Longitudinal study of self-reported hearing loss and subjective cognitive function decline in women

Abstract

INTRODUCTION

We investigated the relation between self-reported hearing loss and risk of subjective cognitive function (SCF) decline among women.

METHODS

We conducted a longitudinal study of 20,193 women in the Nurses’ Health Study aged ≥66 years who reported their hearing status and had no subjective cognitive concerns in 2012. SCF scores were assessed by a 7-item questionnaire in 2012 and 2014. SCF decline was defined as a new report of at least one cognitive concern during follow-up.

RESULTS

Self-reported hearing loss was associated with higher risk of SCF decline. Compared with women with no hearing loss, the multivariable-adjusted odds ratios (95%CI) for incident SCF score ≥1 were 1.35(1.25,1.47), 1.39(1.24,1.56), and 1.40(1.21,1.75) among women with mild, moderate and severe hearing loss, respectively. Recent progression of hearing loss was associated with even higher risk.

DISCUSSION

Self-reported hearing loss was associated with higher risk of incident subjective cognitive function decline in women.

1. Introduction

The disease burden due to Alzheimer’s disease (AD) and related dementias is substantial and a growing public health challenge.¹ Identifying those at higher risk and uncovering early preventive interventions to preserve memory and delay cognitive aging are critical. In 2017, almost 47 million Americans had evidence of preclinical AD that potentially could progress to clinical disease.²

Subjective concerns regarding cognitive function in older adults have been shown to be an indication of non-normative cognitive decline and may be a marker of eventual progression to dementia.^3–6 Associations of subjective cognitive function (SCF) with biologic and neuroimaging markers of AD, such as loss of gray matter volume, cerebral hypometabolism, and deposition of amyloid, have been demonstrated, even in the absence of objective cognitive dysfunction.^7–13 Cognitive concerns may be expressed by individuals up to 15 years before cognitive impairment is clinically manifest^14,15 and may identify a potential target population for early intervention.¹⁶

Hearing loss is common and the prevalence increases with age.^17,18 Among older adults, hearing loss has been associated with poorer cognitive performance and may be a risk factor for accelerated cognitive decline.^19–22 Hearing loss and cognitive decline are each complex, multifactorial and progressive, ostensibly due to auditory and neurodegenerative damage that accumulates over a lifetime.^23,24 Understanding whether hearing loss influences subjective cognitive function could provide needed insight into determinants of early cognitive decline, identify a high-risk group, and reveal a novel approach to curb its trajectory. Self-reported hearing loss was associated with a substantially higher risk of subjective cognitive decline in men;²⁵ however the relation has not been studied in women. Women have a greater lifetime risk of developing AD, which is not fully explained by greater longevity.²⁶ Despite the well-described differences in risk of hearing loss between men and women, published studies have not reported whether the association between hearing loss and cognitive decline varies by sex. A recent American Geriatrics Society and National Institute on Aging Conference recommended that in future studies “data need to be stratified according to sex.”²⁷ Therefore, we investigated the relation of self-reported hearing loss and incident subsequent SCF decline in an ongoing cohort of 20,193 women in the Nurses’ Health Study (NHS).

2. Methods

2.1. Study Population

The NHS began in 1976 when 121,700 female registered nurses, aged 30–55 years, completed a mailed questionnaire about their health and lifestyle. These questionnaires (available at https://www.nurseshealthstudy.org/participants/questionnaires) have been administered every other year, and the 40-year follow-up exceeds 90% of eligible person-time. Most variables are updated every 2 years and diet information is updated every 4 years. We included women who provided information on their hearing on the 2012 questionnaire and also answered the questions on SCF on the 2012 and 2014 questionnaires (n=53,564). In our primary analysis evaluating the relation between self-reported hearing loss and incident SCF decline, we excluded women who reported 1 or more cognitive concerns in 2012 (n=18,334), and women who had cancer (other than non-melanoma skin cancer), Parkinson’s disease or a stroke (n=15,037), to minimize potential residual confounding due to varying severity of these conditions or treatment related effects. The number of women included in the primary analysis was 20,193.

2.2. Ascertainment of Hearing Loss

Self-reported hearing loss was determined based on the response on the 2012 long-form questionnaire that asked whether the participant had a hearing problem and, if so, the severity of the hearing problem (mild, moderate, or severe). In secondary analyses that evaluated the relation of recent onset of hearing loss or progression of mild hearing loss and risk of SCF decline, we used information on hearing status that was collected on the 2008 long-form questionnaire and examined change in self-reported hearing between 2008 and 2012. Questionnaire assessments of hearing loss in large populations have been found to be reasonably reliable in previous studies.^28–30 An Australian validation study of self-reported hearing loss demonstrated the sensitivity of a single question to assess hearing loss among older women was 79% for detecting mild hearing loss, defined as the better ear pure tone average hearing thresholds at 0.5, 1, 2, 4 kHz (BEPTA_{0.5,1,2,4 kHz}) >25 dB HL, 95% for detecting moderate hearing loss (BEPTA_{0.5,1, 2,4 kHz} >40 dB HL), and 100% for detecting severe hearing loss (BEPTA_{0.5,1,2,4 kHz} >60 dB HL), and the specificity was 72%, 65% and 64%, respectively.³⁰ Moreover, accumulating evidence shows auditory deterioration and hearing difficulty may not be fully captured by the conventional audiogram.^23,31 Self-reported functional hearing ability may provide an ecologically valid assessment of hearing in real-world settings and identify a larger group of adults with meaningful hearing impairment. Findings using self-reported hearing loss in this and similar cohorts have been published in several major medical journals. ^25,32–34

