ORIGINAL CONTRIBUTION

Hearing Loss and Incident Dementia

Frank R. Lin, MD, PhD; E. Jeffrey Metter, MD; Richard J. O’Brien, MD, PhD;
Susan M. Resnick, PhD; Alan B. Zonderman, PhD; Luigi Ferrucci, MD, PhD

Objective: To determine whether hearing loss is asso- Main Outcome Measure: Incident caces of all-cause
ciated with incident all-cause dementia and Alzheimer dementia and AD until May 31, 2008.
disease (AD).
Results: During a median follow-up of 11.9 years, 58 cases
Design: Prospective study of 639 individuals who un- of incident all-cause dementia were diagnosed, of which
derwent audiometric testing and were dementia free in 37 cases were AD. The risk of incident all-cause dementia
1990 to 1994. Hearing loss was defined by a pure-tone increased log linearly with the severity of baseline hearing
average of hearing thresholds at 0.5, 1, 2, and 4 kHz in loss (1.27 per 10-dB loss; 95% confidence interval, 1.06-
the better-hearing ear (normal, ⬍25 dB [n = 455]; mild 1.50). Compared with normal hearing, the hazard ratio
loss, 25-40 dB [n=125]; moderate loss, 41-70 dB [n=53]; (95% confidence interval) for incident all-cause dementia
and severe loss, ⬎70 dB [n = 6]). Diagnosis of incident was 1.89 (1.00-3.58) for mild hearing loss, 3.00 (1.43-
dementia was made by consensus diagnostic confer- 6.30) for moderate hearing loss, and 4.94 (1.09-22.40) for
ence. Cox proportional hazards models were used to severe hearing loss. The risk of incident AD also increased
model time to incident dementia according to severity with baseline hearing loss (1.20 per 10 dB of hearing loss)
but with a wider confidence interval (0.94-1.53).
of hearing loss and were adjusted for age, sex, race, edu-
cation, diabetes mellitus, smoking, and hypertension.
Conclusions: Hearing loss is independently associated
with incident all-cause dementia. Whether hearing loss
Setting: Baltimore Longitudinal Study of Aging. is a marker for early-stage dementia or is actually a modi-
fiable risk factor for dementia deserves further study.
Participants: Six hundred thirty-nine individuals aged
36 to 90 years. Arch Neurol. 2011;68(2):214-220

T
HE PREVALENCE OF DEMEN- pertension.4 Some researchers have also
tia is projected to double suggested that hearing loss, by reducing
every 20 years such that by stimulatory input and hampering social in-
2050, more than 100 mil- teraction, may be associated with demen-
lion people or nearly 1 in 85 tia,5,6 but, to our knowledge, this hypoth-
Author Affiliations: persons will be affected worldwide.1,2 The esis has never been prospectively studied.
Department of devastating impact of dementia on af- Given the growing number of people with
Otolaryngology–Head and Neck fected individuals and the burden im- hearing loss7 and the array of technologi-
Surgery, The Johns Hopkins posed on their families and society has cal interventions currently available for au-
School of Medicine (Dr Lin),
made the prevention and treatment of de- ral rehabilitation, understanding whether
Center on Aging and Health,
Johns Hopkins Medical mentia a public health priority. Interven- hearing loss is a risk factor for dementia
Institutions (Dr Lin), tions that could merely delay the onset of is important. We performed the present
Longitudinal Studies Section, dementia by 1 year would lead to a more study to investigate the prospective asso-
Clinical Research Branch, than 10% decrease in the global preva- ciation of hearing loss with incident de-
National Institute on Aging lence of dementia in 2050. 3 Unfortu- mentia within the cohort of the Balti-
(Drs Metter and Ferrucci), and nately, there are no known interventions more Longitudinal Study of Aging (BLSA).
Departments of Neurology and that currently have such effectiveness.
Medicine, Johns Hopkins Epidemiologic approaches have fo-
Bayview Medical Center METHODS
cused on the identification of putative risk
(Dr O’Brien), Baltimore,
factors that could be targeted for preven- SUBJECTS
Maryland; and Laboratory of
Behavioral Neuroscience, tion based on the assumption that demen-
Intramural Research Program, tia is easier to prevent than to reverse. Can- Subjects were participants in the BLSA, an on-
National Institute on Aging, didate factors include low involvement in going prospective study of the effects of aging
Bethesda, Maryland leisure activities and social interactions, that was initiated in 1958 by the National In-
(Drs Resnick and Zonderman). sedentary state, diabetes mellitus, and hy- stitute on Aging.8 The BLSA cohort consists of

