Evaluating a Dichotomized Measure of Self-Reported Hearing Loss Against Gold Standard Audiometry

Abstract

Objectives: To evaluate a harmonized binary measure of self-reported hearing loss against gold standard audiometry in an older adult population. Method: Seven nationally representative population-based studies were harmonized and pooled (n = 23,001). Self-report items were recoded into a dichotomous format. Audiometric hearing loss was defined by averaged pure-tone thresholds greater than 25-decibel hearing level in the better ear. We compared age and sex stratified prevalence rates of hearing loss estimated by self-report and audiometric measures. Results: Overall, 56% of men and 43% of women had audiometric hearing loss. There were moderate associations between self-reported and audiometric hearing loss. However, prevalence based on self-report was overestimated for adults aged below 70 years and underestimated for adults aged above 75. Discussion: Self-report of hearing loss is insensitive to age effects and does not provide a reliable basis for estimating prevalence of age-related hearing loss, although may indicate perceived hearing disability.

Keywords

Hearing loss is among the most prevalent chronic age-related health conditions. It occurs in up to 90% of American adults aged above 80 (Cruickshanks et al., 2003), and is estimated to currently affect 75% of Australians aged above 70 years (Access Economics, 2006). Overall, Australian prevalence rates are projected to increase with population aging. In 2005, the financial cost and burden of disease arising from hearing loss was estimated to be more than AU$22 billion in Australia (Access Economics, 2006) and in excess of €213 billion Europe (Shield, 2006). The lifetime cost of severe to profound hearing loss in the United States has been estimated to be US$297,000 per person (Mohr et al., 2000). These estimates are also expected to increase with rising prevalence. Age-related hearing loss has been linked with reduced quality of life (Chia et al., 2007; Hogan, O’Loughlin, Miller, & Kendig, 2009), poor mental health (Gopinath, 2009b; Kramer, Kapteyn, Kuik, & Deeg, 2002), diabetes (Mitchell et al., 2009), smoking (Gopinath et al., 2010), cognitive impairment (Tay et al., 2006), reduced social participation, increased use of community support services (Schneider et al., 2010), and increased risk of mortality (Karpa et al., 2010). Despite its impact on health and well-being, age-related hearing loss is known to be both underrecognized and undertreated (Reuben, Walsh, Moore, Damesyn, & Greendale, 1998). Recent reviews have found that globally only a small number of surveys were suitable for estimating hearing impairment in the general population (Pascolini & Smith, 2009). There are currently sparse national data on hearing impairment in older people, in Australia and elsewhere, with recent investigations of risk factors for incidence of age-related hearing loss being underpowered (Gopinath et al., 2010; Gopinath, Schneider, Rochtchina, Leeder, & Mitchell, 2009a; Mitchell et al., 2009).

The gold standard method for measuring hearing loss is pure-tone audiometry. The World Health Organisation (WHO) defines mild hearing impairment as unaided pure-tone audiometric hearing thresholds (PTA) greater than 25-decibel hearing level (dB HL) in the better ear, averaged across the tone frequencies of 0.5, 1, 2, 4 kHz. Hearing thresholds greater than 40dB HL are described by WHO as disabling (World Health Organization, 1999). Definitions of moderate and severe hearing loss are slightly different in the United States. Because the costs and logistics involved in conducting audiometry assessments are prohibitive for many epidemiological surveys, self-report measures are often used instead. These self-reported measures of hearing loss have previously been thought to be reliable and predictive of measured hearing loss while also providing an ecologically valid measure of perceived hearing difficulties (Johnson, 2010; Nondahl et al., 1998).

Contrary to the belief that self-reported hearing items are strongly associated with measured hearing loss, evidence is equivocal. Self-reported difficulty of hearing a conversation in a quiet room has been shown to have serious misclassification problems of audiometric hearing loss and has been argued to be inappropriate for providing prevalence estimates of hearing impairment over a broad age sample (range 15 to 71+; Wilson et al., 1999). Response propensity may also be influenced by individual differences, as one study has shown that self-reported hearing loss is more strongly associated with personality than measured hearing loss (Cox, Alexander, & Gray, 2007). In contrast, other studies (Nondahl et al., 1998; Sindhusake et al., 2001) have reported that a more general question asking “Do you feel you have hearing loss?” with binary response is a more reliable self-report measure for assessing age-related hearing loss in adults aged between 45 and 100, with better predictive properties compared to other self-report scales such as the Hearing Handicap for the Elderly (HHIE; Ventry & Weinstein, 1982), hearing aid use, and hearing ratings on a 5 point Likert-type scale, although these comparisons failed to consider how age moderates the association between self-report and audiometric hearing measures. Although responses are generally coded on a 3- to 5-point Likert-type scale, it is not uncommon to recode responses into a dichotomous variable format, whereby any degree of reported hearing difficulty (e.g., “a little trouble,” “a lot of trouble,” “deaf”) is taken to be indicative of hearing loss (Caban, Lee, Gomez-Marin, Lam, & Zheng, 2005).

Given the lack of large-scale data sets including information on hearing loss, much benefit can be gained from combining information from the available data sets that include items on hearing loss. This data-pooling approach requires harmonizing and validating measures of self-reported hearing loss so they are comparable across surveys. In this study, we report on a method of dichotomizing self-reported hearing loss items in a pooled data set and evaluate the reliability and utility of such harmonized self-reported hearing loss in comparison to the gold standard measure of pure-tone audiometric thresholds. The aims of the present study were, first, to compare prevalence estimates of hearing loss between those based on dichotomized self-report and standardized audiometric measurements. Second, this study aimed to evaluate the utility of a harmonized binary measure of self-reported hearing loss. The findings will enable future research into prevalence and incidence of hearing impairment, its risk factors and outcomes, in a nationally representative pooled data set.

