Agreement between audiometric hearing loss and self-reported hearing difficulty on the Revised Hearing Handicap Inventory differs by demographic factors

Abstract

Background:

New standardized measures of self-reported hearing difficulty can be validated against audiometric hearing loss. This study reports the influence of demographic factors (age, sex, race, socioeconomic position [SEP]) on the agreement between audiometric hearing loss and self-reported hearing difficulty.

Methods:

Participants were 1558 adults (56.9% female; 20.0% racial Minority; mean age 63.7 [SD 14.1] years) from the Medical University of South Carolina Longitudinal Cohort Study of Age-Related Hearing Loss (1988-current). Audiometric hearing loss was defined as the average of pure-tone thresholds at frequencies 0.5, 1.0, 2.0, and 4.0 kHz >25 dB HL in the worse ear. Self-reported hearing difficulty was defined as ≥6 points on the Revised Hearing Handicap Inventory (RHHI) or RHHI screening version (RHHI-S). We report agreement between audiometric hearing loss and the RHHI(-S), defined by sensitivity, specificity, accuracy, positive and negative predictive value, and observed minus predicted prevalence. Estimates were stratified to age group, sex, race, and SEP proxy.

Results:

The prevalence of audiometric hearing loss and self-reported hearing difficulty were 49.0% and 48.8%, respectively. Accuracy was highest among participants aged <60 (77.6%) versus 60–70 (71.4%) and 70+ (71.9%) years, for White (74.6%) versus Minority (68.0%) participants and was similar by sex and SEP proxy. Generally, agreement of audiometric hearing loss and RHHI(-S) self-reported hearing difficulty differed by age, sex, and race.

Conclusions:

Relationships of audiometric hearing loss and self-reported hearing difficulty vary by demographic factors. These relationships were similar for the full (RHHI) and screening (RHHI-S) versions of this tool.

INTRODUCTION

Hearing loss is common in aging adults and is a growing public health concern given its wide-ranging impacts on health and well-being.[1–3] Epidemiological studies have measured hearing by pure-tone audiometry, which is generally considered the ‘gold standard,’ and several studies have evaluated hearing with a single question, but few have utilized standardized questionnaires of self-reported hearing difficulty,[1,4–6] which capture functional impacts of hearing loss on individuals in their daily lives.[7–8] These standardized questionnaires are quick and inexpensive to administer in any location, are predictive of hearing healthcare seeking and treatment outcomes such as hearing aid use,[9–10] and provide more information on the impacts of hearing loss than a single question.

The Revised Hearing Handicap Inventory (RHHI) is a standardized measure of hearing difficulty adapted from the widely used Hearing Handicap Inventory for the Elderly (HHIE) and Adults (HHIA) [HHIE/A].[11–13] The RHHI overcomes some limitations of the HHIE/A. For example, it can be administered to adults of all ages, whereas the HHIE and HHIA are separately administered to older and younger adults, respectively.[11–13] The RHHI is available in full-length (18-item) and screening (10-item; RHHI-S) versions.[11]

Past studies reported the agreement between audiometric hearing and single self-report questions,[6] but few compared agreement using standardized questionnaires of self-reported hearing difficulty.[11,14–15] Studies also show demographic factors, such as age, sex, race, and socioeconomic factors, are related to audiometric hearing and self-reported hearing difficulty on standardized questionnaires.[14–16] However, previous studies were limited in their ability to consider racial differences in agreement between audiometric hearing and standardized questionnaires due to racially homogenous samples of White participants and did not report differences in agreement by socioeconomic factors.[14–15] Importantly, demographic factors may influence interpretation of complex relationships among audiometric hearing loss, self-reported hearing difficulty, and hearing loss outcomes.[17]

A previous study in this cohort reported the sensitivity and specificity of the RHHI(-S) to detect audiometric hearing loss,[11] but did not report whether agreement differs by demographic factors. This information is needed to support the use of the RHHI(-S) as a research tool in epidemiological studies and to correctly interpret research findings related to the RHHI(-S). Understanding these relationships is also relevant when the RHHI(-S) is used as a screening tool to identify the potential presence of hearing loss or as a patient-reported outcome measure.

In this study, we determine the influence of demographic factors, including age, sex, race, and socioeconomic position (SEP), on the agreement between audiometric hearing loss and self-reported hearing difficulty measured by the RHHI and RHHI-S.