(REPRINTED) ARCH NEUROL / VOL 68 (NO. 2), FEB 2011 WWW.ARCHNEUROL.COM

214
Downloaded from www.archneurol.com at Emory University, on February 17, 2011
©2011 American Medical Association. All rights reserved.
community-dwelling volunteers who travel to the National In-
stitute on Aging in Baltimore biennially for 21⁄2 days of inten-
1305 BLSA participants with ≥1 study visit,
sive testing. From 1990 through 1994, 1305 participants com- 1990 to 1994
pleted at least 1 study visit, of whom 976 underwent audiometry
and 749 had both audiometry and cognitive testing. Some par- 976 BLSA participants with audiometry
ticipants had missing audiometry or cognitive testing data be- testing, 1990 to 1994
cause of inadequate time for testing or tester unavailability dur-
ing study visits. After excluding individuals with prevalent
dementia (n=58), those with more than 3 errors on the Blessed 749 BLSA participants with audiometry and
cognitive testing, 1990 to 1994
Information Memory Concentration Test (n=39), and those with
suspected dementia (n=13), our baseline cohort consisted of
110 Participants excluded
639 participants who were followed up until May 31, 2008 (me- 58 With prevalent dementia
dian participant follow-up of 11.9 years) (Figure 1). For par- 39 With >3 errors on the Blessed Test
ticipants with more than 1 visit during this period, data from 13 With suspected dementia
the first assessment were used. All participants provided writ-
ten informed consent, and the BLSA study protocol was ap- 639 Participants in baseline study cohort 1990 to 1994
proved by the institutional review board.
58 Participants diagnosed as having incident all-cause
COGNITIVE TESTING AND dementia, including 37 with AD, 1991 to 2008

DIAGNOSIS OF DEMENTIA
Figure 1. Selection of participants for study inclusion. AD indicates
The protocol for adjudication of dementia in the BLSA has been Alzheimer disease; Blessed Test, Blessed Information Memory Concentration
used continuously since 1986 and has been described previ- Test; and BLSA, Baltimore Longitudinal Study of Aging.
ously.9 Participants 65 years or older underwent a complete neu-
rological and neuropsychological examination using a stan- OTHER COVARIATES
dard battery of tests. Participants younger than 65 years first
underwent screening with the Blessed Information Memory Con- A diagnosis of diabetes mellitus was based on a fasting glucose
centration Test and underwent further examination if they made level of more than 125 mg/dL (to convert to millimoles per liter,
3 or more errors. Dementia diagnosis was established during a multiply by 0.0555), a pathologic oral glucose tolerance test re-
multidisciplinary consensus diagnostic conference using the Di- sult, or history of a physician diagnosis plus treatment with oral
agnostic and Statistical Manual of Mental Disorders (Third Edi- antidiabetic drugs or insulin. The diagnosis of hypertension was
tion Revised) for diagnosis of dementia10 and the National In- based on a systolic blood pressure of greater than 140 mm Hg
stitute of Neurological and Communicative Disorders and and/or diastolic blood pressure of at least 90 mm Hg or treat-
Stroke–Alzheimer Disease and Related Disorders Association ment with antihypertensive medications. Race (white/black/
criteria for diagnosis of Alzheimer disease (AD).11 If partici- other), education (in years), smoking status (current/former/
pants were determined to have clinically significant cognitive never), and hearing aid use were based on self-report.
decline (typically memory) but did not meet criteria for de-
mentia, they were classified as having suspected dementia, which
corresponds to the current diagnosis of mild cognitive impair-
STATISTICAL ANALYSES
ment.12 Participants initially underwent evaluation for demen-
Baseline characteristics of cohort members were compared using
tia every 2 years during their routine BLSA follow-up visits. In
1-way analysis of variance for continuous variables and ␹2 or
1997, follow-up was shifted to a sliding-scale schedule to re-
Fisher exact test for categorical variables. Cox proportional haz-
duce participant burden and improve data collection. Partici-
ards models were used to study time to incident all-cause de-
pants older than 80 years were examined annually; those aged
mentia or AD. Participants not diagnosed as having dementia
60 to 80 years, biennially; and those younger than 60 years,
were censored at the time of their last negative cognitive evalu-
every 4 years.
ation finding. Time-on-study (ie, time of entry into the base-
line study cohort) was used as the time scale with the excep-
AUDIOMETRY tion of 1 model that used age as the time scale.
All Cox models included covariates of sex, age, race, edu-
Audiometry was performed in the BLSA from 1958 to 1994. cation, diabetes, smoking, and hypertension. Diabetes and hy-
During the entire period from 1990 through 1994, when the pertension were included as covariates in the analysis because
baseline evaluation for this analysis was performed, hearing they have been found to be risk factors for dementia.4 Addi-
thresholds were measured using an automated testing device tional models included baseline Blessed scores (residual vari-
(Audiometer Model 320; Virtual Equipment Co, Portland, Or- ability in cognition after definition of the baseline cohort) and
egon) in a soundproof chamber under unaided conditions. A hearing aid use. All covariates were treated as time-constant
pure-tone average (PTA) of air conduction thresholds at 0.5, variables. Cox model proportionality assumptions and the lin-
1, 2, and 4 kHz was calculated for each ear, and the PTA in the ear association between hearing loss and dementia were tested
better-hearing ear was used for subsequent analyses because using the Schoenfeld residuals method.13 To examine the graphi-
that ear would be the principal determinant of hearing and cal association between hearing threshold and dementia, we used
speech perception ability on an everyday basis. We used the a smoothing spline for the hearing threshold and age in the Cox
PTA in decibels as both a continuous variable and a categori- proportional hazards model.14 A locally weighted scatterplot
cal variable defined by the following commonly used levels of smoother (loess smoother) was then applied to the exponen-
hearing loss: normal (⬍25 dB), mild loss (25-40 dB), moder- tial of the partial residuals derived from the hazards model against
ate loss (41-70 dB), and severe loss (⬎70 dB). Before 1990, au- the hearing threshold. A bootstrap procedure was used to gen-
diometric testing was performed using a Bekesy audiometer (GSI erate 10 000 data sets that were then used to estimate the 95%
1701; Grason Stadler, Littleton, Massachusetts), and these data confidence interval (CI) for the loess smoother. Analysis of hear-
were used in analyses of prebaseline hearing trajectories. ing loss trajectories before baseline was performed using a ran-

