The Revised Hearing Handicap Inventory and Pure-Tone Average Predict Hearing Aid Use Equally Well

Abstract

Purpose:

This study aimed to (a) compare the Revised Hearing Handicap Inventory (RHHI) and pure-tone average (PTA) in their abilities to predict hearing aid use and (b) report the optimal cut-point values on the RHHI and PTA that predict hearing aid use.

Method:

Participants were from a community-based cohort study. We evaluated the ability of the RHHI and PTA as (a) continuous variables and (b) binary variables characterized by the optimal cut point determined by the Youden Index to predict hearing aid use. RHHI scores range from 0 to 72, and PTA was defined as averaged thresholds at frequencies 0.5, 1.0, 2.0, and 4.0 kHz in the worse ear. We used logistic regression models and receiver operating characteristic curves with corresponding concordance statistics (c-statistics) and 95% confidence intervals (CIs) to determine the predictive ability of models and chi-square tests to determine whether c-statistics were significantly different.

Results:

This study included 581 participants (M_age = 72.9 [SD = 9.9] years; 59.9% female; 14.3% Minority race). The c-statistics for the RHHI (0.79, 95% CI [0.75, 0.83]) and PTA (0.81, 95% CI [0.78, 0.85]), as continuous variables, were not significantly different (p = .25). The optimal cut points for the RHHI and PTA to predict hearing aid use were 6 points and 32.5 dB HL, respectively. The c-statistics for the RHHI (0.72, 95% CI [0.68, 0.76]) and PTA (0.75, 95% CI [0.71, 0.79]), as binary variables, were not significantly different (p = .27).

Conclusion:

The RHHI and PTA are similar in their ability to predict hearing aid use.

Hearing loss is an important public health concern and is common among middle-aged to older adults (Cruickshanks et al., 1998; Nash et al., 2011). Hearing loss has been associated with several negative consequences, particularly those related to communication abilities, psychosocial well-being, health-related quality of life, and cognitive function (Chisolm et al., 2007; Dalton et al., 2003; Dillard et al., 2023; Gopinath et al., 2009; Schubert et al., 2017, 2019). Importantly, some evidence suggests that negative impacts of hearing loss could, in part, be mitigated through treatment with hearing aids (Dillard et al., 2023; Lin et al., 2023; Maharani et al., 2018). However, prevalence and incidence estimates of hearing aid uptake and use among eligible older adults are low. For example, among eligible individuals, the prevalence of hearing aid use has been estimated to range from approximately 14% to 30% (Popelka et al., 1998; Reed et al., 2023; Wu et al., 2021), and estimates of 10-year incidence of hearing aid uptake are approximately 36% (Fischer et al., 2011). Several epidemiological studies conducted in samples of the general population have evaluated factors associated with hearing aid use and, among other factors, have reported that both audiometric and self-reported hearing are predictive of hearing aid use (Dillard et al., 2021; Fischer et al., 2011; Fisher et al., 2015; Hartley et al., 2010; Weycker et al., 2021). However, to the authors' knowledge, no studies have directly compared the ability of audiometric hearing and self-reported hearing difficulty to predict hearing aid use. Such information may inform decision making related to the selection of measurement tools where the goal is to determine hearing aid candidacy or to predict treatment outcomes related to hearing loss.

Although audiometry is often considered the gold standard method for measuring hearing loss and determining hearing aid candidacy, it has some limitations that can restrict its applications. For example, audiometric testing is often expensive and requires expertise and equipment that is not always available in resource-limited settings. For these same reasons, it may not be feasible to incorporate audiometry into some research protocols. On the other hand, measures of self-reported hearing difficulty can be administered via paper and pencil, online, or through mobile or tablet applications, and their administration and interpretation does not require highly trained professionals. Therefore, measures of self-reported hearing difficulty can offer a cost-effective, simple, and practical way to evaluate individuals' perceived hearing and can be used in several clinical and research settings and could contribute to determining hearing aid candidacy.