METHODS

Study Population

The Medical University of South Carolina (MUSC) Longitudinal Cohort Study of Age-Related Hearing Loss is an ongoing (1988-current) community-based cohort study based in Charleston, South Carolina, USA. The cohort has been described in detail in previous publications.[18–21] Briefly, participants are recruited from the community and continuously enrolled into the cohort. They must be 18 years or older and in good general health, with no evidence of conductive hearing loss or active otologic or neurologic disease. This cross-sectional study uses data from participants’ baseline examination.

At the time of analysis, 1,776 participants had baseline data, including pure-tone audiometry. To be included, participants must also have had complete baseline data for the RHHI(-S) (described below). All protocols for this study were approved by the MUSC Institutional Review Board.

Audiometric Assessment

Pure-tone audiometric thresholds at frequencies 0.25, 0.5, 1.0, 2.0, 3.0, 4.0, 6.0, and 8.0 kHz were measured with an audiometer equipped with TDH-39 headphones (Telephonics Corporation, Farmingdale, NY, USA) in a sound-treated booth. All audiological equipment was calibrated annually to appropriate ANSI standards by manufacturers’ representatives.[22]

Thresholds were measured in 5-dB steps following American Speech-Language-Hearing Association standards.[23] A pure-tone average (PTA) was calculated from thresholds at 0.5, 1.0, 2.0, and 4.0 kHz in each ear. Audiometric hearing loss was defined as a PTA >25 dB HL in the worse ear to capture asymmetrical and unilateral, in addition to bilateral hearing losses.[1,4,11,14] For descriptive purposes, hearing loss severity was classified as mild (>25–40 dB HL), moderate (>40–55 dB HL), moderately severe (>55–70 dB HL), and severe or profound (>70 dB HL).[24]

RHHI and RHHI-S

The HHIE/A was administered by paper and pencil prior to audiometric testing, reporting of hearing health history, and participants’ knowledge of hearing-related test results.[11–13] The HHIE/A each consist of 25 questions (3 questions differ between tools). The RHHI was adapted from the 22 items common to the HHIE/A via psychometric analyses (no new questions were added)[11,25,26] to create an 18-item unidimensional scale of self-reported hearing difficulty. The RHHI-S is the corresponding 10-item unidimensional screening tool, which contains a subset of questions from the RHHI.[11] RHHI and RHHI-S (RHHI-[S]) scores were derived from HHIE/A responses. Like the HHIE/A, response options for the RHHI(-S) are yes, sometimes, or no, which are assigned scores of 4, 2, and 0, respectively, and the total score is the sum of all responses. Therefore, RHHI and RHHI-S scores range from 0 to 72 and 0 to 40, respectively. RHHI(-S) self-reported hearing difficulty is defined as a score ≥6.[11]

Demographic Factors

Participants self-reported age, sex assigned at birth (male/female), race,[27] number of years of education and occupation. For this study, race was classified as White or racial Minority to ensure appropriate statistical power for analyses. A proxy for SEP was determined from participants’ education years and baseline occupation, and was categorized as low, mid, or high (online supplemental file 1).[16,21,28]

Statistical Methods and Measures of Agreement

Statistical analyses were conducted in SAS version 9.4 software (Cary, NC). Demographic differences in sample characteristics between participants with RHHI data (included in this study) and without RHHI data (not included) were determined by chi-square for categorical variables and one-way analysis of variance for continuous variables.

Hot-deck imputation, using the simple random samples with replacement method, was used to account for missing demographic data to avoid loss of statistical power for stratified analyses (described below).[29] Observed values from the sample (donors) were used to impute missing values (recipients). Donor units were randomly selected based on their similarity to recipient units in terms of hearing, demographics, and health history.[18,21,30]

The following measures were used to evaluate the ability of the RHHI(-S) (score ≥6) to detect audiometric hearing loss (PTA >25 dB HL in the worse ear): sensitivity, specificity, accuracy, and observed minus predicted prevalence. Definitions are as follows. Sensitivity is the percentage of participants with audiometric hearing loss who were correctly identified by the RHHI(-S). Specificity is the percentage of participants without audiometric hearing loss who were correctly identified as not having hearing loss by the RHHI(-S). Accuracy is the percentage of participants correctly classified by the RHHI(-S) as having hearing loss or not. Observed minus predicted prevalence is the (observed) prevalence estimate of audiometric hearing loss minus the (predicted) prevalence estimate of RHHI(-S) hearing difficulty.