(REPRINTED) ARCH NEUROL / VOL 68 (NO. 2), FEB 2011 WWW.ARCHNEUROL.COM

215
Downloaded from www.archneurol.com at Emory University, on February 17, 2011
©2011 American Medical Association. All rights reserved.
 Table 1. Demographic and Clinical Characteristics of Baseline Study Cohort by Hearing Loss Status

Hearing Loss Status a

Normal Mild Moderate Severe

(n=455) (n = 125) (n = 53) (n = 6) P Value
Male sex 225 (49.5) 94 (75.2) 36 (67.9) 5 (83.3) ⬍.001
Age, mean (SD), y 59.9 (12.2) 71.1 (8.6) 77.0 (8.4) 77.7 (4.8) ⬍.001
Race
White 404 (88.8) 121 (96.8) 49 (92.5) 6 (100.0)
Black 44 (9.7) 4 (3.2) 4 (7.5) 0 .17
Other 7 (1.5) 0 0 0
Education, mean (SD), y 16.6 (2.8) 16.2 (3.0) 16.7 (3.6) 16.2 (4.0) .74
Diabetes mellitus 62 (13.6) 20 (16.0) 12 (22.6) 1 (16.7) .27
Smoking
Current 19 (4.2) 1 (0.8) 1 (1.9) 0
Former 244 (53.6) 85 (68.0) 35 (66.0) 3 (50.0) .05
Never 192 (42.2) 39 (31.2) 17 (32.1) 3 (50.0)
Hypertension 204 (44.8) 79 (63.2) 38 (71.7) 6 (100.0) ⬍.001
Hearing aid use b 6 (1.5) 14 (11.9) 39 (78.0) 4 (66.7) ⬍.001
Blessed Information Memory
Concentration Test score
0 265 (58.2) 65 (52.0) 31 (58.5) 0
1 112 (24.6) 32 (25.6) 13 (24.5) 2 (33.3)
.08
2 49 (10.8) 19 (15.2) 6 (11.3) 3 (50.0)
3 29 (6.4) 9 (7.2) 3 (5.7) 1 (16.7)
Development of all-cause dementia 20 (4.4) 21 (16.8) 15 (28.3) 2 (33.3) ⬍.001
during follow-up

a Hearing loss is defined by the pure-tone average (PTA) of 0.5, 1, 2, and 4 kHz, with tones presented by air conduction in the better-hearing ear. A PTA of less
than 25 dB indicates normal hearing; 25 to 40 dB, mild loss; 41 to 70 dB, moderate loss; and greater than 70 dB, severe loss. Unless otherwise indicated, data are
expressed as number (percentage) of participants. Percentages have been rounded and might not total 100.
b Data on hearing aid use were missing for 72 individuals. Participants with hearing aid use data per hearing loss category included 393 with normal hearing,
118 with mild loss, 50 with moderate loss, and 6 with severe loss.