2.3. Ascertainment of Subjective Cognitive Function

Subjective cognitive function (SCF) was assessed on questionnaires administered in 2012 and 2014. The SCF score is based on 7 questions (yes/no): (1) Have you recently experienced any change in your ability to remember things? (2) Do you have more trouble than usual remembering recent events? (3) Do you have more trouble than usual remembering a short list of items, such as a shopping list? (4) Do you have trouble remembering things from one second to the next? (5) Do you have any difficulty in understanding or following spoken instructions? (6) Do you have more trouble than usual following a group conversation or a plot in a TV program due to your memory? (7) Do you have trouble finding your way around familiar streets? The SCF score ranged from 0 to 7, with one point assigned for each cognitive concern reported by the participant. Consistent with a previous study using this measure in men,²⁵ incident subjective cognitive function decline was defined as the new report of at least one SCF concern over follow-up. Thus, among those with SCF score = 0 (did not report any cognitive concerns) in 2012, incident subjective cognitive function decline was defined as an SCF score of 1 or greater (reported one or more cognitive concerns) in 2014.

These SCF questions have been validated against objective features of dementia and clinical cognitive measures.³⁵ In clinically normal individuals, subjective cognitive concerns have been associated with biomarkers of AD pathophysiology^{11,12,36–38} and higher rates of progression to mild cognitive impairment (MCI) and dementia, compared with those with no cognitive concerns.^39,40 Particularly in individuals with a higher level of education, subjective cognitive concerns are associated with higher risk for AD.⁴¹

2.4. Ascertainment of Covariates

We selected covariates that have been purported to be risk factors for hearing loss and/or for cognitive decline. Age (continuous), race, education (registered nurse, bachelor’s degree, masters or doctorate degree), socioeconomic status (represented by spouse’s level of education), body mass index, waist circumference, smoking (never, past, current), physical activity, hypertension, diabetes, hypercholesterolemia, tinnitus, Alternate Mediterranean diet (AMED) pattern score, aspirin use (days/week), nonsteroidal anti-inflammatory drug (NSAID) use (days/week), acetaminophen use (days/week), Mental Health Index (MHI-5) score on the Medical Outcomes Study 36-item Short-Form Survey (SF36) and depression (defined as the use of anti-depressants in the past 2 years or self-reported diagnosis). Dietary intakes and dietary pattern scores were calculated from semi-quantitative food frequency questionnaires that were administered every 4 years. Information on other covariates was available from the biennial questionnaires, including diagnoses of hypertension, diabetes mellitus, hypercholesterolemia, tinnitus, and use of analgesics. Questionnaire-derived information has been validated for many of the covariates by comparison with directly measured values or detailed diaries.^42–44

2.5. Statistical Analysis

The data were collected prospectively; we used information on hearing that was collected before the occurrence of SCF decline. Logistic regression models were used to compute the age- and multivariable-adjusted relative odds for incident SCF decline (score ≥1) according to severity of self-reported hearing loss: no hearing loss (referent group), mild, moderate, or severe hearing loss. We also evaluated the risk for more pronounced cognitive decline by computing the odds ratio for incident score ≥2 according to severity of self-reported hearing loss.

We performed several additional analyses. Of the seven questions included in the SCF score, two questions specified tasks that directly involved hearing. To evaluate the possibility that responses to these questions could be influenced by hearing ability, we performed sensitivity analyses that removed two questions: “Do you have difficulty in understanding things or following spoken instructions?” and “Do you have more trouble than usual following a group conversation or a plot in a TV program due to your memory?” In a separate analysis, we expanded our inclusion criteria to include women with either no or one cognitive concern at baseline (score ≤1) and examined risk of subjective cognitive decline, defined as worsening of SCF score ≥ 2 points. To evaluate the relation of recent onset of hearing loss or progression of mild hearing loss and risk of SCF decline, we used information on hearing status collected on the 2008 questionnaire and examined the association with incident SCF decline according to reported hearing in 2012 versus reported hearing in 2008. We a priori categorized recent onset of hearing loss as new onset of self-reported “mild” or worse (i.e. “moderate” or “severe”) hearing loss in 2012 for those who reported no hearing loss in 2008, and progression of mild hearing loss as the progression from self-reported “mild hearing loss” in 2008 to “moderate or worse hearing loss” in 2012; the referent group was “no hearing loss” in both 2008 and 2012.