One example of such a tool is the Revised Hearing Handicap Inventory (RHHI), an 18-item, unidimensional, and psychometrically robust questionnaire that measures hearing difficulty (Cassarly et al., 2020). The RHHI was adapted from the widely used Hearing Handicap Inventory for the Elderly (HHIE) and Adults (HHIA [HHIE/A]) and contains a subset of questions that are also on the HHIE/A (Cassarly et al., 2020; Newman et al., 1990; Ventry & Weinstein, 1982). One advantage of the RHHI is that it can be administered to adults of all ages, whereas the HHIE and HHIA are administered to older and younger adults, respectively (Cassarly et al., 2020; Newman et al., 1990; Ventry & Weinstein, 1982). Importantly, it is beneficial and possible to translate and culturally adapt tools of self-reported hearing difficulty to other languages (Kaspar et al., 2021). Therefore, there are potential global applications for tools of self-reported hearing difficulty that may aid in predicting hearing aid candidacy and, thus, referral to a hearing care provider. Beyond the practical benefits of tools of self-reported hearing difficulty, they are highly valuable because they capture the effects of hearing loss on individuals' functional abilities and likely include constructs related to hearing health that are not captured by audiometry alone (Hickson et al., 2008; Humes & Weinstein, 2021).

One way to enhance the utility of continuous measures of hearing loss, including audiometric hearing and self-reported hearing difficulty, is to determine cut-point values that predict an outcome of interest, such as hearing aid use. Establishing a cut-point value that is predicted to elicit an action can provide added value and simplify the interpretation of continuous measures of hearing loss. For example, in a population-based or primary care–based screening program, an individual's threshold or RHHI score that exceeds a given cut-point value could prompt a referral for audiology services. Furthermore, using statistically derived cut-point values could reduce the reliance on descriptive categories of audiometric hearing loss or self-reported hearing handicap or difficulty, which are not statistically derived (Clark, 1981). Using measures that are simple to interpret in hearing loss screening programs and other settings could result in earlier identification and intervention for hearing loss, which may ultimately reduce the burden of hearing loss on individuals (U.S. Preventive Services Task Force, 2021).

To the authors' knowledge, no studies have compared the ability of audiometric and self-reports of hearing difficulty to predict hearing aid use. Knowledge on the predictive ability of audiometric versus self-reported hearing could inform decision making related to testing protocols that aim to determine hearing aid candidacy. Therefore, the purpose of this study was twofold. First, this study compares the ability of the RHHI and audiometry, defined by a pure-tone average (PTA), to predict hearing aid use. Second, this study reports the optimal cut-point values on the RHHI and PTA that predict hearing aid use.

Method

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, SC. The cohort has been described in detail in previously published articles (Dubno et al., 2008; Lee et al., 2005; Matthews et al., 1997; Simpson et al., 2019). Briefly, participants are continuously enrolled into the cohort and are recruited from the community through advertisements and subject referral. They must be aged 18 years or older and in good general health with no evidence of conductive hearing loss or active otologic or neurologic disease. All protocols for this study were approved by the institutional review board at MUSC (Approval ID: E-607R), and data were collected under informed written consent.

The baseline examination consists of three to six visits that include comprehensive measures of hearing and health and hearing-related history. The battery of tests includes pure-tone air-conduction audiometry, speech recognition measures in quiet and noise, middle ear measurements, otoacoustic emissions, auditory brainstem responses, and surveys focused on demographics and general and hearing-related health, including those related to self-reported hearing difficulty.

After the baseline examination, participants attend annual follow-up visits. At these visits, participants undergo pure-tone audiometry, and update their hearing and health history, including details on hearing aid uptake and use. Every 2–3 years after the baseline examination, participants attend a comprehensive follow-up visit where most of the measures described above, including those related to self-reported hearing difficulty, are repeated.

Currently, there are 1,776 participants with baseline data. To be included in this study, participants were required to have (a) complete RHHI and PTA data from at least two examinations that were at least 6 months apart and, (b) for hearing aid users only, have complete data on the time (year) they acquired hearing aids. Participants were excluded from this study if they reported hearing aid use at the baseline examination. These inclusion and exclusion criteria were implemented to determine incident hearing aid use and to ensure that RHHI and PTA data were collected prior to the time of hearing aid uptake.