Positive predictive value (PPV) and negative predictive value (NPV) quantified how well the RHHI(-S) identified individuals with and without audiometric hearing loss. PPV is the probability of an individual having audiometric hearing loss, as determined by their RHHI(-S) score. NPV is the probability of an individual not having audiometric hearing loss, as determined by their RHHI(-S) score.

The parameters (sensitivity, specificity, accuracy, PPV, NPV) with corresponding 95% confidence intervals (CI) were estimated using the Clopper Pearson (exact) method.[31,32] All estimates were stratified to age group (<60, 60–70, 70+ years), sex (female, male), race (White, Minority) and SEP proxy (low, mid, high). The McNemar statistic was used to test for statistical significance for the difference between observed (audiometric hearing loss) minus predicted (RHHI[-S] hearing difficulty) prevalence. Differences are presented as P-values and statistical significance was defined as P<0.05. First, we present analyses related to the agreement between audiometric hearing loss and RHHI self-reported hearing difficulty, then we present the same analyses using the screening version (RHHI-S) to determine differences in agreement with audiometric hearing loss when using full (RHHI) or screening (RHHI-S) versions.

Sensitivity Analyses

Two ad hoc sensitivity analyses were conducted. The first evaluated the agreement between audiometric hearing loss and RHHI self-reported hearing difficulty in Black or African American participants only, because most racial Minority participants were Black or African American. The second evaluated the agreement when race and SEP proxy data were not imputed.

RESULTS

Of the 1776 participants with baseline audiometric data, 1558 also had complete RHHI(-S) data and were included in this study. Participants with missing RHHI(-S) data (n=218, 12.3%) were more likely (vs. participants with complete RHHI(-S) data) to be younger (P<0.01) and have a better PTA (P<0.01), but did not differ by sex, race, or SEP proxy (P>0.05).

The mean age of participants in this study was 63.7 (SD 14.1) years; 56.9% were female and 20.0% were Minority (18.9% of total sample was Black or African American) (Table 1). The prevalence of audiometric hearing loss was 49.0% (n=763).

Table 1:

Study sample characteristics of participants of the Medical University of South Carolina Longitudinal Cohort Study of Age-Related Hearing Loss (1988-current) (n=1558)

Characteristic	Mean (SD) or n (%)
Age (years)	63.7 (14.4)
18–60	415 (26.6%)
60–70	598 (38.4%)
70+	545 (35.0%)
Sex
Female	887 (56.9%)
Male	671 (43.1%)
Pure tone average, worse ear (dB HL)a	27.3 (16.2)
Normal (≤25)	795 (51.0%)
Mild (>25–40)	437 (28.0%)
Moderate (>40–55)	246 (15.8%)
Moderately severe (>55–70)	61 (3.9%)
Severe or profound (>70)	19 (1.2%)
Race
White	1246 (80.0%)
Minority	312 (20.0%)
Socioeconomic position proxy
Low	350 (22.5%)
Mid	390 (25.0%)
High	818 (52.5%)
Year of baseline examination
1988–2000	497 (31.9%)
2001–2010	483 (31.0%)
2011-present	578 (37.1%)

^aCalculated from pure-tone audiometric thresholds at frequencies 0.5, 1.0, 2.0, and 4.0 kHz.

Agreement between audiometric hearing loss and RHHI (full version) self-reported hearing difficulty

The prevalence of RHHI self-reported hearing difficulty was 48.8% (n=760). Table 2 shows the overall and stratified estimates for sensitivity, specificity, accuracy, PPV, NPV, and observed (audiometry) minus predicted (RHHI) prevalence. The observed (audiometry) prevalence was similar to the predicted (RHHI) prevalence (difference 0.2%; p=0.88).

