Association of the use of hearing aids with the conversion from mild cognitive impairment to dementia and progression of dementia: A longitudinal retrospective study

Abstract Introduction Hearing aid usage has been linked to improvements in cognition, communication, and socialization, but the extent to which it can affect the incidence and progression of dementia is unknown. Such research is vital given the high prevalence of dementia and hearing impairment in older adults, and the fact that both conditions often coexist. In this study, we examined for the first time the effect of the use of hearing aids on the conversion from mild cognitive impairment (MCI) to dementia and progression of dementia. Methods We used a large referral‐based cohort of 2114 hearing‐impaired patients obtained from the National Alzheimer's Coordinating Center. Survival analyses using multivariable Cox proportional hazards regression model and weighted Cox regression model with censored data were performed to assess the effect of hearing aid use on the risk of conversion from MCI to dementia and risk of death in hearing‐impaired participants. Disease progression was assessed with Clinical Dementia Rating Sum of Boxes (CDR‐SB) scores. Three types of sensitivity analyses were performed to validate the robustness of the results. Results MCI participants that used hearing aids were at significantly lower risk of developing all‐cause dementia compared to those not using hearing aids (hazard ratio [HR] 0.73, 95% confidence interval [CI], 0.61 to 0.89; false discovery rate [FDR] P = 0.004). The mean annual rate of change (standard deviation) in CDR‐SB scores for hearing aid users with MCI was 1.3 (1.45) points and significantly lower than for individuals not wearing hearing aids with a 1.7 (1.95) point increase in CDR‐SB per year (P = 0.02). No association between hearing aid use and risk of death was observed. Our findings were robust subject to sensitivity analyses. Discussion Among hearing‐impaired adults, hearing aid use was independently associated with reduced dementia risk. The causality between hearing aid use and incident dementia should be further tested.


BACKGROUND
The escalating costs and devastating psychological and emotional impact of dementia on affected individuals, their families, and caregivers makes the prevention, diagnosis, and treatment of dementia a national public health priority worldwide. 1 While the process of drug development to delay the onset of dementia has been slower than initially hoped, there is evidence that behavioral and lifestyle interventions might reduce dementia risk. [2][3][4] Numerous studies have investigated the effects of physical exercise; healthy diet; and management of medical conditions, such as diabetes and heart disease, on cognitive decline and risk of developing dementia. 2,5 However, there is a paucity of research on hearing impairment and dementia. Such research is vital, given the high prevalence of dementia and hearing impairment in older adults, and the fact that both conditions often coexist.
The prevalence of hearing loss is higher among older than younger individuals, with more than 70% of adults aged 70 and older having hearing loss in at least one ear. 6 In the United States, mild hearing loss affects 23% of the population over the age of 12, with moderate hearing loss more prevalent in those over 80 years. 6 More importantly, approximately 23 million adults with hearing loss in the United States do not use hearing aids even though the negative impact of untreated hearing loss has been widely documented. [7][8][9] The low level of hearing aid adoption is associated with stigma and affordability of hearing aids. 8,15 Adults with impaired hearing, who do not wear hearing aids, demonstrate significantly higher rates of depression, anxiety, and other psychosocial disorders. 9,10 Hearing loss has also been associated with increased risk of incident dementia. [11][12][13][14] Despite the observed link between hearing loss and cognition, surprisingly few studies have investigated the association between the use of hearing aids, cognitive decline, and risk for developing dementia. [16][17] One study found little evidence that hearing aids promoted cognitive function but acknowledged that they may be effective in reducing hearing handicap. 18 Other work has showed that hearing aid use was associated with better cognition while controlling for confounding by age, sex, general health, and socioeconomic status. 19 A recent randomized pilot study examined the changes in cognition due to treatment of hearing loss, with some promising results for proximal outcomes (perceived hearing handicap, loneliness) that may mediate a relationship between hearing and cognition. 10 Better understanding of the relationship between use of hearing aids, cognitive function, and risk of dementia has the potential to significantly impact public health, as hearing aids represent a minimally invasive, cost-effective treatment to mitigate the impact of hearing loss on dementia. In fact, it is postulated that up to 9% of dementia cases could be prevented with proper hearing loss management. 14 Slower conversion from mild cognitive impairment (MCI) to dementia and progression of dementia after hearing aid use could potentially lead to the reduced incidence of dementia and extended preservation of functional independence in people with dementia.
In this study, we used data from a large referral-based cohort to examine for the first time the effect of the use of hearing aids on the conversion from MCI to dementia and risk of death. We tested if the use of hearing aids is independently associated with a decreased risk of incident all-cause dementia diagnosis for MCI patients and reduced risk of death in individuals with dementia. We also examined if the rate of cognitive decline is slower for hearing aid users when compared to those not using hearing aids.