Predictors: Audiometric Hearing and RHHI Score

Pure-tone thresholds at frequencies 0.25, 0.5, 1.0, 2.0, 3.0, 4.0, 6.0, and 8.0 kHz were measured with a clinical audiometer equipped with TDH-39 headphones (Telephonics Corporation) in a sound-treated booth. All audiological equipment is calibrated annually to the appropriate American National Standards Institute (ANSI) standards by manufacturers' representatives (ANSI, 2018). Thresholds were measured in 5-dB steps following American Speech-Language-Hearing Association (2005) standards. A PTA was calculated from threshold values at frequencies 0.5, 1.0, 2.0, and 4.0 kHz for each ear. PTA in the worse ear was used for all analyses (Cassarly et al., 2020; Cruickshanks et al., 1998).

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. Before 2015, participants aged 60 years and older and younger than 60 years were administered either the HHIE or HHIA, respectively (Newman et al., 1990; Ventry & Weinstein, 1982). The study protocol was modified in 2015, so that all participants, regardless of age, were administered a version of the HHIE/A that included all 28 items from both tools. The HHIE/A each consist of 25 questions (three questions differ between tools). Responses include three possible answers, yes, sometimes, or no (indicating either disagreement or not applicable), which are assigned scores of 4, 2, and 0, respectively. Scores range from 0 to 100 points and are determined by summing the numbers corresponding to responses, with higher scores indicating greater perceived hearing difficulties.

As described in Cassarly et al. (2020), the RHHI was created using the 22 items common to both the HHIE and HHIA via psychometric analyses, including nonparametric item response theory and Mokken Scale Analysis, which was used to guide decision making for manifest invariant item ordering and scale length reduction (Mokken, 1971; Molenaar, 1971). This approach resulted in the RHHI, an 18-item unidimensional scale of self-reported hearing difficulty, and the RHHI-S, a 10-item screening tool of self-reported hearing difficulty. Like the HHIE/A, response options for the RHHI are yes (4), sometimes (2), and no (0), and the total score is the sum of all responses. Therefore, RHHI scores range from 0 to 72, and RHHI-S scores range from 0 to 40. RHHI scores were derived from HHIE/A responses. The RHHI and RHHI-S questionnaires are in Supplemental Materials S1 and S2, respectively.

Outcome: Hearing Aid Use

At each examination, participants reported whether they currently own hearing aids. If they reported being hearing aid owners, they also reported (a) the year hearing aids were obtained, (b) whether they consider themselves successful hearing aid users and (c) whether they use hearing aids at least twice per week. In this study, hearing aid users were defined as participants who self-reported successful hearing aid use and/or hearing aid use twice per week. All other participants, except for baseline hearing aid users who were excluded from this study (as described above), were classified as nonusers.

For each participant, we used outcome and predictor data from a single time point. For hearing aid users, we used data from the closest examination before participants obtained their hearing aids. For nonusers, we used data from participants' most recent examination with RHHI data.

Statistical Method

All statistical analyses were conducted in SAS (Version 9.4) software. For all analyses, statistical significance was defined by an α level of p < .05. We used chi-square for categorical variables and one-way analysis of variance for continuous variables to determine differences by age, sex, race, and worse ear PTA for participants that were included and excluded from this study. We used the same statistical tests to determine differences in sample characteristics in hearing aid users versus nonusers.

We used logistic regression models and receiver operating characteristic (ROC) curves with corresponding concordance statistics (c-statistics) to determine the performance, in terms of predictive ability, of several models. First, we evaluated the ability of the RHHI and PTA, used as continuous variables, to predict hearing aid use. Second, we evaluated the ability of the RHHI and PTA, used as binary variables characterized by the cut point determined by the Youden Index, to predict hearing aid use. The Youden Index is used to determine cut-point values by selecting the optimal cut point where sensitivity and specificity are equally important (Youden, 1950). For both approaches, RHHI and PTA are modeled separately and then together in a single model. To compare our results with those presented by Cassarly et al. (2020), we also determined the optimal cut point, using the Youden Index, for the RHHI score that would predict audiometric hearing loss, defined as PTA at frequencies 0.5, 1.0, 2.0, and 4.0 kHz > 25 dB HL in the worse ear.

Two planned sensitivity analyses were conducted. First, we determined the optimal cut point for the RHHI screening version (RHHI-S) using the same methods described above, to determine if there were differences in the optimal cut-point value to predict hearing aid use between the RHHI and RHHI-S, or their ability to predict the outcome of hearing aid use. Second, we determined the optimal cut-point values for the RHHI and PTA, using the same methods above, in adults aged 50 years and older only. That age was chosen given World Health Organization (WHO) recommendations that hearing screening could begin in adults aged 50 years and older (WHO, 2021).