To examine whether hearing aid use may influence the relation between hearing loss and subjective cognitive function, we used information on hearing loss and hearing aid use that was collected in 2006 (information on hearing aid use was not collected on the 2012 questionnaire). We included women who provided information on their hearing on the 2006 long-form questionnaire and also answered the questions on SCF on the 2012 and/or 2014 questionnaires (n=60,765), and excluded women with a history of cancer (other than non-melanoma skin cancer) or stroke (n=13,478). The number of women included in the analysis was 47,287. Consistent with previous studies using this measure in a similar cohort,⁴⁵ we computed the SCF score as an average of the two assessments with the possible range 0 to 7 points. If a participant responded to the SCF questions on only one of the questionnaires, we computed the SCF score using only 1 time point. The SCF score was used to categorize participants into one of 3 levels of SCF: “good” subjective cognitive function (SCF score = 0), “moderate” subjective cognitive function (SCF score = 0.5–2.5), or “poor” subjective cognitive function (SCF score = 3+). We used logistic regression models to compute the age- and multivariable-adjusted odds ratios for moderate versus good SCF and for poor versus good SCF according to severity of self-reported hearing loss (no hearing loss, mild, moderate, severe hearing loss with no hearing aid use, and severe hearing loss with hearing aid use). All P values are two-tailed and considered statistically significant at P<0.05. Statistical tests were performed with SAS statistical software, version 9.4 (SAS Institute Inc., Cary, North Carolina).

3. Results

The age-standardized characteristics of the study participants in 2012, according to self-reported hearing, are shown in Table 1. Women who reported more severe hearing loss were more likely to be older and to have tinnitus. Women who were excluded due to missing information on subjective cognitive function or hearing loss were slightly older, less physically active, and less likely to have depression than those included in the study and in the overall NHS population. Otherwise, the characteristics of the women included in the study population did not differ substantially from the overall cohort or those who were excluded due to missing information (Supplemental Table 1).

Table 1:

Age-standardized characteristics of women who completed subjective cognitive function questions in NHS I according to reported hearing status (2012)

Characteristic	No Hearing Loss	Mild Hearing Loss	Moderate Hearing Loss	Severe Hearing Loss
N	12692	5075	2000	426
Age, years*	73.3 (5.5)	75.0 (6.2)	77.7 (6.6)	78.7 (6.9)
White race,%	94.8	95.6	95.0	93.9
Body mass index, kg/m2	26.5 (5.4)	26.5 (5.3)	26.7 (5.5)	27.2 (5.8)
Waist circumference,cm†	84.9 (13.1)	85.6 (13.2)	86.3 (13.0)	87.6 (12.5)
Never smoker,%	49.1	49.5	46.0	45.9
Past smoker,%	46.1	46.2	50.1	50.7
Current smoker,%	4.8	4.2	3.8	3.4
Physical activity, METs‡	22.2 (26.6)	20.5 (24.1)	20.1 (23.5)	21.8 (26.8)
Hypertension,%	66.2	67.2	68.6	73.1
Diabetes,%	12.6	13.4	13.6	14.3
Hypercholesterolemia, %	73.1	75.3	76.4	76.9
Tinnitus¶,%	14.5	34.3	42.0	49.0
AMED§ score	4.2 (2.0)	4.2 (2.0)	4.1 (2.0)	4.1 (2.0)
Aspirin, regular use#,%	13.3	12.9	13.6	15.4
NSAID, regular use#,%	25.1	24.8	25.0	27.0
Acetaminophen, regular use#, %	27.2	30.4	31.3	35.4
Mental Health Index**	83.7 (11.5)	82.2 (12.2)	82.2 (12.1)	79.7 (15.2)
Depression,%	9.1	12.8	14.0	13.9
Education
RN††, %	58.7	59.9	58.8	57.4
Bachelor,%	20.0	20.2	22.6	20.5
Master,%	9.4	9.4	8.7	11.3
Doctorate,%	1.2	1.1	1.1	0.1
Spouse’s Education‡‡
<High School, %	0.9	0.9	0.9	0.5
Some High School, %	2.0	2.3	2.2	1.9
High School Graduate,%	27.9	29.6	28.6	32.0
College Graduate, %	23.7	23.7	24.1	24.2
Graduate School, %	21.4	20.1	21.0	17.1
Missing, %	24.1	23.3	23.3	24.3

Values are means(SD) or medians for continuous variables; percentages for categorical variables, and are standardized to the age distribution of the study population.

Values of polytomous variables may not sum to 100% due to rounding

All participants had an SCF score = 0 at baseline.