Participants
We conducted a retrospective analysis of the demographic and clinical data obtained from the National Alzheimer's Coordinating Center (NACC). 20

Clinical assessment and diagnosis
The incidence of MCI and all-cause dementia was determined based on the clinical diagnosis made by a single clinician or a consensus panel. In addition, we used the Clinical Dementia Rating Sum of Boxes (CDR-SB) scores to assess a decline due to cognitive changes in six functional domains, namely: memory, orientation, judgment and problem solving, community affairs, home and hobbies, and personal care. 26 CDR-SB has a total score range from 0 to 18 points, with higher scores indicating greater cognitive impairment. CDR-SB has been commonly used as a reliable tool for assessing dementia severity. 27,28

Hearing assessment
The information on presence of hearing loss and use of hearing aids were extracted from the NACC-UDS Physical Evaluation form. 29 Information on hearing loss was collected via self-report using a single hearing screening question: "Without a hearing aid(s), is the subject's hearing functionally normal?," which provided possible responses of "yes'' or "no.'' 29 Participants responding "yes'' were defined as

HIGHLIGHTS
• High prevalence of dementia and hearing impairment in older adults. • Hearing aid (HA) use is associated with a lower risk of incident dementia.
• Slower cognitive decline in users than non-users of HA with mild cognitive impairment.
• The relationship between hearing impairment and dementia should be further tested. individuals with hearing loss and those that responded "no'' were excluded from the analysis. The determination of severity of hearing loss was not a part of a standardized clinical evaluation given to NACC participants. Although participants were allowed to wear hearing aids during the cognitive assessments, it was not reported to NACC whether a participant was wearing a hearing aid during cognitive testing. Missing codes were, however, entered when ADRCs had a reason to believe that the test was invalid, including if they were aware that the participant was unable to hear properly. All participants with missing codes were excluded from the analysis.
Information on hearing aid use was collected via self-report using a single question: "Does the subject usually wear a hearing aid(s)?," which provided possible responses of "yes'' or "no.'' 29 Participants responding "yes'' were classified as hearing aid users and those that responded "no'' were classified as non-users of hearing aids. The consistent use or non-use of hearing aids was established if the participant answered F I G U R E 1 Selection of participants for study inclusion. MRI, mild cognitive impairment "yes" or "no" to the above question at every consecutive visit. There was no additional information in the NACC-UDS on the number of hours of daily use of hearing aids, the type of hearing aids used, or the history of hearing aid use before the enrollment in the NACC-UDS.
Accordingly, data from individuals identified with impaired hearing at every annual clinical evaluation, that consistently reported non-use or use of hearing aids as well as functionally normal hearing when wearing hearing aids served as the sample for our study. The subject's hearing was characterized as functionally normal with a hearing aid if there was no evidence of reduced ability to do everyday activities such as listening to the radio/television or talking with family/friends. 29 This information was based on self-report. On the other hand, the group of participants excluded from the analysis consisted of subjects without hearing impairment or with inconsistent hearing impairment labeling as indicated by the records from each follow-up visit. Participants with inconsistent records relating to the use or non-use of hearing aids or without functionally normal hearing when wearing a hearing aid were also excluded. The baseline characteristics of participants included and excluded from the study are shown in Table S1 in supporting information. Among 2114 patients with hearing impairment included in the study, 636 subjects in Group 1 and 710 in Group 2 were classified as using hearing aids.

Statistical analysis
Baseline summary statistics are presented as proportions for categorical data and means with standard deviations (SD) for continuous variables. Unadjusted analyses for comparison of demographic and clinical features between individuals with hearing impairment that used and did not use hearing aids were performed with the Fisher's exact test and unpaired t test. 30 The average annual rate of change in CDR-SB score in individuals using and not using hearing aids was compared by applying Mann-Whitney U method. 30 The assumption of normality of CDR-SB data was tested using Shapiro-Wilk test. 30 A Cox model with censored data was used to study time to incident dementia diagnosis for MCI patients (Group 1) and death for individuals diagnosed with dementia (Group 2). 31  To avoid the inflation of false-positive findings, the Benjamini-Hochberg false discovery rate (FDR) procedure was used to adjust for multiple hypothesis-testing. 34 False discovery-adjusted P values (FDR P) < 0.05 were considered statistically significant.