Method

Sample

Data were drawn from the Dynamic Analyses to Optimise Ageing Project (DYNOPTA) which has harmonized and pooled nine Australian longitudinal studies of aging (baseline n = 50,652). A comprehensive description of the study design has been reported previously (Anstey et al., 2010). Baseline details of the contributing studies included in this article are shown in Table 1 and comprise the Australian Longitudinal Study of Ageing (ALSA; Luszcz et al., 2007), the Australian Longitudinal Study of Women’s Health (ALSWH; Lee et al., 2005), the Blue Mountains Eye Study (BMES; Mitchell et al., 2009), the Canberra Longitudinal Study (CLS; Christensen et al., 2004), the Melbourne Longitudinal Studies on Healthy Ageing Program (MELSHA; Browning & Kendig, 2010), the Path Through Life Study (PATH; Anstey et al., 2011) and the Sydney Older Persons Study (SOPS; Piguet et al., 2003). Baseline waves were collected between 1992 and 2001. Table 2 shows the pooled sample profile reported in this study (n = 23,001; 22% men). At baseline, the mean age was 71.6 years (SD = 6.2, range = 55 to 103), 9,190 (40%) participants left secondary school before the age of 15 years and 1,759 (8%) participants were tertiary qualified. The preponderance of women aged between 70 and 75 years reflects the inclusion of the ALSWH old cohort (n = 12,432).

Table 1.

Contributing Study Baseline Information

Study	Location	Baseline year	n	Age range
Australian Longitudinal Study of Ageing (ALSA)	Adelaide, SA	1992	2087	65-103
Australian Longitudinal Study of Women’s health (ALSWH)	Australia (national)	1996	13706	45-51
Blue Mountains Eye Study (BMES)	Blue Mountains, NSW	1997-00	2334	50-98
Canberra Longitudinal Study (CLS)	Canberra, ACT Queanbeyan, NSW	1990-91	1134	70-103
Melbourne Longitudinal Study Healthy Ageing (MELSHA)	Melbourne, Vic	1994	1000	65-94
Personality and Total Health through life (PATH)	Canberra, ACT Queanbeyan, NSW	2005-06	2222	64-70
Sydney Older Person’s Study (SOPS)	Sydney, NSW	1991-93	630	75-97

Table 2.

DYNOPTA Sample Profile at Baseline, Demographics (Row %) by 5-Year Age Groups (N = 23,001)

		55-59	60-64	65-69	70-74	75-79	80-84	85+
	n	%	%	%	%	%	%	%
Sex
Men	5061	4.5	28.8	12.6	17.1	17.4	11.4	8.1
Women	17940	1.7	8.0	4.6	71.1	8.2	3.5	2.9
Age left school
Left school age 14 or younger	9190	0.7	4.8	5.3	64.0	12.4	7.2	5.6
Left school age 15 or older	13215	3.5	18.5	7.4	55.5	8.7	3.9	2.5
Highest qualification
Secondary school	14708	1.3	5.4	4.3	70.8	10.2	4.7	3.3
Nontertiary	5204	4.8	24.8	11.9	37.2	10.6	6.6	4.1
Tertiary	1759	2.9	42.3	9.2	31.3	7.2	4.0	3.1
Career occupation
Managers and professionals	3933	1.7	26.1	7.5	43.9	10.2	6.1	4.5
Clerical and associate professional	5349	2.8	12.4	6.3	64.0	8.4	3.7	2.4
Tradespersons	1466	2.9	11.0	8.8	44.5	15.7	10.1	7.1
Sales, service, production, transport and laborers	5640	2.6	15.2	6.8	56.0	10.9	4.7	3.7

Note. Pooled baseline sample from ALSA, ALSWH old cohort, BMES, CLS, MELSHA, PATH.

Measures

Measured hearing loss was assessed by uncorrected pure-tone air-conduction audiometric thresholds (PTA) in waves 1, 3, 6, and 7 of ALSA and waves 2 and 3 of BMES. We defined hearing impairment as PTA greater than 25 dB HL in the better ear averaged across air-conduction tone frequencies of 0.5, 1, 2, and 4 kHz (PTA_{0.5,1,2,4 kHz}). Levels of hearing impairment were defined as no hearing loss (PTA_{0.5,1,2,4 kHz} < 25 dB), mild hearing loss (PTA_{0.5,1,2,4 kHz}: 25-40 dB), moderate hearing loss (PTA_{0.5,1,2,4 kHz}: 41-60 dB), and severe hearing loss (PTA_{0.5,1,2,4 kHz} > 60 dB).

Self-reported hearing loss was obtained by ALSA (Waves 1, 3, 6, 7), ALSWH (Wave 1), BMES (Waves 1, 2, 3), CLS (Waves 1, 2, 3, 4), MELSHA (Waves 1, 10), PATH (Wave 2) and SOPS (Waves 1, 2, 4, 5). Original item wording and response formats varied between studies and are presented in Table 3. A harmonized measure of self-reported hearing loss comprising of two levels was created (“no self-reported hearing loss,” and “any self-reported hearing loss”), whereby any degree of reported hearing losses, hearing difficulty, or hearing problems were recoded to reflect self-reported hearing loss. Age left school was coded as a binary variable indicating school leaving age being 15 years or older.

Table 3.