Table 2:

Agreement between audiometric hearing loss and RHHI (full version) self-reported hearing difficulty, defined by sensitivity, specificity, accuracy, positive predictive value (PPV), and negative predictive value (NPV), and observed minus predicted prevalence, overall and stratified by demographic factors. Estimates are shown with corresponding 95% exact confidence intervals (CI).

	n	Sensitivity	Specificity	Accuracy	PPV	NPV	Observed-Predicted Prevalence
		% (95% CI)	% (95% CI)	% (95% CI)	% (95% CI)	% (95% CI)	% (P-value)a
Total	1558	72.5 (69.3, 75.7)	74.0 (70.9, 77.0)	73.2 (71.0, 75.4)	72.8 (69.6, 75.9)	73.7 (70.6, 76.7)	0.2 (P=0.88)
Age (yr)
<60	415	82.6 (74.5, 90.6)	76.3 (71.7, 80.9)	77.6 (73.3, 81.5)	47.7 (39.6, 55.7)	94.4 (91.6, 97.1)	−15.2 (P<0.01)
60–70	598	73.1 (67.9, 78.3)	70.0 (65.0, 75.0)	71.4 (67.6, 75.0)	67.5 (62.1, 72.8)	75.3 (70.5, 80.2)	−3.8 (P=0.08)
70+	545	69.9 (65.4, 74.4)	77.6 (70.8, 84.5)	71.9 (68.0, 75.7)	89.8 (86.4, 93.1)	47.8 (41.4, 54.3)	16.3 (P<0.01)
Sex
Female	887	71.3 (66.6, 76.0)	74.8 (71.1, 78.5)	73.4 (70.4, 76.3)	65.8 (61.1, 70.5)	79.3 (75.8, 82.9)	−3.4 (P=0.05)
Male	671	73.5 (69.2, 77.8)	72.3 (66.9, 77.7)	73.0 (69.5, 76.4)	80.1 (76.0, 84.1)	64.3 (58.9, 69.8)	4.9 (P=0.01)
Race
White	1246	75.3 (72.0, 78.5)	73.7 (70.1, 77.3)	74.6 (72.0, 77.0)	77.0 (73.8, 80.0)	71.9 (68.2, 75.5)	1.2 (P=0.40)
Minority	312	52.2 (42.0, 62.4)	74.6 (68.8, 80.3)	68.0 (62.5, 73.1)	46.2 (36.6, 55.7)	78.9 (73.3, 84.4)	−3.8 (P=0.23)
Socioeconomic position proxy
Low	350	72.0 (65.1, 78.8)	68.8 (62.2, 75.5)	70.3 (65.2, 75.0)	67.1 (60.1, 74.0)	73.6 (67.0, 80.1)	−3.4 (P=0.24)
Mid	390	74.9 (69.1, 80.7)	79.3 (73.4, 85.3)	76.9 (72.4, 81.0)	81.3 (75.9, 86.7)	72.5 (66.2, 78.7)	4.3 (P=0.07)
High	818	71.4 (66.8, 75.9)	74.0 (69.8, 78.1)	72.7 (69.5, 75.8)	71.0 (66.5, 75.5)	74.3 (70.2, 78.4)	−0.2 (P=0.89)

^aP-value generated by McNemar’s test.

Results stratified to demographic factors are shown in Table 2. The sensitivity of the RHHI to detect audiometric hearing loss was highest for participants under 60 years and decreased with age. Specificity was similar across age groups, with participants 60–70 years old having the lowest specificity. Accuracy was highest in adults under 60 years. PPV was lowest in adults under 60 years and increased with age. As shown by the observed-predicted prevalence estimates, adults under 60 years had higher prevalence of RHHI self-reported difficulty than audiometric hearing loss (P<0.01), whereas adults aged 70 or older had higher prevalence of audiometric hearing loss than RHHI self-reported hearing difficulty (P<0.01). The prevalence of RHHI and audiometric hearing loss was similar for participants aged 60 to 70 years.

Sensitivity, specificity, and accuracy were similar for males and females. However, females had lower PPV estimates than males. Females had slightly higher prevalence of RHHI self-reported hearing difficulty than audiometric hearing loss (p=0.05), whereas males had higher prevalence of audiometric hearing loss than RHHI hearing difficulty (p=0.01).

Sensitivity was lower in Minority (vs White) participants and specificity was similar in White and Minority participants. In Minority participants (vs White), accuracy and PPV were lower. For both Minority and White participants, prevalence estimates for audiometric hearing loss and RHHI self-reported hearing difficulty were similar (White: p=0.40; Minority: p=0.23).