Sensitivity analysis
To validate the robustness of the main findings, we performed three different sensitivity analyses: (1) propensity score matching to control for measured confounding, (2) the analysis for unmeasured confounding to assess the sensitivity of our main conclusions with respect to confounders not included in our study, and (3) the inverse probability of treatment weighting method to reduce selection bias within our study population. [35][36][37] Propensity scores were generated for hearing aid status using multivariable logistic regression model and adjusting for baseline covariates, including age, sex, education, and CDR-SB. The standardized mean difference between two patient groups, ie, patients with and without hearing aids, was then calculated for each covariate and compared before and after the matching process to determine covariate balance between the two groups. 34 A standardized difference of < 0.1 was considered negligible in the prevalence of a covariate. 34 Sensitivity analysis for unmeasured confounding was conducted to measure the potential influence an unmeasured covariate might have on the HR estimates of the association between hearing aid use and: (1) incident dementia in Group 1; and (2) death in Group 2. 36 We considered prevalence rates for the confounder of 5%, 10%, and 20% in the

Participant characteristics
Baseline demographic and clinical characteristics of participants by hearing aid status are presented in Table 1. Statistically significant differences in the use of hearing aids were found between men and women both in Group 1 (P = 0.01) and Group 2 (P = 0.02), with higher rates of hearing aid use in males. Hearing aid users with baseline MCI were significantly older than MCI individuals not using hearing aids (P < 0.001). Age was comparable for users/non-users of hearing aids in Group 2 (P = 0.32). Hearing aid users with dementia had more years of education completed (P < 0.001). The CDR-SB score was significantly lower in both individuals with baseline MCI and dementia who used hearing aids (P = 0.01 and P < 0.001, respectively).

Hearing aid status and risk of incident all-cause dementia in MCI participants
During the study follow-up, 497 MCI subjects in Group 1 developed dementia. The median time to incident dementia was 2 years for nonhearing aid users and 4 years for hearing aid users. The 5-year overall survival rate, which is the percentage of participants that did not convert to dementia 5 years after the baseline MCI diagnosis, was 19% for non-hearing aid users and 33% for individuals using hearing aids.
In the multivariable Cox proportional hazards regression model, the major risk factor for MCI-to-dementia conversion was the CDR-SB score (HR 1.39, 95%CI, 1.30 to 1.48, FDR P < 0.001; Table 2  In the weighted Cox regression model accounting for time-varying effects, the relationship between the use of hearing aids and mortality risk was not statistically significant (HR 0.98; 95% CI, 0.78 to 1.24; FDR P = 0.89; Table 3). Higher CDR-SB scores were associated with the increased risk of death (HR 1.08; 95% CI, 1.05 to 1.11; FDR P < 0.001).

Hearing aid status and mortality risk in participants with dementia
The average (SD) annual rate of change in CDR-SB score of 0.96 (1.02) for hearing aid users with dementia was not significantly TA B L E 3 Risk of death by hearing aid status for individuals with dementia after adjustment for age, sex, race, education, and CDR-SB score

Sensitivity analysis
The distributions of potential confounders were similar between the hearing aid and non-hearing aid user groups after propensity score matching (standardized difference < 0.1). Again, we found that MCI individuals using hearing aids were at lower risk of developing dementia compared to non-users of hearing aids (HR 0.67, 95%CI, 0.50 to 0.80, FDR P = 0.01; Table 4). No link between hearing aid use and risk of death was found for individuals with dementia (Table 4).
Sensitivity analyses for unmeasured confounding performed for each of two studied groups produced virtually unchanged findings ( Abbreviations: CDR-SB, Clinical Dementia Rating Sum of Boxes; CI, confidence interval; HR, hazard ratio. a All models were adjusted for sex, age, education, and CDR-SB score.
device. As such, we performed additional analysis on a group of hearing aid users that included subjects who still experienced auditory difficulties when wearing a hearing aid (Table S3 in