Self-Reported Hearing Items Harmonized in DYNOPTA

		Item	Response
Study	Wave	Harmonized self-reported hearing	Normal hearing	Hearing impairment
ALSA (Luszcz et al., 2007)	1	Do you have any difficulty with your hearing?^a	No	Yes
	1	If yes, how much difficulty do you have with your hearing?^a		Slight	Moderate	Severe
	3, 6, 7	How much difficulty, if any, do you have with your hearing (even if you are wearing your hearing aid)?^a	None	Slight difficulty	Moderate difficulty	Great difficulty
ALSWHold (Lee et al., 2005)	1	In the past 12 months have you had any hearing problems	Never	Rarely	Sometimes	Often
BMES (Mitchell et al., 2009)	1	Have you ever had a problem with your hearing	No	Yes
	1	Assessment of hearing problem		Mild	Moderate	Severe
	2, 3	Do you feel you have hearing loss^a	No	Yes
	3	Do you have difficulties with your hearing?^a	No	Sometimes	Yes
CLS (Christensen et al., 2004)	1, 2, 3, 4	Would you say your hearing (with a hearing aid) is generally good, fair or poor?	good	fair	poor
MELSHA (Browning & Kendig, 2010)		Is your hearing (with your hearing aid)	Excellent or Good	Fair	Poor
PATH (Anstey et al., 2011)	2	How you would rate your hearing on the following scale?	Adequate for all purposes	Slight inconvenience	Definite inconvenience	Definite handicap
SOPS (Piguet et al., 2003)	1, 2	Do you have any loss of hearing	No	Yes

Validated against gold standard of PTA (pure tone thresholds) averaged over 0.5, 1, 2, 4 kHz in the better ear.

Analyses

Self-reported hearing measures were obtained concurrently with audiometric assessments in ALSA Waves 1, 3, 6, and 7 and BMES waves 2 and 3. Evaluation of self-reported hearing measures was made by investigating their association with audiometry within these samples and waves. First, the sample was classified according to the categories of the self-report measure, then the means and standard deviations for PTA_{0.5,1,2,4 kHz} were calculated to enable calibration between subjective and objective measures. Second, polychoric correlations between the defined levels of impaired hearing thresholds (none, mild, moderate, severe) in the better ear and self-reported hearing were estimated using Mplus v5 (Muthen & Muthen, 2007). Finally, the discriminative properties of dichotomized measures were evaluated using receiver operator characteristic (ROC) curves, sensitivity, and specificity.

Baseline prevalence estimates for hearing loss were stratified by 5-year age groups and sex. Prevalence were estimated for pooled audiometric data from ALSA Wave 1 and BMES Wave 2, pooled dichotomized self-report data from ALSA Wave 1 and BMES Wave 2, and pooled dichotomized self-report data from all contributing studies. To examine how well self-reported hearing data could replicate findings based on analysis of PTA_{0.5,1,2,4 kHz} in multivariate analyses, odds ratios (OR) were estimated from a logistic regression model testing the effects of age, sex, and age left school on audiometric measures of hearing loss and on dichotomized self-reported hearing loss. Results were compared between all studies, using Wave 1 from the ALSA, ALSWH, CLS, MELSHA, and SOPS samples, and Wave 2 from the BMES and PATH samples.

Results

Comparison of Self-Report and Audiometric Hearing Loss

The broader age range and inclusion of participants younger than 65 years meant that the overall average hearing thresholds were lower in the BMES sample (Mean [M] = 24.9 dB, Standard Deviation [SD] = 14.7) compared to the ALSA sample (M = 35.4 dB, SD = 16.1). Table 4 compares how well four dichotomized self-report items predicted hearing loss as measured by PTA_0.5,1,2,4kHz within ALSA and BMES. Three of the self-report items were moderately associated with audiometric hearing loss. Their polychoric correlations ranged between 0.56 and 0.69, and area under the curve (AUC) ranged between 0.75 and 0.82. Furthermore, the average PTA_0.5,1,2,4kHz was higher for respondents who reported some level of hearing loss or hearing difficulty compared to respondents who reported no hearing loss or no hearing difficulties (ALSA Wave 1: mean difference = 16.5 dB, t₍₁₅₉₃₎ = 21.8, p < .01; ALSA Wave 3: mean difference = 13.0 dB, t₍₁₃₈₇₎ = 17.3, p < .01; ALSA Wave 6: mean difference = 12.8 dB, t₍₄₉₀₎ = 11.5, p < .01; ALSA Wave 7: mean difference = 11.5dB, t₍₃₄₉₎ = 8.4, p < .01; BMES Wave 2: mean difference = 14.2 dB, t₍₁₉₁₉₎ = 23.2, p < .01; BMES Wave 3: mean difference = 12.5 dB, t₍₁₅₆₂₎ = 18.5, p < .01). However, one item from Wave 3 in BMES, “do you feel you have difficulties with your hearing?” with responses originally coded on a 3-point scale: yes, sometimes, no, showed poor discrimination between impaired and nonimpaired hearing as defined by PTA_{0.5,1,2,4 kHz} > 25 dB HL. When recoded into dichotomous variable format where a response of either yes or sometimes indicated self-reported hearing loss, there was little difference in hearing thresholds between categories (mean difference = 0.3dB, t₍₁₃₀₁₎ = 0.4, p = .72; polychoric correlation = .03; AUC = .51, standard error [SE] = .02). Subsequent regression analysis revealed that original responses for this question were not associated with PTA_{0.5,1,2,4 kHz} (F _{(2, 1300)} = 0.1, p = .90). This self-report item was not used in coding the harmonized self-reported hearing variable.

Table 4.