^*Value is not age adjusted

^†Waist circumference assessed in 2000

^‡Metabolic equivalents from recreational and leisure-time activities assessed in 2010

^¶Tinnitus defined as occurring several days per week or more

^§AMED: Alternate Mediterranean diet score

^#Regular use defined as 2 or more times per week; NSAID: nonsteroidal anti-inflammatory drug

^**Mental Health Index assessed in 2000

^††RN: Registered nursing degree

^‡‡Spouse’s level of education used as a proxy for socioeconomic status

In 2014, 3,991 participants newly reported 1 or more cognitive concerns (score ≥1) and 1,115 newly reported 2 or more concerns (score ≥2); the mean (SD) SCF score was 0.29 (0.72). For most of the specific cognitive concerns, the percentage of participants who endorsed the concern was greater among participants with self-reported hearing loss, compared with no hearing loss. Relatively few participants in any category of self-reported hearing loss endorsed the cognitive concern regarding “trouble finding your way on familiar streets.” The most commonly reported cognitive concerns in participants with self-reported hearing loss at baseline were “change in ability to remember things,” “trouble remembering short lists” and “trouble remembering one second to the next,” but these were also the most commonly reported cognitive concerns in participants with no hearing loss (Supplemental Table 2). Self-reported hearing loss of any severity was associated with higher risk of incident SCF decline (Table 2).

Table 2:

Self-reported Hearing loss (2012) and risk of incident subjective cognitive function decline^* in the Nurses’ Health Study

	SCF Score ≥ 1				SCF Score ≥ 2
Hearing Status		Cases	Age-adjusted OR (95% CI)	Multivariable-adjusted OR† (95% CI)		Cases	Age-adjusted OR (95% CI)	Multivariable-adjusted OR† (95% CI)
No hearing loss		2188	1.00 (ref)	1.00 (ref)		738	1.00 (ref)	1.00 (ref)
Mild hearing loss		1180	1.40 (1.30, 1.52)	1.35 (1.25, 1.47)		305	1.47 (1.29, 1.68)	1.39 (1.22, 1.59)
Moderate hearing loss		510	1.47 (1.31, 1.64)	1.39 (1.24, 1.56)		59	1.61 (1.35, 1.92)	1.48 (1.24, 1.78)
Severe hearing loss		113	1.53 (1.23, 1.90)	1.40 (1.21, 1.75)		13	1.40 (0.98, 1.99)	1.23 (0.86, 1.76)
p-trend			<0.001	<0.001			<0.001	<0.001

^*Incident subjective cognitive function decline defined as Subjective Cognitive Function Score (SCF) ≥ 1 or ≥ 2 at follow-up

^†Adjusted for: Age, race, education, socioeconomic status (spouse’s education), BMI, waist circumference, physical activity, smoking, hypertension, diabetes, hypercholesterolemia, depression, Mental Health Index (MHI-5), tinnitus, aspirin use, non-steroidal anti-inflammatory drug (NSAID) use, acetaminophen use, and AMED adherence score.

Abbreviations:

OR: odds ratio

BMI: body mass index

AMED: Alternate Mediterranean diet

Compared with women with no hearing loss, the multivariable-adjusted odds (MVOR,95%CI) for incident SCF decline (score ≥1) were 1.35 (1.25,1.47), 1.39 (1.24,1.56), and 1.40 (1.21,1.75) among women with mild, moderate and severe hearing loss, respectively. For incident SCF decline (score ≥2), the respective MVORs for mild and moderate hearing loss were 1.39 (1.22,1.59) and 1.48 (1.24,1.78); for severe hearing loss, there were only 13 incident SCF decline cases [MVOR 1.23 (0.86,1.76)]. In a sensitivity analysis that removed questions potentially related to hearing ability from the score, hearing loss remained associated with higher risk of SCF decline, but the magnitudes of the association were modestly attenuated (Table 3). In sensitivity analyses in which individuals with Parkinson’s, cancer or stroke were not excluded, but these conditions were adjusted for, the results did not appreciably differ (data not shown).

Table 3:

Self-reported Hearing loss (2012) and risk of incident subjective cognitive function decline^* in the Nurses’ Health Study I, after excluding two questions that potentially could be influenced by hearing status

	SCF Score ≥ 1		SCF Score ≥ 2
Hearing Status	Cases	Multivariable-adjusted OR† (95% CI)	Cases	Multivariable-adjusted OR† (95% CI)
No hearing loss	2146	1.00 (ref)	590	1.00 (ref)
Mild hearing loss	1124	1.31 (1.21, 1.42)	341	1.32 (1.14, 1.52)
Moderate hearing loss	466	1.28 (1.13, 1.44)	151	1.33 (1.09, 1.61)
Severe hearing loss	107	1.35 (1.08, 1.70)	33	1.24 (0.85, 1.81)
p-trend		<0.001		<0.001

^*Incident subjective cognitive function decline defined as Subjective Cognitive Function Score (SCF) ≥ 1. Score does not include questions on “difficulty in understanding or following spoken instructions” or “trouble following a group conversation or plot in a TV program.”