Sensitivity was similar across SEP proxy groups, although specificity and accuracy were slightly lower for low SEP proxy, and mid SEP proxy had the highest estimates for sensitivity, specificity, and accuracy. Mid SEP proxy had the highest PPV estimate. Across all SEP groups, prevalence estimates for audiometric hearing loss and RHHI self-reported hearing difficulty were similar (Low: p=0.24; Mid: p=0.07; High: p=0.89).

Agreement between audiometric hearing loss with RHHI-S self-reported hearing difficulty: Differences between the screening (RHHI-S) and full (RHHI) versions

The prevalence of RHHI-S self-reported hearing difficulty was 44.6% (n=695). Sixty-five participants (4.2%) were classified as having hearing difficulty on the RHHI but not the RHHI-S, and no participants were classified as having hearing difficulty on the RHHI-S but not the RHHI. Table 3 shows overall and stratified estimates for sensitivity, specificity, accuracy, PPV, NPV, and observed (audiometry) minus predicted (RHHI-S) prevalence. The observed (audiometry) prevalence was higher than the predicted (RHHI-S) prevalence (difference 4.4%; P<0.01).

Table 3:

Agreement between audiometric hearing loss and RHHI-S (screening version) self-reported hearing difficulty, defined by sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV), and observed minus predicted prevalence, overall and stratified by demographic factors. Estimates are shown with corresponding exact 95% confidence intervals (CI).

	n	Sensitivity	Specificity	Accuracy	PPV	NPV	Observed-Predicted Prevalence
		% (95% CI)	% (95% CI)	% (95% CI)	% (95% CI)	% (95% CI)	% (P-value)a
Total	1558	69.2 (0.66, 0.72)	79.0 (76.2, 81.8)	74.2 (72.0, 76.4)	76.0 (72.8, 79.2)	72.8 (69.8, 75.7)	4.4 (P<0.01)
Age (yr)
<60	415	80.2 (71.8, 88.7)	80.2 (76.0, 84.6)	80.2 (76.1, 84.0)	51.5 (43.0, 60.0)	94.0 (91.2, 96.7)	−11.6 (P<0.01)
60–70	598	68.7 (63.3, 74.2)	75.9 (71.2, 80.5)	72.6 (68.8, 76.1)	70.8 (65.3, 76.2)	74.0 (69.3, 78.7)	1.3 (P=0.53)
70+	545	67.2 (62.6, 71.8)	83.2 (77.1, 89.3)	71.4 (67.4, 75.1)	91.8 (88.7, 95.0)	47.4 (41.2, 53.6)	19.8 (P<0.01)
Sex
Female	887	68.0 (62.9, 72.5)	80.3 (76.9, 83.7)	75.2 (72.2, 78.0)	70.0 (65.2, 74.9)	78.5 (75.1, 82.0)	1.4 (P=0.42)
Male	671	70.5 (66.1, 75.0)	76.4 (71.3, 81.5)	72.9 (69.3, 76.2)	81.9 (77.9, 85.9)	63.2 (57.9, 68.4)	8.3 (P<0.01)
Race
White	1246	72.0 (68.6, 75.4)	78.8 (75.4, 82.1)	75.1 (72.6, 77.5)	79.8 (76.6, 83.0)	70.7 (67.2, 74.2)	5.3 (P<0.01)
Minority	312	48.9 (38.7, 59.1)	79.6 (74.2, 84.9)	70.5 (65.1, 75.5)	50.0 (40.0, 60.0)	78.8 (73.5, 84.2)	0.6 (P=0.83)
Socioeconomic position proxy
Low	350	67.7 (60.5, 74.8)	74.2 (67.9, 80.5)	71.1 (66.1, 75.8)	69.8 (62.7, 77.0)	72.3 (65.9, 78.6)	1.4 (P=0.62)
Mid	390	72.1 (66.1, 78.1)	83.8 (78.4, 89.2)	77.4 (73.0, 81.5)	84.2 (79.0, 89.5)	71.4 (65.3, 77.5)	7.9 (P<0.01)
High	818	68.2 (63.6, 72.9)	79.1 (69.6, 82.9)	74.0 (70.8, 76.9)	74.4 (69.9, 79.0)	73.6 (69.6, 77.6)	3.9 (P=0.03)

^aP-value generated by McNemar’s test.

