, being a generic measure, was less useful in assessing QoL impairment solely attributed to hearing loss.
In view of the increasing prevalence of hearing loss and its significant detrimental impact, we advocate hearing loss screening, creation of support groups, and timely medical intervention (e.g. referral for hearing aids).
The Hearing Handicap Inventory for the Elderly Screening Version questionnaire is a good, disease-specific screening tool for hearing loss. In view of the imperfect relationship between physiological and functional hearing loss, the combined use of both functional (i.e. the Hearing Handicap Inventory for the Elderly Screening Version questionnaire) and physiological (i.e. audiometry) assessment may provide the most holistic evaluation for interventional purposes.
Acknowledgements
The authors would like to thank Mr Arul Ernest (Principal Medical Statistician, Clinical Research Unit, Tan Tock Seng Hospital) and the audiologists of the Department of Audiology, Tan Tock Seng Hospital, for their assistance with this study.