^†Adjusted for: Age, race, education, socioeconomic status (spouse’s education), BMI, waist circumference, physical activity, smoking, hypertension, diabetes, hypercholesterolemia, depression, Mental Health Index (MHI-5), tinnitus, aspirin use, non-steroidal anti-inflammatory drug use, acetaminophen use, and AMED adherence score.

Abbreviations:

OR: odds ratio

BMI: body mass index

AMED: Alternate Mediterranean diet

In an analysis among women who reported either no or one cognitive concern at baseline (score ≤1) that examined risk of subjective cognitive decline defined as worsening of SCF score ≥ 2 points, self-reported hearing loss was associated with higher risk of SCF decline, and the magnitude of the risk tended to be greater with increasing severity of hearing loss (Table 4). Compared with women with no hearing loss, the MVORs for cognitive decline were 1.37 (1.23,1.53) among women with mild hearing loss, 1.34 (1.15,1.55) among women with moderate hearing loss, and 1.50 (1.14,1.96) among women with severe hearing loss.

Table 4:

Self-reported Hearing loss (2012) and risk of subjective cognitive function decline^* in the Nurses’ Health Study, among women with SCF≤1 in 2012

		SCF Score Increase ≥ 2 Points
Hearing Status	Cases	Age-adjusted OR (95% CI)	Multivariable-adjusted OR† (95% CI)
No hearing loss	885	1.00 (ref)	1.00 (ref)
Mild hearing loss	619	1.44 (1.30, 1.61)	1.37 (1.23, 1.53)
Moderate hearing loss	277	1.43 (1.24, 1.66)	1.34 (1.15, 1.55)
Severe hearing loss	68	1.65 (1.27, 2.16)	1.50 (1.14, 1.96)
p-trend		<0.001	<0.001

^*Subjective cognitive function decline defined as worsening of Subjective Cognitive Function Score (SCF) ≥ 2 points.

^†Adjusted for: Age, race, education, socioeconomic status (spouse’s education), BMI, waist circumference, physical activity, smoking, hypertension, diabetes, hypercholesterolemia, depression, Mental Health Index (MHI-5), tinnitus, aspirin use, non-steroidal anti-inflammatory drug (NSAID) use, acetaminophen use, and AMED adherence score.

Abbreviations:

OR: odds ratio

BMI: body mass index

AMED: Alternate Mediterranean diet

Among women who reported no hearing loss or mild hearing loss in 2008, a 4-year worsening of hearing was associated with substantially higher risk of SCF decline (Table 5). Compared with women with no hearing loss in 2008 and 2012, the MVOR for incident SCF decline (score ≥1) among women with no hearing loss in 2008 that progressed to mild or worse hearing loss by 2012 was 1.28 (1.13,1.45); among women with mild hearing loss in 2008 that progressed to moderate or worse hearing loss by 2012, the MVOR for incident SCF decline was 1.55 (1.29,1.85). For more pronounced SCF decline (score ≥2), the magnitudes of the associations were even greater. Among women with no hearing loss that progressed to mild or worse hearing loss, the MVOR was 1.55 (1.28,1.88) and among women with mild hearing loss that progressed to moderate or worse hearing loss, the MVOR was 1.68 (1.28, 2.22).

Table 5:

Change in self-reported hearing (2008–2012) and risk of incident subjective cognitive function decline^* in the Nurses’ Health Study

	SCF Score ≥ 1			SCF Score ≥ 2
Hearing Status (2012)		Cases	Multivariable-adjusted OR† (95% CI)		Cases	Multivariable-adjusted OR† (95% CI)
No hearing loss		1899	1.00 (ref)		546	1.00 (ref)
New onset mild or worse hearing loss		399	1.28 (1.13, 1.45)		142	1.55 (1.28, 1.88)
Mild hearing loss progressed to moderate or worse hearing loss		185	1.55 (1.29, 1.85)		63	1.68 (1.28, 2.22)

^*New onset mild or worse hearing loss was defined as a new report of mild, moderate or severe hearing loss in 2012 for those who reported no hearing loss in 2008. Mild hearing loss that progressed to moderate or worse hearing loss was defined as the progression from self-reported mild hearing loss in 2008 to moderate or severe hearing loss in 2012. Incident subjective cognitive function decline defined as Subjective Cognitive Function Score (SCF) ≥ 1 or ≥ 2 at follow-up.

^†Adjusted for: Age, race, education, socioeconomic status (spouse’s education), BMI, waist circumference, physical activity, smoking, hypertension, diabetes, hypercholesterolemia, depression, Mental Health Index (MHI-5), tinnitus, aspirin use, non-steroidal anti-inflammatory drug (NSAID) use, acetaminophen use, and AMED adherence score.