Hearing Thresholds (dB HL) for Levels of Self-Reported Hearing and Their Association

	n	Age range	Mean PTA^b	SD	ρ (se)	AUC (se)	Sens.	Spec.	PPV	NPV
ALSA wave 1 (1992)^a
hearing loss	741	65-101	43.0	17.6	.69	.82 (.01)	.62	.85	.90	.51
no hearing loss	854	65-103	26.5	12.4
ALSA wave 3 (1994)^c
hearing loss	722	66-105	41.6	15.1	.56	.76 (.01)	.63	.78	.88	.45
no hearing loss	667	66-100	28.6	12.7
ALSA wave 6 (2000)^c
hearing loss	262	73-101	42.7	13.8	.56	.77 (.02)	.62	.77	.90	.37
no hearing loss	230	72-99	29.8	10.6
ALSA wave 7 (2003)^c
hearing loss	204	75-102	41.8	12.8	.59	.75 (.02)	.67	.77	.92	.38
no hearing loss	147	76-96	30.3	12.0
BMES wave 2 (1997-2000)^d
hearing loss	985	54-98	31.6	16.4	.66	.78 (.01)	.78	.67	.61	.82
no hearing loss	936	50-96	17.4	9.2
BMES wave 3 (2001-2004)^d
hearing loss	1000	59-99	30.1	14.7	.61	.77 (.01)	.85	.53	.59	.81
no hearing loss	564	55-96	17.5	9.2
BMES wave 3 (2001-2004)^e
hearing loss	787	55-99	26.1	14.3	.03	.51 (.02)	.63	.41	.47	.57
no hearing loss	516	55-96	25.8	14.9

Do you have any difficulty with your hearing?

PTA: average pure tone thresholds (decibels hearing levels dB HL) from tone frequencies of 0.5, 1, 2, 4 kHz in the better ear.

How much difficulty, if any, do you have with your hearing (even if you are wearing your hearing aid)?

Do you feel you have hearing loss?

Do you have difficulties with your hearing?

ρ : Polychoric Correlations between self-reported hearing and defined levels of hearing impairment (none, mild, moderate, severe) based on PTA_{0.5,1,2,4 kHz} in the better ear; AUC: Area Under the Curve, Receiver Operator Characteristic of self-reported hearing loss with PTA_{0.5,1,2,4 kHz} (dB) in the better ear; Sens.: Sensitivity; Spec: Specificity; PPV: Positive predictive value; NPV: Negative predictive value.

There were between-study differences in the ratio of positive predictive value (PPV) relative to negative predictive value (NPV). In all ALSA waves, self-reported hearing loss had high PPV (Wave 1 = .90, Wave 3 = .88, Wave 6 = .90, Wave 7 = .92) and comparably lower NPV (Wave 1 = .51, Wave 3 = .45, Wave 6 = .37, Wave 7 = .38). In contrast, both BMES waves had lower PPV (Wave 2 = .61, Wave 3 = .59) and higher NPV (Wave 2 = .82, Wave 3 = .81). The ALSA item had higher specificity than sensitivity at all waves whereas the item used in BMES showed lower specificity than sensitivity at all waves. The difference between studies can largely be explained by their respective age distributions. To demonstrate this we compared PPV and NPV for an older age cohort (aged 70 years and older, n = 947) relative to a younger aged cohort (aged 69 years and younger, n = 974) in the Wave 2 sample of BMES. In the older age cohort, the PPV of 0.76 was higher than the NPV of 0.67. This was reversed in the younger age cohort where the PPV of 0.41 was smaller than the NPV of 0.93. This could be expected as audiometric hearing loss is less prevalent at younger ages, consequently a negative test result is more likely in the younger cohort whereas the probability of a positive test result increases with age.

Comparison of Prevalence Rates Based on Self-Report and Audiometric Hearing Loss

Figure 1 shows the prevalence hearing loss based on audiometric assessment and self-report. Within the pooled ALSA and BMES sample, overall prevalence rates derived from self-reported hearing loss was 56% (95% CI = [53, 58]) for men and 43% (95% CI = [41, 45]) for women, which only slightly underestimated true overall prevalence estimated from PTA_{0.5,1,2,4 kHz}, of 59% (95% CI = [57, 61]) for men and 46% (95% CI = [44, 48]) for women. However, the age gradient for self-report items was not as steep as the age gradient for measured hearing impairment. The prevalence of self-reported hearing loss increased by only 4.1% for every 5-year increase in age. In contrast, prevalence of measured hearing impairment increased by 13.5% for every 5-year increase in age. Although prevalence rates based on self-reported items were reasonably accurate for adults aged between 65 and 74 years, prevalence based on self-report data greatly overestimated measured hearing loss prevalence for younger age groups but greatly underestimated prevalence rates for older age groups. For example, within the pooled ALSA and BMES sample 44% (95% CI = [35, 53]) of men aged between 55 and 59 reported some degree of hearing loss, whereas only 9% (95% CI = [4, 15]) of men in this age group had average PTA_{0.5,1,2,4 kHz} greater than 25 dB HL in the better ear. In contrast, 69% (95% CI = [64, 75]) of men aged over 85 years reported some level of hearing difficulty, whereas 89% (95% CI = [85, 94]) of men in this age group had average PTA_{0.5,1,2,4 kHz} greater than 25 dB HL in the better ear. A similar pattern was observed for women. Prevalence of hearing impairment based on self-report by women aged above 85 years was 63% (95% CI = [56, 69]) which was significantly less than prevalence of 87% (95% CI = [82, 9]) of women aged over 85 who had average PTA_{0.5,1,2,4 kHz} greater than 25 dB HL in the better ear. A comparable age bias was observed for prevalence rates based on harmonized self-report items from baseline samples of all studies. This age bias is such that the self-reported items are insensitive to age effects and overestimate prevalence for younger ages whilst underestimating prevalence for older ages.