DISCUSSION
Despite the prevalence of auditory impairment in dementia, hearing loss is often not diagnosed and not treated even though hearing loss has been shown to be an independent risk factor for poorer cognitive function, 38 depression and loneliness, 39 and diminished functional status. 39 Several longitudinal studies indicated that individuals with hearing impairment experience substantially higher risk of inci-dent all-cause dementia. 12,[40][41][42] For instance, the study of Lin et al. 11 observed a strong relationship between hearing loss and risk of developing dementia. While the authors observed no association between use of hearing aids and reduced risk of dementia, they found a strong link between degree of hearing loss and dementia incidence. 11 In Ray et al., 42 the association between cognitive impairment and degree of hearing loss was also observed but only in individuals who did not use hearing aids.
Hypothesized mechanisms explaining the association between hearing impairment and cognitive function included the reallocation of cognitive resources to auditory perceptual processing, [43][44][45] cognitive deterioration due to long-term deprivation of auditory input, 43,46 a common neurodegenerative process in the aging brain, 43 and social isolation caused by both sensory and cognitive loss. 45 In addition, recent findings have suggested that hearing impairment manifested as central auditory dysfunction may be an early marker for dementia. 47 Previous studies concluded that intervention in the form of hearing aids may have a positive effect on cognition 19 and reduce the impact of behavioral and psychological symptoms of dementia. 48 In this study, we investigated the relationship between the use of hearing aids with incidence and progression of dementia. Our results clearly suggest that the use of hearing aids is independently associated with a decreased risk of incident all-cause dementia diagnosis in MCI participants . Statistically significant differences in cumulative survival functions by hearing aid status were found in Group 1. We also observed accelerated cognitive decline, as indicated by change in the CDR-SB score, in non-users of hearing aids in the MCI group. The use of hearing aids was not associated with reduced risk of death in people with dementia. Three different sensitivity analyses confirmed the robustness of our findings.
So far, hearing aid usage has been linked to improvements in cognition as well as psychological, social, and emotional functioning. 13 Amieva et al. 49 showed that non-use of hearing aids was associated with faster cognitive decline as indicated by the accelerated rate of change in Mini-Mental State Examination score. Yet, no significant difference in cognitive decline was observed between hearing-impaired subjects using hearing aids and healthy individuals. 49 The recent study of Maharani et al. 16 adopted a different approach in examining differences in cognitive outcomes of hearing aid use. To prevent potential residual confounding caused by demographic differences between hearing aid users and non-users, the authors analyzed rates of cognitive change before and after hearing aid use in the same individuals.
They reported a significantly slower decline in episodic memory scores after patients started to use hearing aids.
The potential mechanisms behind the association between the use of hearing aids and cognitive loss, in particular the decreased risk for incident dementia, remain to be determined. Possible explanations include optimized communication and increased social engagement, with resulting lower rates of depression and loneliness caused by the use of hearing aids and/or changes to the brain, associated with the reduced impact of sensory deprivation on brain function. 50 There is also the possibility that facilitated access to auditory information for individuals using hearing aids may result in a reduction in cognitive resources consumed by listening and, hence, lead to improved cognitive ability. 50 We acknowledge that biases in our analysis could arise from mul- We minimized this bias by considering only participants with a consistent record of hearing impairment and use/non-use of hearing aids. In this way, we could obtain the true effect of hearing aids on incidence and progression of dementia. The inclusion of participants with noisy labels could likely result in an over-or underestimation of the effect between exposure (hearing aid use) and outcome (incident dementia or death).
It is also worth highlighting that the proportion of participants wearing hearing aids in the final groups was considerably higher than the prevalence of hearing aid use in the general population. 7  Nevertheless, we implemented measures to account for the potential impact of confounding and selection bias in our study. Propensity score matching was applied to control for measured confounding; the analysis for residual confounding was implemented to assess the sensitivity of our main conclusions with respect to confounders not included in our study; and potential selection bias was addressed via the inverse probability of treatment weighting method. Our results remained robust under different assumptions.
Finally, this study relies on self-reported hearing loss, which is far less reliable than audiometric screening. The use of information on hearing impairment via self-report prevents any adjustment for the effect of degree of hearing loss when investigating the impact of hearing aid usage on incidence and progression of dementia. It is also worth noting that even though ADRCs are anticipated to enter missing codes when they have reason to believe that the cognitive test is invalid, including if they are aware that the participant is unable to hear properly, we cannot exclude the possibility that hearing impairment, rather than cognitive function, impacted the ability to complete tasks on cognitive tests in participants with hearing loss who did not have hearing aids. Because verbal instructions that are used during cognitive testing depend significantly on hearing, hearing loss might have contributed to the overestimation of the level of cognitive impairment in some hearing-impaired individuals.
Irrespective of the limitations associated with the present analyses, the fact that significant benefit appears to be derived in delaying conversion of MCI to dementia with hearing aid use warrants further exploration. Properly designed clinical trials will definitively measure the potential benefit of hearing correction in those experiencing hearing loss.

CONCLUSION
Slower conversion from MCI to dementia in individuals using hearing aids suggests that effective identification and treatment of hearing loss may reduce the cumulative incidence of dementia. The competing risk of all-cause mortality and dementia among those with MCI should be examined in future work. One of the findings reviewed above Public health campaigns are needed to demonstrate the scale and impact of hearing loss and increase awareness regarding effective prevention strategies, consequences of inaction, and potential benefits of timely audiological intervention.

ACKNOWLEDGMENTS
The NACC database is funded by NIA/NIH Grant U01 AG016976.
NACC data are contributed by the NIA-funded ADCs: P30 AG019610