Abbreviations:

OR: odds ratio

BMI: body mass index

AMED: Alternate Mediterranean diet

There were 47,287 women who provided information on their hearing and hearing aid use in 2006 and reported complete SCF data in 2012 or 2014. The average SCF score was 1.2 (SD 1.4). Among endorsed SCF concerns, a “change in ability to remember things” was the most common, reported by 38% and 36%, respectively, on the 2 questionnaires. Among the participants, 42% were categorized as having good SCF (0 points), 42% as moderate SCF (0.5–2.5 points), and 16% as having poor SCF (3+ points). Women who reported hearing loss in 2006 had substantially and independently higher odds of poor or moderate SCF compared with women who reported no hearing loss (Table 6). In the analysis that compared poor versus good SCF, the multivariable-adjusted odds ratio (MVOR, 95% CI) for poor SCF was 2.08 (1.95, 2.23) for women with mild hearing loss, 2.48 (2.20, 2.81) for women with moderate hearing loss, and 2.51 (1.96, 3.21) for women with severe hearing loss who did not use hearing aids. Among women with severe hearing loss who wore hearing aids, the MVOR for poor versus good SCF was 2.16 (1.90, 2.45). In the analysis that compared moderate versus good SCF, compared with women with no hearing loss, the MVOR (95% CI) was 1.57 (1.50, 1.65) for women with mild hearing loss, 1.66 (1.50, 1.83) for women with moderate hearing loss, and 1.53 (1.24, 1.88) for women with severe hearing loss who did not use hearing aids. Among women with severe hearing loss who wore hearing aids, the MVOR for moderate versus good SCF was 1.40 (1.26, 1.55).

Table 6:

Prevalence and adjusted odds ratios (95% confidence intervals) for subjective cognitive function according to 2006 hearing status

	No hearing loss	Mild hearing loss	Moderate hearing loss	Severe hearing loss, no hearing aid	Severe hearing loss, hearing aid
Good SCF (N,%)	14409 (48.0)	3855 (32.3)	678 (26.9)	157 (28.1)	648 (28.7)
Moderate SCF (N,%)	12084 (40.3)	5642 (47.3)	1179 (46.8)	250 (44.7)	998 (44.2)
Poor SCF (N,%)	3520 (11.7)	2440 (20.4)	661 (26.3)	152 (27.2)	614 (27.2)
Moderate versus good SCF
Age-adjusted OR (95% CI)	1.00 (ref)	1.64 (1.56, 1.72)	1.77 (1.60, 1.95)	1.70 (1.39, 2.08)	1.49 (1.34, 1.65)
MV-adjusted1 OR (95% CI)	1.00 (ref)	1.57 (1.50, 1.65)	1.66 (1.50, 1.83)	1.53 (1.24, 1.88)	1.40 (1.26, 1.55)
Poor versus good SCF
Age-adjusted OR (95% CI)	1.00 (ref)	2.26 (2.12, 2.41)	2.79 (2.48, 3.15)	3.01 (2.37, 3.81)	2.42 (2.14, 2.74)
MV-adjusted1 OR (95% CI)	1.00 (ref)	2.08 (1.95, 2.23)	2.48 (2.20, 2.81)	2.51 (1.96, 3.21)	2.16 (1.90, 2.45)

¹Covariates in the multivariable model were age, race, smoking history, education, socioeconomic status (spouse’s education), diabetes, hypertension, Mental Health Index (MHI-5), depression, hypercholesterolemia, physical activity level, body mass index, waist circumference, Alternate Mediterranean Diet score, tinnitus, aspirin use, non-steroidal anti-inflammatory drug (NSAID) use, and acetaminophen use.

MV-adjusted: multivariable-adjusted

OR: odds ratio

4. Discussion

In this large study among older women in the Nurses’ Health Study, we found that self-reported hearing loss was associated with higher risk of incident subjective cognitive decline, which may be an early marker of higher risk for dementia. There was a suggestion that the magnitude of the risk may be greater among those with more severe hearing loss. Recent progression of hearing loss was associated with an even higher risk. Early detection of hearing loss and evaluation of subjective cognitive function could help identify individuals at greater risk for cognitive decline and reveal a potential opportunity for intervention.

The subjective cognitive function score appraises cognitive concerns that reflect changes in memory or thinking experienced by individuals who may not meet the criteria for MCI or exhibit abnormal performance on cognitive tests.⁴⁶ Assessments of subjective cognitive function may capture an individual’s self-perceived experience of cognitive decline before evidence of cognitive impairment manifests on standardized neuropsychological tests and may provide the earliest noticeable indication of preclinical Alzheimer’s disease (AD) and other dementias.^4,47–50 Adults with subjective cognitive concerns have a higher risk of progression to MCI³⁵ and SCF score predicts dementia years before physician diagnosis.⁵¹ In addition, increasing numbers of subjective memory concerns were associated with strong linear trends of increasingly worse scores on cognitive tests, with approximately 20% higher odds of cognitive impairment with each additional cognitive concern endorsed.⁵² Our primary analyses focused on women with no cognitive concerns at baseline to evaluate the longitudinal association between self-reported hearing loss and incident SCF decline. We also evaluated the association with more pronounced cognitive decline to reduce the likelihood of misclassification of the outcome.