Figure 1.

Comparison of prevalence (with 95% CI) of objective (PTA_{0.5,1,2,4 kHz}) and subjective hearing loss by age group and sex at baseline

Comparison of Differentials of Hearing Loss Based on Self-Report and Audiometric Measures

To test how well use of a binary outcome measure of self-reported hearing loss replicated effects from a multivariate model where audiometric hearing impairment was the outcome, we compared age, sex, and education effects estimated from a logistic regression model of data from each individual study sample. When audiometric hearing loss was modeled as the outcome, there was little variability between studies with both ALSA and BMES showing that for every year lived respondents were 1.13 (95% CI = [1.11, 1.15]) times more likely to have hearing impairment. Men were 1.47 (95% CI = [1.17, 1.84]) times more likely to have hearing impairment in the ALSA sample, and 1.56 (95% CI = [1.26, 1.91]) times more likely to have hearing impairment in the BMES sample. Although participants who left school before the age of 15 were more likely to have audiometric hearing impairment in both studies, there were differences between these two samples in the magnitude of the OR with early school leavers in the ALSA sample 1.43 (95% CI = [1.14, 1.79]) times more likely to have hearing impairment. Although early school leavers in the BMES sample were 1.17 (95% CI = [0.93, 1.48]) times more likely to have hearing impairment, the OR was not reliably different from 1 (p = .18).

When a binary measure of self-reported hearing was modeled as outcome the OR for age was reduced relative to audiometric hearing loss and ranged between 0.98 (95% CI = [0.93, 1.04]) in the PATH study and 1.08 (95% CI = [1.03, 1.12]) in the SOPS study. The OR of 0.98 for the PATH study was in the opposite direction compared to other studies but was not reliably different from 1 (p = .51) and could be a result of the narrow-aged cohort for this sample (aged between 64 and 70).

The greater relative risk of self-reported hearing loss associated with men was also reasonably consistent between studies. ORs ranged from 1.51 (95% CI = [1.14, 1.99]) for the MELSHA study, to 1.97 (95% CI = [1.66, 2.34]) for the PATH study, and showed that men are more likely to experience hearing loss. In general the association between sex and hearing loss was stronger for self-report items in comparison with audiometric hearing impairment, and there was consistent overlap of confidence intervals for all studies.

The relative risk of self-reported hearing impairment associated with school-leaving age, was also incongruent between studies and with audiometric hearing loss. Within the baseline ALSA sample, participants who left school before the age of 15 were 1.43 times (95% CI = [1.14, 1.79]) more likely to have audiometric hearing loss, whereas there was no significant difference in the likelihood of reporting hearing difficulties for individuals who left school before the age of 15 compared to those who left school at the age of 15 or older (OR = 0.96, 95% CI = [0.80, 1.15], p = .69). A similar, although less divergent pattern was observed in BMES. Although not statistically significant, early school leavers in BMES were 1.17 times more likely to have audiometric hearing loss (95% CI = [0.92, 1.48]), whereas there was no difference in self-reported hearing loss between early or late school leavers (OR = 0.99, 95% CI = [0.79, 1.24]). Of the studies that only collected self-report hearing data, all ORs trended to be greater than 1 indicating that participants who left school before age 15 were more likely to report hearing difficulties, however these effects were statistically significant only for ALSWH (OR = 1.14, 95% CI = [1.05, 1.23]) and MELSHA (OR = 1.41, 95% CI = [1.07, 1.87]).

Discussion

This study aimed to report on the harmonization of dichotomized self-reported hearing loss items and evaluate them against audiometric hearing loss. We compared prevalence estimates derived from self-reported hearing measures with estimates derived from standardized audiometric data in the Australian population. Analyses showed that within the ALSA and BMES studies, a dichotomized measure of self-reported hearing loss appeared to be reasonably sensitive and displayed moderate associations with audiometric assessment. Self-report data, however, did not provide a reliable basis for estimating prevalence in the general population, and depending on the sample characteristics or question wording, had conflicting sensitivity and specificity. In particular, self-reported hearing appeared to overestimate hearing impairment (PTA_{0.5,1,2,4 kHz} > 25 dB HL) in younger age cohorts whereas underestimating hearing impairment in older age cohorts. Although previous comparisons of self-report with audiometric measures have supported conclusions that self-report data may be sufficient for estimating overall prevalence of hearing loss, these studies only compared prevalence in broad age cohorts and failed to consider an age bias in self-reported health measures (Gates, Cooper, Kannel, & Miller, 1990; Nondahl et al., 1998). Indeed, in this study, the difference between self-report and audiometric-based prevalence for all adults aged 55 years and older was minimal.

Social comparison theory (Willis, 1981) provides one explanation for the failure of self-reported hearing to detect age differences. Social comparison theory maintains that older adults tend to overrate their perceived health because they make implicit downward comparisons with negative old-age stereotypes (Heckhausen & Brim, 1997; Sargent-Cox, Anstey, & Luszcz, 2008). The age bias inherent in self-reported hearing items could therefore reflect the downward social comparisons older adults are surmised to make when rating their health despite loss of functioning (Heckhausen & Brim, 1997). A similar explanation was given for the poorer performance of the HHIE in estimating prevalence in adults aged 65 years and older compared to adults aged between 48 and 64. Nondahl et al. (1998) suggested that older adults are more likely to be accepting of hearing impairment as they do not consider it an unusual aspect of ageing. The high prevalence and common experience leads to hearing loss becoming normalized in older adults. Furthermore, as hearing decline is generally a gradual process, many adults have time to adjust to hearing loss.