ROC curves plot sensitivity (true positive rate) against 1 minus specificity (false positive rate) for different cut-point values, where the positive rate refers to the correct prediction of hearing aid use. C-statistics were calculated as the area under the ROC curve and measured the ability of the models to discriminate between hearing aid users and nonusers and are presented with corresponding 95% confidence intervals (CIs; Kuhn & Johnson, 2013). A c-statistic of < .5 suggests no discrimination, whereas a c-statistic .7 to < .8 is considered acceptable discrimination, ≥ .8 to .9 is considered excellent discrimination, and ≥ .9 is considered outstanding discrimination (Hosmer & Lemeshow, 2000). We used chi-square tests to determine whether c-statistics across different models were significantly different (DeLong et al., 1988). That is, for the models described above, we compared the c-statistics from (a) the RHHI models versus the PTA models and (b) the models with both RHHI and PTA versus RHHI and PTA, separately. Finally, for the models that characterize the RHHI and PTA by the optimal cut points, we report odds ratios (OR) with corresponding 95% CIs, which estimate the association of the predictor of interest (RHHI and/or PTA) with hearing aid use.

Results

Of the 1,776 participants with baseline data, 581 met the inclusion criteria described above and were included in this study. Participants included in this study, as compared to the 1,195 who were excluded, were more likely to be older (p < .01), and White race (p < .01) but did not differ by sex (p = .51), PTA (p = .19), or RHHI score (p = .35).

The characteristics of the 581 participants in the entire sample and by nonusers and hearing aid users, separately, included in this study are in Table 1. In the entire sample, participants' mean age was 72.9 (SD = 9.9) years, 59.9% were female, and 14.3% were Minority race (13.8% of the sample was Black or African American). In the entire sample, the mean PTA and RHHI scores were 30.7 (SD = 14.2) dB HL, and 8.8 (SD = 10.9) points, respectively. The number of hearing aid users was 121 (20.8%). Hearing aid users, as compared to nonusers, did not differ by age (p = .60) but were more likely to be female (p < .01) and have higher PTA (p < .01) and RHHI scores (p < .01).

Table 1.

Study sample characteristics.

Characteristic	Entire sample (N = 581)		Nonusers (n = 460)		Hearing aid users (n = 121)		p valuea
	M or n	SD or %	M or n	SD or %	M or n	SD or %
Age	72.9	9.9	72.8	10.6	73.3	6.6	.60
Female	349	59.9%	168	36.5%	65	53.7%	< .01
Minority race	83	14.3%	—	—	—	—	—
PTA (dB HL)	30.7	14.2	27.5	13.2	42.7	11.4	< .01
RHHI score	8.8	10.9	6.6	9.4	17.0	12.4	< .01

Note. Pure-tone average (PTA) is defined as the average of thresholds 0.5, 1.0, 2.0, and 4.0 kHz in the worse ear. We did not report numbers of nonusers and hearing aid users who were Minority race given low sample size of Minority race hearing aid users. RHHI = Revised Hearing Handicap Inventory.

^ap value presents results from chi-square and one-way analysis of variance tests that determine differences in nonusers versus hearing aid users.

The relationship between the RHHI and PTA for hearing aid users and nonusers is shown in Figure 1. Among both hearing aid users and nonusers, RHHI scores increase with increasing PTA. As supported by results in Table 1, hearing aid users have higher PTA and RHHI scores, on average, than nonusers.

Figure 1.

Relationship between audiometry and RHHI in hearing aid users and nonusers. PTA is defined as the average of thresholds 0.5, 1.0, 2.0, and 4.0 kHz in the worse ear. RHHI = Revised Hearing Handicap Inventory; PTA = pure-tone average.

Predicting Hearing Aid Use With RHHI and PTA as Continuous Predictors

Figure 2 displays the ROC curves and corresponding c-statistics for relationships between the RHHI and PTA, used as continuous variables, and hearing aid use. The c-statistics for the RHHI and PTA, modeled separately, were .79 (95% CI [0.75, 0.83]) and .81 (95% CI [0.78, 0.85]), respectively, which were not significantly different (p = .25). When the RHHI and PTA were modeled together, the c-statistic improved to .84 (95% CI [0.81, 0.87]) and was significantly higher as compared to those for PTA (p < .01) or the RHHI (p < .01) alone.