dom effects analysis and adjusted for age. Population- Independent of age, in the 15 years before baseline
attributable risk (PAR) was calculated using the following assessment (520 participants with 2678 observations),
equation15: participants who later developed incident dementia ex-
PAR=(Pexposed([R–1])/(1⫹Pexposed[RR–1]), perienced an average PTA loss of 0.52 dB/y (95% CI, 0.34-
0.70 dB/y) compared with 0.27 dB/y (0.21-0.33 dB/y) in
where Pexposed was the prevalence of baseline hearing loss of at those who did not develop dementia.
least 25 dB and RR was the rate ratio (hazard ratio [HR]) of In Cox proportional hazards models adjusted for sex,
dementia risk associated with hearing loss. Participants with age, race, education, diabetes, smoking, and hyperten-
missing data were excluded from analyses; this represented sion (base model), the excess risk of incident dementia per
less than 0.2% of the study sample (1 participant) for all
10 dB of hearing loss was 1.27 (95% CI, 1.06-1.50)
analyses except for analyses incorporating hearing aid use,
in which there were more extensive missing data (typically (Table 3). The risk of incident dementia became evi-
among normal-hearing participants who did not respond). dent for hearing loss of greater than 25 dB and thereaf-
Significance testing for all analyses was 2 sided with a type I ter increased log linearly with more severe loss
error of .05. The statistical software used was a free available (Figure 2). This association remained significant after
software environment (R, version 2.9.1; http://www.r-project censoring participants who developed dementia within
.org). a 2-, 4-, or 6-year washout period from baseline (P=.008,
P =.003, and P=.04, respectively).
RESULTS Confirmatory analyses from models including base-
line Blessed error score (to account for baseline cogni-
Baseline demographic characteristics of participants by tive function) or models using age as the time scale rather
hearing loss category are presented in Table 1. In gen- than time-on-study (to account for residual confound-
eral, participants with greater hearing loss were more likely ing between age and hearing loss) produced virtually un-
to be older, male, and hypertensive. Blessed scores did changed findings (cf Table 3). Restricting the analytical
not differ by hearing loss category (P=.08), although the cohort to participants 65 years or older at baseline
range of errors was narrow (0-3) because participants with (n=315) or excluding participants at baseline with a his-
more than 3 errors were excluded from the study cohort tory of stroke or transient ischemic attack (n=19) also
at baseline. did not substantially change the main findings (Table 3).
Baseline covariates associated with an increased risk There was no evidence to suggest that self-reported hear-
of incident all-cause dementia are hearing loss, age, hy- ing aid use was associated with a reduction in dementia
pertension, hearing aid use, and Blessed score (Table 2). risk (HR, 0.97; P =.92).

(REPRINTED) ARCH NEUROL / VOL 68 (NO. 2), FEB 2011 WWW.ARCHNEUROL.COM

216
Downloaded from www.archneurol.com at Emory University, on February 17, 2011
©2011 American Medical Association. All rights reserved.
 Table 2. Demographic and Clinical Characteristics Table 3. Cox Proportional Hazards Models for Incident
of Baseline Study Cohort by Incident Dementia a All-Cause Dementia per 10 dB of Hearing Loss