Younger age groups could overestimate their hearing difficulties for a number of reasons. First, they are more likely to be actively participating in the workforce and have greater work-related demands on their hearing. Certainly after the retirement age of 65, self-report data no longer overestimates audiometric hearing impairment. Also, hearing ability for pure-tone thresholds below 4kHz begin to decline in the 50s, and initial losses may be more noticeable at these ages (Wiley, Chappell, Carmichael, Nondahl, & Cruickshanks, 2008). Low levels of hearing aid utilization in younger age groups could also contribute to the differences between self-report and audiometric measures. It has been reported that adults may experience hearing difficulties for up to 10 years before they recognize their hearing to be a problem and access hearing services during their mid 70s (Davis, Smith, Ferguson, Stephens, & Gianopoulos, 2007). Finally, this could reflect a cohort effect whereby younger cohorts are more likely to report health problems and functional difficulties. It should not be discounted that the apparent overreporting to hearing loss is a real effect reflecting the poorer health status of younger cohorts. Seeman, Merkin, Crimmins and Karlamangla (2010) found significant increases in disability over a 16-year period in a cohort aged 60-69 years whereas those aged 70-79 years showed no significant changes in disability and those aged 80 years and above showed lower prevalence of functional limitations.

Hearing impairment was more likely to occur in men, older adults, and early school leavers regardless of whether hearing impairment was defined by pure-tone thresholds or self-report. However, when estimates were based on self-report items rather than audiometric measures, there were considerable differences in the strength of these associations and there was more variability between studies. In particular, self-reported hearing items were not sensitive to the relation between age left school and hearing impairment. Self-reported hearing items consistently underestimated the strength of the age-related risk of hearing impairment and tended to overestimate the association between sex and hearing impairment. It is possible that some of the discrepancy in findings related to the self-reported hearing impairment items was due to design difference in the contributing studies, including sample composition and age structure.

Of particular interest is the influence of the response scale on the reliability of self-reported hearing loss. Two almost identical questions: “do you have any difficulty with your hearing?” (Binary response: yes, no) and “do you have difficulties with your hearing” (three level: yes, sometimes, no) were shown to have conflicting associations with measured hearing loss. The latter item is of a similar form to the HHIE (Ventry & Weinstein, 1982), and was used as a screening question prior to implementing the HHIE. Binary responses and responses framed in reference to the grade of hearing difficulty (none, slight, moderate severe) were good predictors of measured hearing loss. In contrast, nonbinary responses framed in reference to the frequency or duration of hearing difficulty (yes, sometimes, no) was extremely poor predictors of measured hearing loss and not at all associated with audiometric assessment. This suggests such measures are not an adequate alternative to audiometric screening and that the context in which a question is asked may also influence response propensities.

Detailed, self-reported hearing loss could be useful in screening for hearing impairment, and in accounting for the multiple, independent, impacts now described for this sensory impairment (Gopinath et al., 2010; Gopinath, 2009b; Karpa et al., 2010; Mitchell et al., 2009; Schneider et al., 2010; Tay et al., 2006). Hickson et al. (1999) argue that objective and subjective measures may be tapping different disability profiles and that both are needed to select clients for rehabilitation. In their study, 15% of participants classified with normal hearing by audiometry reported hearing difficulty, whereas 17.5% of those with moderate or greater audiometric hearing loss reported no hearing difficulty. Our findings do not discount the efficacy of self-reported hearing in identifying disability due to hearing loss (Newman, Weinstein, Jacobson, & Hug, 1990). Perceived health is often reported to be a stronger predictor of quality of life and well-being than actual illness. Therefore, self-report items do have clinical utility and play a role in determining social burden of hearing loss.

There were study differences in self-reported hearing loss. Some items made specific reference to hearing ability with a hearing aid, whereas others did not explicitly distinguish between perceived hearing ability either with or without a hearing aid. There were qualitative differences in the nature of hearing loss (“problems,” “difficulty,” “loss,” “adequacy”), temporal reference frame (“ever,” “12 months,” “currently”) and response options (“binary” vs. rating scale of either the “frequency” of hearing problems or “severity” of hearing problems). This variation in self-report measures across studies is indicative of two different measures of hearing loss. Questions referring to “hearing loss” with a binary response are more likely to be perceived as directly asking about physical hearing loss; whereas questions referring to the frequency of hearing difficulties, or require a graded judgment concerning hearing-related problems, may lead the respondent to consider in what contexts and how regularly they experience difficulties due to hearing loss. A comparison of the base-line self-report item with subsequent waves within ALSA also revealed that, despite the sample becoming more homogonous with respect to age, reference to hearing aids reduced the association between self-report and audiometric data. Thus, consideration of hearing difficulties when using a hearing aid (if owned) is also more likely to elicit a response that reflects hearing disability.

The need to dichotomize responses to self-reported hearing loss items when creating harmonized measures resulted in information loss and reduced variability in the data. However, harmonization improves the comparability of data across studies, enabling future investigations into hearing disability using the pooled data set. Pooled analyses of the harmonized self-report variable will require careful interpretation and consideration of how study differences may influence findings.