Results stratified to demographic factors are shown in Table 3. In general, the RHHI-S showed similar trends to the RHHI in its agreement with audiometric hearing loss after stratification to demographic factors, although relationships were slightly different for observed minus predicted prevalence estimates. Differences include i) females had similar observed and predicted (RHHI-S) prevalence estimates (p=0.42), ii) White participants had higher prevalence of audiometric hearing loss than RHHI-S self-reported hearing difficulty (P<0.01), and iii) participants classified as mid or high SEP proxy had higher prevalence of audiometric hearing loss than RHHI-S self-reported hearing difficulty (P<0.01, p=0.03, respectively).

Sensitivity analyses

The agreement between audiometric hearing loss and RHHI self-reported hearing difficulty were similar for i) Black or African American participants only (online supplemental file 2) versus all racial Minority participants (Table 2), and ii) non-imputed (online supplemental file 3) versus imputed data (Table 2).

DISCUSSION

In this study, the agreement of audiometric hearing loss and self-reported hearing difficulty measured by the RHHI(-S) differed by demographic factors. Demographic differences were particularly salient for age, sex, and race, and less so for SEP. Agreement between self-reported hearing difficulty and audiometric hearing loss, and differences by demographic factors, were similar for the RHHI (full version) and screening RHHI-S (screening version). Understanding the influence of demographic factors on the agreement between audiometric hearing loss and self-reported hearing difficulty is integral to appropriately interpreting research using the RHHI(-S).[17]

We compare our findings to the few epidemiological studies that reported demographic differences in the agreement of audiometric hearing loss, defined as a worse ear[14] or better ear[15] PTA, with HHIE-S scores, a standardized measure of hearing handicap, as it is similar to the RHHI(-S).[14,15] Those studies defined HHIE-S self-reported hearing difficulty as a score >8, whereas this study defines RHHI(-S) self-reported hearing difficulty as a statistically derived score ≥6.[11] Consistent with our findings, studies in the population-based Epidemiology of Hearing Loss Study (EHLS) and Blue Mountains Hearing Study (BMHS) showed younger adults self-reported more hearing difficulty than audiometric hearing loss, whereas older adults self-reported less. It has been hypothesized older adults may consider hearing difficulty a normal process of aging, whereas younger adults may consider it a more serious health problem.[14,15] Similar to our findings, a study in EHLS reported the PPV of the HHIE-S to detect audiometric hearing loss was higher for males than females.[14] In contrast to our findings, that study reported marked sex differences in sensitivity, specificity, and accuracy, whereas those estimates were similar by sex in this study. Furthermore, the study in EHLS reported the HHIE-S had limited sensitivity (34%) to detect audiometric hearing loss in the whole cohort, whereas our estimate of sensitivity was much higher (73%) and more like the estimate for individuals with moderate hearing loss reported in a study in BMHS (80%). Differences in findings across cohorts may be explained by different measurement tools and definitions of audiometric hearing loss and self-reported hearing difficulty, sample size, and/or differences in sample characteristics, such as a higher proportion of Minority participants in this cohort.[14,15] Studies conducted in EHLS or BMHS did not report racial or socioeconomic differences in agreement between audiometric hearing loss and self-reported hearing difficulty.[14,15]

Parameter estimates are interrelated and influenced by prevalence estimates, which vary by definitions used. For example, we defined audiometric hearing loss as a PTA in the worse ear,[11,14,16] which captures more cases, including unilateral or asymmetrical hearing losses, than a definition based on the better ear.[15] Unilateral or asymmetrical hearing losses can be impactful, including on communication-related outcomes.[33] In this sample, 89% of participants with audiometric hearing loss defined by worse ear PTA also had hearing loss defined by better ear PTA. Defining audiometric hearing loss by better (instead of worse) ear thresholds would have resulted in minor changes to sensitivity (+4.1%) and specificity (−5.6%), which is consistent with a lower prevalence of audiometric hearing loss, and would not have changed trends by demographic factors.

