Hearing Impairment and Cognitive Decline: Alternative Explanations to Causality

Abstract

Despite growing interest in hearing impairment as a potentially modifiable risk factor for dementia, the association is poorly understood. This has implications for whether treating hearing impairment can prevent or delay onset of dementia, as causation is not the only explanation for the association. In this editorial, we highlight how biases in research studies might account for the reported associations. We suggest future research using different study designs and novel biomarkers to help us overcome methodological limitations. This may allow us to determine the strength of the causal pathways linking hearing impairment to dementia, ultimately informing prevention and treatment strategies.

There is consistent evidence of an association between hearing impairment and cognitive impairment (for meta-analyses see^1,2). Consequently, hearing impairment has been considered a potentially modifiable risk factor for dementia,³ and hearing aids have been suggested as an intervention to help slow cognitive decline.⁴ It is hypothesized that hearing impairment could cause cognitive impairment through several mechanisms: a.) higher cognitive demands due to reduced auditory signal quality,⁵ b.) brain atrophy in regions linked to sensory pathways,⁶ and c.) indirectly through reduced social engagement, mental health, and physical activity.^7–9 These explanations are not mutually exclusive and are described in detail elsewhere.^10–14 However, alternate explanations to causality are plausible.¹⁵ Both, hearing and cognitive impairment are strongly associated with age, have a gradual onset, and have shared etiologies.¹⁰ Moreover, each may impact the measurement of the other.¹⁶ Thus, multiple biological pathways linking hearing impairment to dementia as well as biasing pathways due to methodologic limitations are likely. Estimates for the impact of hearing impairment on dementia will be biased when studies cannot fully account for alternative explanations and sources of bias.¹⁵ Depending on the direction of such biases, the burden of dementia attributable to hearing impairment and the potential of hearing aids to prevent or slow dementia may be over- or under-estimated.

While much focus has been dedicated to how hearing loss may cause cognitive impairment, there has been less focus on describing and studying alternative explanations for the association. Thus, this editorial focuses on describing in our opinion, four main alterative explanations and/or sources of bias: a.) measurement error,^16,17 b.) common causes–shared risk factors including biological aging processes and vascular disease,¹⁰ c.) reverse causation–dementia causing the hearing impairment, and d.) selection bias due to selective survival and dropout in studies.¹⁸

Measurement Error

Differing sensitivities of hearing versus cognitive tests might affect the ability to correctly detect the temporal order of changes. For example, hearing impairment might appear to precede cognitive impairment because measures of auditory function are more sensitive to changes in hearing than cognitive tests are to changes in cognition. Hearing changes can be detected via pure-tone audiometry in midlife and even younger adulthood.^19,20 Many standard cognitive tests, such as memory tests, have a restricted set of items and may thus lack sensitivity to subtle changes. Similarly, studies using health record data could be affected by this issue since hearing tests are often easier to perform, more widely implemented, and conducted earlier in life than cognitive testing and dementia ascertainment.

Moreover, most hearing tests including detection of pure tones in quiet require at least some degree of cognitive capacity/processing, e.g., the person needs to comprehend the task instruction, pay attention to the stimulus presentation and generate a behavioral response.²¹ Accordingly, the performance on a battery of auditory processing measures in older hearing-impaired listeners has been shown to be systematically related to individual differences in cognitive function.²² This is a task-impurity problem, i.e., the outcome (cognitive ability) might affect the measure of the determinant (hearing performance), which might lead to an overestimation of the association.²³ Across studies, the strengths of associations between hearing and cognition vary by the measure of hearing that is being utilized (e.g., pure-tone audiometry vs. speech recognition).^1,24,25 Hearing tests, which involve more cognitive processing, such as speech-in-noise, show stronger associations with cognition and may be more sensitive to this task impurity compared to pure tone testing.²⁶ Some auditory processing measures are more strongly associated with cognitive function than measures of auditory function.²² It is additionally problematic for interpreting results, when cognitive measures are being used that are based on auditory stimulus material. In studies on hearing aid use, the use of hearing aids will increase audibility of the auditory stimulus material and may reduce the auditory cognitive listening effort (or cognitive load) a person experiences while undergoing cognitive testing. A recent study reported better cognitive test performance in a sub-set of hearing aid users versus a control group.⁴ It remains unclear whether cognitive benefits in some individuals are due to the immediately improved auditory signal or long-term changes reflecting disease-modifying effects. Studies could determine this by comparing unaided performance before and after a period of hearing aid usage, as well as investigating immediate improvements in cognitive test performance following hearing aid fitting.²⁷ Alternatively, studies could compare cognitive performance between treatment and control group purely on non-verbal or adapted tests such as the MOCA-H.²⁸ This could reveal if benefits of hearing aids transfer to overall brain health.