Figure 2.

Superimposed ROC curves that represent the ability of the RHHI and pure-tone average (PTA) as continuous variables, separately and together, to predict hearing aid use. PTA is defined as the average of thresholds 0.5, 1.0, 2.0, and 4.0 kHz in the worse ear. ROC = receiver operating characteristic; CI = confidence interval; RHHI = Revised Hearing Handicap Inventory.

Optimal Cut-Point Selection

As determined by the Youden Index, the optimal cut point for the RHHI to predict hearing aid use was 6 points, which had a corresponding sensitivity and specificity of 82.6% and 62.0%, respectively. A sensitivity analysis determined the optimal cut point for the RHHI-S (screening version) to predict hearing aid use, which was also 6 points, with a sensitivity of 79.3% and specificity of 66.7%. The optimal cut point for PTA to predict hearing aid use was 32.5 dB HL, the corresponding sensitivity was 87.6%, and the specificity was 62.6%. A second sensitivity analysis that determined the optimal cut points for the RHHI and PTA in participants aged 50 years and older returned the same cut points of 6 points and 32.5 dB HL, respectively.

To facilitate the comparison of results from this study with another study that published data from this cohort (Cassarly et al., 2020), we also determined the cut point for the RHHI to predict audiometric hearing loss, defined as PTA of thresholds at 0.5, 1.0, 2.0, and 4.0 kHz > 25 dB HL in the worse ear, to be 4 points. This corresponded to a sensitivity of 73.9% and specificity of 75.0%. Therefore, our cut point for the RHHI to predict audiometric hearing loss was 2 points lower than the cut point of 6 points reported by Cassarly et al. (2020), and the estimates for sensitivity and specificity were similar to those in that study (sensitivity: 73.2%, specificity 73.8%). Potential reasons for the slight difference in cut points are discussed later.

Predicting Hearing Aid Use With RHHI and PTA Defined by Optimal Cut Points

Figure 3 displays the ROC curves and c-statistics for relationships of hearing aid use with the RHHI cut point of 6 points and the PTA cut point of 32.5 dB HL. The c-statistics for the RHHI and PTA, dichotomized as ≥ 6 versus < 6 points, and ≥ 32.5 versus < 32.5 dB HL, were .72 (95% CI [0.68, 0.76]) and 0.75 (95% CI [0.71, 0.79]), respectively. The c-statistics for RHHI and PTA were not significantly different (p = .27). When the RHHI (score ≥ 6) and PTA (≥ 32.5 dB HL) were modeled together, the c-statistic improved to 0.81 (95% CI [0.77, 0.84]), and was significantly higher as compared to PTA (p < .01) or the RHHI (p < .01) alone.

Figure 3.

Superimposed ROC curves that represent the ability of the optimal cut-point values for the RHHI (≥ 6 points) and pure-tone average (PTA; ≥ 32.5 dB HL), separately and together, to predict hearing aid use. PTA is defined as the average of thresholds 0.5, 1.0, 2.0, and 4.0 kHz in the worse ear. ROC = receiver operating characteristic; CI = confidence interval; RHHI = Revised Hearing Handicap Inventory.

When modeled separately, an RHHI score of ≥ 6 points was associated with nearly eight-fold higher odds of hearing aid use (OR = 7.76, 95% CI [4.67, 12.87]), and a PTA of ≥ 32.5 dB HL was associated with nearly 12-fold higher odds of hearing aid use (OR = 11.83, 95% CI [6.67, 20.98]). When the RHHI and PTA were modeled together using the cut points defined above, the OR for both the RHHI (OR = 4.07, 95% CI [2.36, 6.99]) and PTA (OR = 7.23, 95% CI [3.97, 13.17]) were slightly attenuated but remained significantly associated with hearing aid use.

Discussion

Findings from this study conducted in a community-based sample of the general population suggest that the RHHI, a measure of self-reported hearing difficulty, has similar ability to PTA to predict hearing aid use. Given that the RHHI is a simple and practical tool that evaluates individuals' perceived hearing, it may be possible to utilize the RHHI to predict hearing aid use in settings where audiometry is not feasible. For example, the RHHI could be implemented in hearing loss screening programs to identify individuals who may benefit from treatment for hearing loss.