Identifying risk factors that may influence cognitive decline at its earliest stages is critical as interventions will most likely be more effective during an early prodromal phase before neurodegenerative or cognitive damage is irreversible.⁵³ Data on the relation between hearing loss and SCF decline are scarce. Our findings are consistent with our previous study among older men that found the risk of subjective cognitive decline was 30% higher among men with self-reported mild hearing loss, and 42–54% higher among men with moderate or worse hearing loss,²⁵ and with a smaller study that found self-reported hearing difficulty (queried as “even with a hearing aid”) was associated with 22% (95% CI: 14–29%) higher odds of poorer subjective cognitive function (assessed using 4 questions) 5 years later.⁵⁴ In a study in Japan (n=406), self-reported hearing loss was associated with higher prevalence of a subjective memory concern assessed with the single question, “Do you have problems with your memory?” Although no longitudinal association between self-reported hearing loss and an incident subjective cognitive concern at 1-year follow-up was observed, the follow-up was short and there were only 48 incident cases thus the power was limited.⁵⁵

Several potential mechanisms could underlie an association between hearing loss and cognitive decline, but the pathophysiology and whether the relation is causal remains unclear. In a neuroimaging study among clinically healthy adults, worse subjective cognitive function was associated with smaller entorhinal cortical and hippocampal volumes, brain regions involved in auditory processing and memory, and with accumulation of early tauopathy in the entorhinal cortex of the medial temporal lobe,⁵⁶ a region involved in the encoding and retrieval of auditory memory.^57,58 Hearing loss may lead to more effortful listening, increased cognitive load, and changes in neural networks.^59,60 To compensate for hearing loss, individuals may rely on cognitive resources that otherwise would be dedicated to task performance, which may contribute to poorer performance on cognitive tests.^61,62 Chronic compensation for hearing loss may lead to reallocation of cognitive resources and neural reorganization, and may deplete cognitive resources.^60,63 In addition, hearing loss can encumber social engagement and result in social isolation and depression, thereby increasing risk of cognitive decline.^64,65 It is likely that there are shared causative processes and sensory-related factors that cumulatively contribute to cognitive decline.^66,67 Common etiopathogenic mechanisms may underlie both auditory and cognitive changes that accrue with advancing age, including vascular, oxidative, and inflammatory processes, as well as genetics.^68–70 A cross-sectional study found that individuals with primary progressive aphasia had poorer pure-tone hearing sensitivities than healthy older individuals,⁷¹possibly suggesting shared neurodegenerative mechanisms.

We also found that progression of hearing loss over a 4-year period was associated with substantially higher risk of SCF decline. Possibly, detection of early hearing decline may help identify those at higher risk for cognitive decline and offer a valuable opportunity for targeted interventions. In a study of older adults at high risk for heart disease, intensive blood pressure control reduced the risk of developing MCI.⁷² Similarly, effective management of hearing loss, and prevention of progression, could possibly contribute to the prevention or delay of dementia. Notably, in a study in the Atherosclerosis Risk in Communities Neurocognitive Study, 20-year rates of decline in memory and global function were greater among participants with moderate/severe hearing loss, compared with participants with no hearing loss, and the rates of decline were greatest in participants who did not wear a hearing aid.⁷³ Further, heightened awareness of the implications of hearing loss on daily activities, work, social engagement, depression, and the added cognitive load and fatigue associated with difficulty hearing in challenging listening situations, could motivate improved surveillance and earlier detection of hearing loss. Further, management of hearing loss with lifestyle and environmental adaptations, hearing rehabilitation and hearing aids, could potentially be early stage interventions to prevent or slow the progression of cognitive decline.

In our cross-sectional examination among all women for whom we had information on hearing, hearing aid use, and SCF, we found that hearing loss was associated with substantially higher odds of reduced subjective cognitive function. The higher odds ratios appear to be attenuated by hearing aid use. Among individuals with hearing loss, hearing aid use may reduce listening effort and cognitive fatigue,⁷⁴ but findings from studies of hearing aid use and cognitive decline have been inconsistent.^73,75–78 Most have focused on older individuals and had important methodologic limitations; information on SCF is scarce. Greater listening effort and fatigue have been demonstrated even among those who wear hearing aids.⁷⁹ Our findings that suggest hearing aid use may attenuate the risk of subjective cognitive function decline among those with severe hearing loss are consistent with a large study in the UK Biobank⁷⁸ and a small study among older adults (mean age 77 years)⁷³ that found higher odds of poorer cross-sectional performance on neurocognitive tests among those with hearing loss who did not wear hearing aids, but differ from those from a study of older adults with hearing loss in the Epidemiology of Hearing Loss Study (n=666) that found no significant differences between hearing aid users and non-users for any of several cognitive measures.⁸⁰ More information, particularly on whether hearing aid use provides cognitive benefits for individuals with mild or moderate hearing loss, is needed.