For a given sensitivity and specificity, PPV estimates increase and NPV estimates decrease as prevalence increases. In this study, the agreement between audiometric hearing loss and RHHI(-S) self-reported hearing difficulty was poorer for Minority (vs White) individuals. In part, this discrepancy may be explained by differences in age distributions between Minority (mean 58 [SD 15.1] years) and White (mean 65 [SD 13.5] years) participants in this sample, and the lower prevalence of audiometric hearing loss in Black/African American individuals in this cohort (Minority: 29.5%; White: 53.9%) and others,[34–35] which exists after adjustment for age (data not shown). Minor differences in parameter estimates between the RHHI and RHHI-S may be explained by the prevalence of RHHI self-reported hearing difficulty being slightly higher (4.6%) than for the RHHI-S.

The agreement between audiometric hearing loss and self-reported hearing difficulty in this cohort was reported as reasonably high but not perfect.[11] This suggests the RHHI(-S) captures functional impacts of hearing loss on individuals not fully explained by audiometric PTA. Disability and health are holistic constructs capturing not only bodily impairment, but also impacts on function, activity limitations, and participation restrictions.[36] To this end, audiometric hearing loss has been described as a measure of bodily impairment whereas self-reported hearing difficulty has been identified as a multidimensional construct capturing function, activity limitations, and participation restrictions, and thus the disease burden of hearing loss.[7] This notion is supported by research indicating self-reported hearing difficulty is a strong predictor of hearing healthcare seeking and treatment uptake.[9–10,37] Research has used self-reported hearing handicap (HHIE/A) to predict hearing-related outcomes,[8,38] and to measure the effectiveness of hearing-related interventions.[39] Taken together, measures of self-reported hearing difficulty, such as the RHHI(-S), may be relevant, versatile, and accessible research tools that can be incorporated into hearing-related epidemiological studies.

Strengths of this community-based cohort study include its large and diverse sample and multiple hearing-related measures. This cohort is comparable to other epidemiological studies in terms of age and audiometric hearing.[4,18,40] Audiometric hearing loss prevalence was high, which supported the evaluation of measures of agreement. However, some limitations exist. Most Minority participants were Black/African American, so differences by other races could not be evaluated. Other demographic factors not considered in this study may influence relationships of audiometric and self-reported hearing loss. This study was conducted in the same cohort from which the RHHI(-S) was developed and therefore, results must be validated in another cohort.[11] Nonetheless, this study builds upon previous research to report demographic differences in the agreement between audiometric hearing loss and RHHI(-S) self-reported hearing difficulty.[11,16]

CONCLUSION

The agreement of audiometric hearing loss and RHHI(-S) self-reported hearing difficulty differed by demographic factors, namely age, sex, and race, in this diverse population across the age range and with a high prevalence of hearing loss.

KEY MESSAGES.

What is already known on this topic

Some validation studies evaluate the agreement between single self-reported questions of hearing difficulty and audiometric hearing, which is often considered the ‘gold standard’ measure of hearing, and how the agreement differs by demographic factors. However, few studies use standardized measures of self-reported hearing difficulty, such as the Revised Hearing Handicap Inventory (RHHI). These standardized measures can be used in epidemiological research as a predictor of hearing-related outcomes, or as an outcome measure to determine the functional impacts of hearing loss on individuals in their daily lives and the impacts of hearing-related interventions, such as hearing aid use. Understanding these relationships is integral for accurate interpretation of research findings from studies that use the RHHI or similar standardized questionnaires of self-reported hearing difficulty.

What this study adds

Results from this community-based cohort study of the general population show that the agreement of audiometric hearing loss and RHHI self-reported hearing difficulty differs by age, sex, and race on measures of sensitivity, specificity, accuracy, positive and negative predicted value, and observed minus predicted prevalence. For example, accuracy was highest among participants aged <60 versus 60–70 and 70+ years, and for White versus Minority participants and was similar by sex and SEP proxy.

How this study might affect research, practice, or policy

Results from this study can be applied to the design and interpretation of epidemiological studies that include standardized self-reported measures of hearing difficulty. Furthermore, when the RHHI is used as a patient-reported outcome measure in clinical settings, results from this study can inform clinical understanding that, on average, self-reported hearing difficulties for some groups of individuals may differ (in both directions) from audiometric hearing loss.

DATA AVAILABILITY STATEMENT

Deidentified participant data are available upon reasonable request to the corresponding author under a data use agreement.