This study determined that a cut point of 6 points on the RHHI had the optimal sensitivity and specificity for predicting hearing aid use. To compare the results to Cassarly et al. (2020), which was also conducted in participants from this cohort, we also determined the cut point to predict audiometric hearing loss (PTA > 25 dB HL in the worse ear) to be 4 points, which is 2 points lower than the cut point to predict audiometric hearing loss (using the same definition) reported by Cassarly et al. The study by Cassarly et al. used cross-sectional baseline data from 1,447 participants, whereas the present study leveraged the longitudinal design of this cohort and therefore included data from 581 participants who had at least two examinations with RHHI data and who were not hearing aid users at the baseline examination. Therefore, our sample size was substantially lower as compared to the study by Cassarly et al. Taken together, results from the two studies indicate that the optimal RHHI cut point to predict hearing aid use is slightly higher than that to detect audiometric hearing loss. Importantly, there is a need to validate cut points on both the RHHI and PTA to detect hearing aid use, and for the RHHI to predict audiometric hearing loss, in other data sets.

Many studies, conducted in samples of the general population or clinical samples, have reported factors that are predictive of hearing aid use. In epidemiological studies of the general population, higher (worse) PTA, poorer self-perceived hearing, and hearing handicap have been consistently associated with hearing aid use or uptake (Dillard et al., 2021; Fischer et al., 2011, 2015; Gopinath et al., 2011; Hartley et al., 2010; Moon et al., 2015; Popelka et al., 1998; Weycker et al., 2021). On the other hand, results from some studies conducted in clinical populations report that PTA is not associated with hearing aid use or satisfaction (Cox et al., 2007; Hickson et al., 1999). This is likely due to important differences in the composition of samples between studies of the general population and clinical populations, namely, that clinical populations are more likely to have poorer audiometric hearing and are often older. In studies of the general population, some additional factors have been associated with hearing aid use, such as older age, education or socioeconomic factors, and health and lifestyle factors (Bainbridge & Ramachandran, 2014; Dillard et al., 2021; Fisher et al., 2015; Gopinath et al., 2011; Hartley et al., 2010; Moon et al., 2015; Popelka et al., 1998; Weycker et al., 2021). Additional factors associated with hearing aid use reported in studies conducted in clinical populations include social support, attitudes and expectations, and self-efficacy (Hickson et al., 1999, 2014; Humes et al., 2003; Kricos, 2006; Singh & Launer, 2016; Singh et al., 2015). Several reviews have comprehensively summarized factors associated with hearing aid use and related outcomes (McCormack & Fortnum, 2013; Meyer & Hickson, 2012; Ng & Loke, 2015; Vestergaard Knudsen et al., 2010).

Although the studies described above evaluate associations of PTA and self-reported hearing difficulty with hearing aid use, no studies (to the authors' knowledge) have focused on comparing the predictive abilities of the RHHI and PTA alone. Doing so is important to determine the predictive abilities of these commonly used measures of hearing. Importantly, in this study, the c-statistics across all models were considered to provide either acceptable (.7 to < .8) or excellent (≥ .8 to .9) discrimination between hearing aid users and nonusers (Hosmer & Lemeshow, 2000). These results suggest that the RHHI and PTA are highly predictive of hearing aid use, which is supported by the finding that participants with an RHHI score of ≥ 6 points or a PTA of ≥ 32.5 dB HL had nearly eight-fold and 12-fold higher likelihood, respectively, of being hearing aid users as compared to those with RHHI scores of < 6 points or PTA of < 32.5 dB HL. However, it is important to note there was substantial uncertainty in those estimates, which may, in part, be attributed to the relatively low number of hearing aid users in this sample.

In this study, the prediction of hearing aid use improved when both the RHHI and PTA were included in the model. This finding suggests that the RHHI and PTA capture some overlapping, yet some distinct constructs related to hearing. Consistent with that finding, the strength of the association of the RHHI (score ≥ 6 points) and PTA (≥ 32.5 dB HL) with hearing aid use when the RHHI and PTA were in a single model, as compared to when they were modeled separately, were slightly attenuated. This interpretation may be consistent with the definitions of disability and health provided by the WHO: International Classification of Functioning, Disability and Health. That is, audiometry may capture bodily impairment, but the RHHI may capture individuals' function, activity limitations, and participation restrictions and, therefore, the burden of hearing loss (Humes & Weinstein, 2021; WHO, 2001).