In summary, our findings suggest that self-reported hearing measures are not sufficiently sensitive to be used to estimate prevalence or incidence of hearing impairment, particularly for adults of working age (below 65 years) and adults aged above 75 years. Instead, these measures may be useful as broad indicators of perceived hearing disability and when investigating impacts of hearing loss on health and well-being. Given the increase in prevalence of hearing impairment due to population aging, there is a need for epidemiological studies to combine both objective measures of hearing loss with measures that evaluate the impact of hearing loss on functioning and healthy aging.

Footnotes

Acknowledgements

The authors would like to thank Julie Byles, Mary A. Luszcz, and Olivier Piguet. They would like to thank the participants for volunteering their time to be involved in the respective studies. Details of all studies contributing data to DYNOPTA, including individual study leaders and funding sources, are available on the DYNOPTA website ().

The data on which this research is based were drawn from several Australian longitudinal studies including the following: the Australian Longitudinal Study of Ageing (ALSA), the Australian Longitudinal Study of Women’s Health (ALSWH), the Blue Mountain Eye Study (BMES), the Canberra Longitudinal Study of Ageing (CLS), the Melbourne Longitudinal Studies on Healthy Ageing (MELSHA), the Personality And Total Health Through Life Study (PATH), and the Sydney Older Persons Study (SOPS). These studies were pooled and harmonized for the Dynamic Analyses to Optimise Ageing (DYNOPTA) project. The findings and views reported in this article are those of the author(s) and not those of the original studies or their respective funding agencies.

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

This work was supported by the National Health and Medical Research Council (NHMRC) Grant # 410215. Professor Anstey is funded by NHMRC Fellowship #366756.

References

Access Economics. (2006). Listen hear. Canberra, Australia: CRC Hear and the Victorian Deaf Society.

Anstey

K. J.

Byles

J. E.

Luszcz

M. A.

Mitchell

Steel

Booth

. . . Kendig

(2010). Cohort profile: The Dynamic Analyses to Optimise Ageing (DYNOPTA) Project. International Journal of Epidemiology, 39(1), 44-51.

Anstey

K. J.

Christensen

Butterworth

Easteal

Mackinnon

Jacomb

. . . Jorm

A. F.

(2011). Cohort profile: The PATH through life project. International Journal of Epidemiology. doi: 10.1093/ije/dyr025

Browning

C. J.

Kendig

(2010). Cohort Profile: The Melbourne Longitudinal Studies on Healthy Ageing Program. International Journal of Epidemiology, 39(5), 44-51

Caban

A. J.

Lee

D. J.

Gomez-Marin

Lam

B. L.

Zheng

D. D.

(2005). Prevalence of concurrent hearing and visual impairment in US Adults: The National Health Interview Survey, 1997-2002. American Journal of Public Health, 95, 1940-1942.

Chia

E. M.

Wang

J. J.

Rochtchina

Cumming

R. R.

Newall

Mitchell

(2007). Hearing impairment and health-related quality of life: The Blue Mountains Hearing Study. Ear and Hearing, 28, 187-195.

Christensen

Mackinnon

Jorm

A. F.

Korten

Jacomb

Hofer

S. M.

Henderson

(2004). The Canberra Longitudinal Study: Design, aims, methodology, outcomes and recent empirical investigations. Aging, Neuropsychology, and Cognition, 11, 169-195.

Cox

R. M.

Alexander

G. C.

Gray

G. A.

(2007). Personality, hearing problems, and amplification characteristics: contributions to self-report hearing aid outcomes. Ear and Hearing, 28, 141-162.

Cruickshanks

K. J.

Tweed

T. S.

Wiley

T. L.

Klein

B. E.

Klein

Chappell

. . . Dalton

D. S.

(2003). The 5-year incidence and progression of hearing loss: the epidemiology of hearing loss study. Archives of Otolaryngology — Head and Neck Surgery, 129, 1041-1046.

10.

Davis

Smith

Ferguson

Stephens

Gianopoulos

(2007). Acceptability, benefit and costs of early screening for hearing disability: a study of potential screening tests and models. Health Technology Assessment, 11, 1-294.

11.

Gates

G. A.

Cooper

J. C.

Jr. Kannel

W. B.

Miller

N. J.

(1990). Hearing in the elderly: the Framingham cohort, 1983-1985. Part I. Basic audiometric test results. Ear and Hearing, 11, 247-256.

12.

Gopinath

Flood

V. M.

McMahon

C. M.

Burlutsky

Smith

Mitchell

(2010). The effects of smoking and alcohol consumption on age-related hearing loss: the Blue Mountains Hearing Study. Ear and Hearing, 31, 277-282.

13.

Gopinath

Schneider

Rochtchina

Leeder

S. R.

Mitchell

(2009a). Association between age-related hearing loss and stroke in an older population. Stroke, 40, 1496-1498.

14.

Gopinath

Wang

J. J.

Schneider

Burlutsky

Snowdon

McMahon

C. M.

. . . Mitchell

(2009b). Depressive symptoms in older adults with hearing impairments: The Blue Mountains Study. Journal of the American Geriatrics Society, 57, 1306-1308.

15.

Heckhausen

Brim

O. G.

(1997). Perceived problems for self and others: self-protection by social downgrading throughout adulthood. Psychology and Aging, 12, 610-619.

16.

Hickson

Lind

Worrall

Yiu

Barnett

Lovie-Kitchin

(1999). Hearing and vision in healthy older Australians: Objective and self-report measures. Advances in Speech-Language Pathology, 1(2), 95-105.

17.

Hogan

O’Loughlin

Miller

Kendig

(2009). The health impact of a hearing disability on older people in Australia. Journal of Aging and Health, 21, 1098-1111.

18.