Common Causes/ Confounding

Another explanation for the association between hearing loss and cognitive decline is that there is an underlying common cause that links hearing impairment to cognitive decline and dementia risk.²⁹ Common causes can induce confounding if not properly adjusted for, which could result in a spurious association between hearing impairment and dementia even if there was no true causal relationship.¹⁵ Plausible common causes for hearing impairment and dementia include aging, other health and lifestyle factors, and social determinants of health. Aging is considered the result of an accumulation of system-wide cellular damage leading to increased vulnerability, impaired function and age-related pathologies that could impact both hearing and cognition.³⁰ Inflammation, metabolic dysregulation, atherosclerosis and vascular-related risk factors (including diabetes, exercise and smoking) contribute to hearing impairment³¹ and to the development of cognitive decline and dementia.^3,32,33 Social determinants of health are also important fundamental common causes, which are strongly associated with a wide range of health outcomes, including hearing impairment,³⁴ cognitive decline and dementia,³⁵ and their proximal causes. Common causes are an important potential source of bias in observational studies because even with statistical adjustment for possible confounders, results may be susceptible to residual and/or unmeasured confounding, limiting causal interpretations.³⁶

Reverse Causation

While hearing impairment may have a causal effect on dementia, reverse causation is also feasible, i.e., that hearing impairment is a consequence of dementia (or dementia pathology). For hearing impairment to cause dementia it must precede the onset of dementia and with enough time etiologically. The decades-long pathophysiologic development of dementia³⁷ makes establishing the temporal order of hearing impairment and dementia extremely difficult in observational research. Neurodegeneration underlying dementia includes regions important for hearing processing,³⁸ and central auditory dysfunction is common in patients with dementia and hearing loss.^39,40 Individuals with high genetic risk for Alzheimer’s disease (AD) have worse speech-in-noise hearing ability compared to those without genetic risk decades prior to typical dementia diagnosis.⁴¹ Thus, reverse causation (which can also be conceptualized as a common cause) is a plausible explanation of the association between hearing impairment and dementia. The issue of temporality may be compounded by measurement error as described above as particularly central auditory tests rely on cognitive function during test conduction²² and changes in memory recall performance are predictive of subsequent hearing impairment.⁴² If reverse causation explains the association then treating hearing impairment will not affect the progression of dementia, but hearing impairment could be a sign of preclinical dementia.

Selection Bias

Selection bias is another bias that can lead to an overestimate, underestimate, or spurious associations in studies on hearing impairment, hearing aid use and dementia. Selection bias occurs when selection into the study sample is driven jointly by the exposure (e.g., hearing impairment) and the outcome (e.g., dementia), or their causes.¹⁸ Selection bias may occur at study enrollment, such as by restricting studies to only those who survived to older ages or by excluding individuals who have already used hearing aids in studies on hearing aid use. In longitudinal studies, loss-to-follow up may bias towards individuals with hearing impairment and/or those developing dementia, which could induce spurious results, or it could lead to an underestimation of effects due to a healthy survivor bias. For example, a study comparing clinic and home visits found that risk factors for dementia were over-estimated if data were limited to clinic visit data.⁴³ However, in another case, simulation of associations between education and dementia suggested selective survival could underestimate effects.⁴⁴ It is unclear how selection bias might have impacted the reported results on hearing impairment and cognition. Further simulation studies could estimate the potential direction and likelihood of bias. In addition, more studies could employ methods to account for or estimate potential biases, such as inverse probability of selection weights.⁴⁵

Future studies are needed to evaluate each of the above alternative explanations and to address potential sources of bias in earlier studies. Well-designed randomized controlled trials (RCTs) are considered the gold standard approach for establishing causality in research. They are less susceptible to confounding and reverse causation by assigning treatment. Multiple additional large RCTs on hearing aids would ideally be conducted, starting in midlife with follow-up through late life. However, such RCTs would be costly, and might not be feasible or ethical given known benefits of hearing aids on communication and quality of life. Observational studies will be needed to help addressing gaps in RCTs. Large longitudinal studies utilizing AD-specific and AD-non-specific neurodegenerative biomarkers (including blood-based and imaging biomarkers) and/or usage of non-behavioral hearing measures such as auditory evoked potentials might help elucidate specific biologic mechanisms, establish temporal order of events, and overcome previous measurement issues.^46–48 Researchers could also use approaches to enhance causal inference in observational studies. Such designs include target trial emulation,⁴⁹ which frame analyses towards mimicking RCTs; and using statistical methods to account for potential time-varying (measured) confounding and selection bias.⁴⁵ Quasi-experimental approaches including natural experiments and Mendelian randomization, which use genetic information, are generally less susceptible to individual-level confounders and can help evaluate unmeasured confounding.^50,51 A few studies using such approaches have been implemented so far with support for different pathways.^41,52–54 These study designs and methods, however, rely on different assumptions and limitations compared to traditional designs which are also important to consider (see elsewhere^15,49,51). If evidence converges across multiple study designs, this will strengthen conclusions around pathways linking hearing impairment and dementia.

In conclusion, most evidence that links hearing impairment with cognitive decline and dementia risk has been derived from studies that are susceptible to biases. Thus, the conclusion that hearing loss causes dementia might be premature. The combination of different study designs and measurements will help to converge evidence and overcome limitations of any individual methodology. It will be integral to study and quantify each of the pathways linking hearing impairment and dementia to adequately inform prevention and treatment strategies for neurodegeneration and dementia.

Data availability

Not applicable