Findings from this study are relevant to public health research and practice. In terms of research, findings suggest that the RHHI is a valuable tool that can predict hearing aid use and may support its use in epidemiological studies of hearing loss. Furthermore, upon validation of cut-point values in other cohorts and/or in clinical studies, results from this study could inform screening cut-point values for both the RHHI and PTA that could be used to identify possible hearing aid candidates and refer them to seek audiologic care. A sensitivity analysis showed that the RHHI and RHHI-S (screening version) returned the same optimal cut-point values (score ≥ 6 points) to predict hearing aid use, and similar sensitivity and specificity, which suggests that either the RHHI or RHHI-S could be used in the situations described above. There may be situations where individuals choose to use the RHHI instead of the RHHI-S and vice versa. For example, given that the RHHI has eight more questions than the RHHI-S, it could be used to provide additional details on individuals' hearing difficulties that are relevant for counseling. On the other hand, the RHHI-S is shorter, so it could be used instead of the RHHI when there are strict time limitations in research or clinical settings.

Because the RHHI is a short questionnaire that can be administered to all adults aged 18 years and above and does not require specialized equipment or expertise such as that needed for audiometry, it would be possible to easily administer the tool in a variety of settings and formats relevant for screening, including in population-based screening programs or in clinical settings, such as primary care offices. Although the goal of this study was to compare the predictive ability of the RHHI and PTA, separately, future research should consider whether considering other factors, such as those mentioned above, could substantially improve prediction of hearing aid use.

Strengths of this community-based cohort study are its large and diverse sample that contains measures of audiometric and self-reported hearing. This cohort study is similar to other epidemiological studies of age-related hearing loss conducted in samples of the general population in terms of age and audiometric hearing loss (Cruickshanks et al., 1998; Gates et al., 1990), which enhances the generalizability of study findings. Studies conducted in samples of the general population, such as this one, are important to advance research that is relevant to hearing loss screening, given that screening programs often occur among the general population, rather than in audiology clinics. However, some limitations exist. Study participants live in a relatively small geographic region, which may limit the generalizability of study results. Furthermore, most Minority participants were Black or African American, so results may not be generalizable to other races. As stated above, optimal cut-point values for the RHHI and PTA to predict hearing aid use must be validated in other cohorts. This is important because the agreement between audiometry and self-reported hearing difficulty may differ by demographic factors, including age, sex, and race (Choi et al., 2016) and, therefore, repeating this study in another cohort may return different cut-point values for the RHHI and PTA to predict hearing aid use, and/or may result in differences regarding the comparisons between the predictive ability of the RHHI and PTA. This study defined PTA with a commonly used definition (Cassarly et al., 2020; Cruickshanks et al., 1998), which does not capture hearing loss above 4.0 kHz or other dysfunctions of the auditory system that may present on other tests, such as speech recognition in noise.

Conclusions

This study conducted in a diverse, community-based sample of the general population indicated that the RHHI and PTA are similar in their ability to predict hearing aid use. The cut-point values that were determined to provide the optimal prediction of hearing aid use were an RHHI score of 6 points and a PTA of 32.5 dB HL. Cut-point values must be validated in other cohorts prior to their use. The RHHI is likely a valuable tool that can predict successful hearing aid use, particularly given its ease of administration and low cost.

Data Availability Statement

The data sets generated and/or analyzed during the current study are not publicly available but data are available from the corresponding author on reasonable request.

Funding Statement

This work was funded (in part) by the National Institues of Health/National Institute on Deafness and Other Communication Disorders Individual Postdoctoral Fellowship Grant F32 DC021078 (awarded to L.K.D.), Institutional Training Grant T32 DC014435 (awarded to J.R.D.), and Clinical Research Center Grant P50 DC000422 (awarded to J.R.D.) and by the South Carolina Clinical and Translational Research Institute, with an academic home at the MUSC, National Center for Advancing Translational Sciences Grant UL1 TR001450. This investigation was conducted in a facility constructed with support from Research Facilities Improvement Program Grant C06 RR14516 from the National Center for Research Resources.