Johnson

S. B.

(2010). Study finds self-report valid for assessing hearing loss incidence in farm population. Hearing Journal, 63(9), 46-47.

19.

Karpa

M. J.

Gopinath

Beath

Rochtchina

Cumming

R. G.

Wang

J. J.

Mitchell

(2010). Associations between hearing impairment and mortality risk in older persons: The Blue Mountains Hearing Study. Annals of Epidemiology, 20, 452-459.

20.

Kramer

S. E.

Kapteyn

T. S.

Kuik

D. J.

Deeg

D. J. H.

(2002). The association of hearing impairment and chronic diseases with psychosocial health status in older age. Journal of Aging and Health, 14(1), 122-137.

21.

Lee

Dobson

A. J.

Brown

W. J.

Bryson

Byles

Warner-Smith

Young

A. F.

(2005). Cohort profile: The Australian Longitudinal Study on Women’s Health. International Journal of Epidemiology, 34, 987-991.

22.

Luszcz

M. A.

Giles

Eckermann

Edwards

Browne-Yung

Hayles

(2007). The Australian Longitudinal Study of Ageing: 15 years of ageing in South Australia. Adelaide, Australia: South Australian Department of Families and Communities.

23.

Mitchell

Gopinath

McMahon

C. M.

Rochtchina

Wang

J. J.

Boyages

S. C.

Leeder

S. R.

(2009). Relationship of type 2 diabetes to the prevalence, incidence and progression of age-related hearing loss. Diabetic Medicine, 26, 483-488.

24.

Mohr

P. E.

Feldman

J. J.

Dunbar

J. L.

McConkey-Robbins

Niparko

J. K.

Rittenhouse

R. K.

Skinner

M. W.

(2000). The societal costs of severe to profound hearing loss in the United States. International Journal of Technology Assessment in Health Care, 16, 1120-1135.

25.

Muthen

L. K.

Muthen

B. O.

(2007). Mplus user guide (5th ed.). Los Angeles, CA: Muthen & Muthen.

26.

Newman

C. W.

Weinstein

B. E.

Jacobson

G. P.

Hug

G. A.

(1990). The Hearing Handicap Inventory for Adults: Psychometric adequacy and audiometric correlates. Ear and Hearing, 11, 430-433.

27.

Nondahl

D. M.

Cruickshanks

K. J.

Wiley

T. L.

Tweed

T. S.

Klein

B. E.

(1998). Accuracy of self-reported hearing loss. Audiology, 37, 295-301.

28.

Pascolini

Smith

(2009). Hearing Impairment in 2008: A compilation of available epidemiological studies. International Journal of Audiology, 48, 473-485.

29.

Piguet

Grayson

D. A.

Creasey

Bennett

H. P.

Brooks

W. S.

Waite

L. M.

Broe

G. A.

(2003). Vascular risk factors, cognition and dementia incidence over 6 years in the Sydney Older Persons Study. Neuroepidemiology, 22, 165-171.

30.

Reuben

D. B.

Walsh

Moore

A. A.

Damesyn

Greendale

G. A.

(1998). Hearing loss in community-dwelling older persons: national prevalence data and identification using simple questions. Journal of the American Geriatrics Society, 46, 1008-1011.

31.

Sargent-Cox

K. A.

Anstey

K. J.

Luszcz

M. A.

(2008). Determinants of self-rated health items with different points of reference: Implications for health measurement of older Adults. Journal of Aging and Health, 20, 739-761.

32.

Schneider

Gopinath

Karpa

M. J.

McMahon

C. M.

Rochtchina

Leeder

S. R.

Mitchell

(2010). Hearing loss impacts on the use of community and informal supports. Age and Ageing, 39, 458-464.

33.

Seeman

T. E.

Merkin

S. S.

Crimmins

E. M.

Karlamangla

A. S.

(2010). Disability trends among older Americans: National Health And Nutrition Examination Surveys, 1988-1994 and 1999-2004. American Journal of Public Health, 100(1), 100-107.

34.

Shield

(2006). Evaluation of the social and economic costs of hearing impairment: 2006. Retrieved from http://www.hear-it.org/multimedia/Hear_It_Report_October_2006.pdf

35.

Sindhusake

Mitchell

Smith

Golding

Newall

Hartley

Rubin

(2001). Validation of self-reported hearing loss. The Blue Mountains Hearing Study. International Journal of Epidemiology, 30, 1371-1378.

36.

Tay

Wang

J. J.

Kifley

Lindley

Newall

Mitchell

(2006). Sensory and cognitive association in older persons: Findings from an older Australian population. Gerontology, 52, 386-394.

37.

Ventry

I. M.

Weinstein

B. E.

(1982). The Hearing Handicap Inventory for the Elderly—A new tool. Ear and Hearing, 3(3), 128-134.

38.

Wiley

T. L.

Chappell

Carmichael

Nondahl

D. M.

Cruickshanks

K. J.

(2008). Changes in hearing thresholds over 10 years in older adults. Journal of the American Academy of Audiology, 19, 281-292.

39.

Willis

T. A.

(1981). Downward comparison principles in social psychology. Psychological Bulletin, 90, 245-227.

40.

Wilson

D. H.

Walsh

P. G.

Sanchez

Davis

A. C.

Taylor

A. W.

Tucker

Meagher

(1999). The epidemiology of hearing impairment in an Australian adult population. International Journal of Epidemiology, 28, 247-252.

41.

World Health Organization. (1999). WHO Ear and Hearing Disorders Survey (No. WHO/PBD/PDH/99.8(1)). Geneva, Switzerland: Author.