While it was hypothesized that individuals who wore hearing aids would be less likely to have reduced SCF compared with those with severe hearing loss but who did not wear them, it was unexpected that the odds ratios among women who used hearing aids were lower than even those with moderate hearing loss. In our previous study among men, the magnitudes of the cross-sectional associations were similarly attenuated by hearing aid use. However, in our longitudinal analysis among men, even though the point estimates among men with severe hearing loss who used hearing aids were slightly lower than among men with hearing loss who did not use hearing aids, the difference was not statistically significant.²⁵ To our knowledge, the only other study of hearing and SCF examined hearing ability “even with the use of hearing aids.”⁵⁴ This requires further investigation.

We assessed SCF by methods used previously in this and similar cohorts.^25,45,81,82 SCF assessment has evolved over time, but the gold standard has not yet been established.⁸³ More recent approaches capture additional aspects regarding memory concerns and self-perception in relation to peers.^83–85 In our sensitivity analysis that excluded questions regarding cognitive concerns potentially influenced by hearing status from the SCF score, we found the magnitude of the relation between hearing loss and cognitive decline was modestly attenuated. Notably, a study that demonstrated that age-related differences in cognitive function varied as a function of hearing loss and the influence of age on cognitive decline may be overestimated if hearing loss is not taken into account⁸⁶ and suggests that some measures commonly used in cognitive research could be sensitive to hearing ability.⁸⁷ Given the high prevalence of hearing loss, hearing ability could be an important methodologic consideration in studies that evaluate cognition.

Potential limitations of this study include the self-report of hearing loss and of cognitive function. Self-reported hearing loss is reliable when compared with objective measures, but hearing decline is often insidious and individuals may not be aware of their hearing loss. This may also apply to self-assessment of cognitive function. Further longitudinal studies that examine relations of objective measures of hearing ability and SCF decline would be informative. Information on hearing was collected prospectively, before the occurrence of SCF decline, and our study population was limited to women who reported no cognitive concerns at baseline. Thus, misclassification of hearing status would likely be at random with respect to the outcome and would attenuate associations towards the null. Our study was predominantly white women with relatively high educational achievement. This is useful to reduce variability in the study measures and to enhance the internal validity of health-related information. Notably, SCF may be a stronger predictor of longitudinal cognitive decline in those with higher educational attainment.⁴¹ However, the findings may not be generalizable, thus replication of this study in other populations would be informative. This study was observational and residual confounding is possible, but the findings were robust to adjustments for numerous potential confounders, which were measured repeatedly using well-validated questionnaires. We examined SCF decline, which has been validated in previous studies and correlates with objective measures of memory loss;⁸⁸ however, our study did not examine objective outcome metrics. Although objective cognitive testing was performed in a subset of participants, the mode of administration was by telephone, which could influence test performance in those with hearing loss, thereby making the test results less reliable. Information on hearing status at the time of administration was not available, thus we would not be able to account for the potential influence of hearing status on performance. Although subjective cognitive function decline has been increasingly recognized as a possible early manifestation of AD, there are other potential explanations for subjective cognitive concerns in cognitively healthy individuals, including affective disorders, systemic illnesses, personality factors, and lifestyle factors.⁸⁴ We only had information on hearing aid use in 2006, thus we were unable to examine subsequent change in hearing aid use. Further study of the influence of hearing aid use on the relation between hearing loss and SCF decline is needed.

5. Conclusion

Self-reported hearing loss was associated with higher risk of incident subjective cognitive decline in women, which may be an early marker of higher risk for dementia. Hearing loss and cognitive decline are often slowly progressive and have both shared and independent mechanisms. Detection of early hearing decline may help identify those at higher risk for cognitive decline and offer a valuable opportunity for targeted interventions.

• Hearing loss is associated with higher risk of subjective cognitive function decline in women

• The risk may be greater with more severe hearing loss

• Recent progression of hearing loss was associated with even higher risk

• Hearing ability could be a valuable consideration in studies of cognition

Research in Context.

1. Systematic review: We reviewed the literature on hearing loss as a potential risk factor for accelerated cognitive decline. Longitudinal data were limited and few studies examined the relation of hearing loss and subjective cognitive function (SCF) decline. A substantially higher risk of SCF decline was reported in men with self-reported hearing loss; however, the relation had not been studied in women.

2. Interpretation: Our findings among women showed that self-reported hearing loss is associated with higher risk of incident SCF decline, and the magnitude of the risk may be greater among those with more severe hearing loss. Recent progression of hearing loss was associated with an even higher risk. Additionally, hearing ability could be a valuable consideration in studies of cognitive decline and dementia.

3. Future directions: Future studies need to address the following: (1) to investigate this relation in additional populations and (2) to examine whether management of hearing loss and prevention of progression could delay the onset of cognitive decline.

Abbreviations

AD

Alzheimer’s disease

AMED

Alternate Mediterranean diet

METs

Metabolic equivalents from recreational and leisure-time activities

MVOR

multivariable-adjusted odds ratio

NHS

Nurses’ Health Study

SCF

Subjective cognitive function