No No. of P
Dementia Dementia Univariate Model Participants HR (95% CI) a Value
(n=581) (n=58) HR (95% CI)
Base b 638 1.27 (1.06-1.50) .008
Hearing loss, mean (SD), 18.8 (13.9) 32.6 (17.0) 1.1 (1.0-1.1) c Base and Blessed 638 1.24 (1.04-1.48) .01
PTA b Information Memory
Hearing loss d Concentration Test score
Normal 435 (74.9) 20 (34.5) 1 [Reference] Base with age as time scale 638 1.29 (1.08-1.53) .005
Mild 104 (17.9) 21 (36.2) 4.9 (2.6-8.8) Base and hearing aid use 566 1.33 (1.07-1.64) .008
Moderate 38 (6.5) 15 (25.9) 12.1 (6.2-23.9)
Severe 4 (0.7) 2 (3.4) 21.9 (5.1-94.2) Abbreviations: CI, confidence interval; HR, hazard ratio.
Male sex 327 (56.3) 33 (56.9) 1.1 (0.6-1.8) a Hearing loss is defined by the pure-tone average of hearing thresholds at
Age, mean (SD), y 62.2 (12.3) 78.3 (6.4) 1.2 (1.2-1.2) 0.5, 1, 2, and 4 kHz, with tones presented by air conduction in the
Race better-hearing ear.
b Base model covariates include sex, age, race, education, diabetes
White 523 (90.0) 57 (98.3) 1 [Reference]
mellitus, smoking, and hypertension.
Black 51 (8.8) 1 (1.7) 0.2 (0.02-1.2)
Other 7 (1.2) 0
Education, mean (SD), y 16.5 (3.0) 16.6 (3.0) 1.0 (0.9-1.1)
Diabetes mellitus 84 (14.5) 11 (19.0) 1.6 (0.9-3.0) 12
Smoking
Current 20 (3.4) 1 (1.7) 1 [Reference]
Former 333 (57.3) 34 (58.6) 2.3 (0.3-17.0) 10
Never 228 (39.2) 23 (39.7) 2.1 (0.3-15.3)
Hypertension 286 (49.2) 41 (70.7) 2.9 (1.7-5.2)
Hearing aid use e 47 (9.2) 16 (30.2) 5.3 (2.9-9.6) 8
Blessed Information
Memory Concentration
Hazard
Test score 6
0 338 (58.2) 23 (39.7) 1 [Reference]
1 140 (24.1) 19 (32.8) 2.1 (1.2-3.9)
2 64 (11.0) 13 (22.4) 2.8 (1.4-5.6) 4
3 39 (6.7) 3 (5.2) 1.2 (0.4-4.1)

2
Abbreviations: CI, confidence interval; HR, hazard ratio; PTA, pure-tone
average.
a Unless otherwise indicated, data are expressed as number (percentage)
of participants. Percentages have been rounded and might not total 100. 0
b Hearing loss is defined by the PTA of hearing thresholds at 0.5, 1, 2, and
0 20 40 60 80 100
4 kHz, with tones presented by air conduction to the better-hearing ear. Hearing Loss, dB
c Indicates hazard per 1 dB of PTA.
d A PTA of less than 25 dB indicates normal hearing; 25 to 40 dB, mild
loss; 41 to 70 dB, moderate loss; and greater than 70 dB, severe loss. Figure 2. Risk of incident all-cause dementia by baseline hearing loss after
e Data on hearing aid use were missing for 72 individuals. Participants with
adjustment for age, sex, race, education, diabetes mellitus, smoking, and
hearing aid use data include 514 with no dementia and 53 with dementia. hypertension. Hearing loss is defined by the pure-tone average of thresholds
at 0.5, 1, 2, and 4 kHz in the better-hearing ear. Upper and lower dashed
lines correspond to the 95% confidence interval.
In subsequent analyses, we categorized hearing loss
according to commonly accepted levels of hearing loss
tive risk (HR) of dementia associated with hearing loss
severity. Compared with those with normal hearing, par-
was 2.32 (95% CI, 1.32- 4.07). Thus, the attributable risk
ticipants with mild hearing loss had an HR for incident
of dementia associated with hearing loss in this subco-
dementia of 1.89 (95% CI, 1.00-3.58; P=.049), those with
hort was 36.4% (95% CI, 12.8%-58.6%).
moderate hearing loss had an HR of 3.00 (1.43-6.30;
P=.004), and those with severe hearing loss (n=6) had
an HR of 4.94 (1.09-22.40; P = .04). COMMENT
When the outcome of the analysis was restricted to in-
cident AD (37 of the 58 cases of dementia), hearing loss In this study, hearing loss was independently associated
was associated with an excess risk of 1.20 per 10 dB of hear- with incident all-cause dementia after adjustment for sex,
ing loss (95% CI, 0.94-1.53). This result is comparable to age, race, education, diabetes, smoking, and hyperten-
the risk seen for all-cause dementia (Table 3) but with a sion, and our findings were robust to multiple sensitiv-
wider CI, possibly owing to the smaller sample size. ity analyses. The risk of all-cause dementia increased log
We estimated the proportion of incident all-cause de- linearly with hearing loss severity, and for individuals older
mentia risk that was attributable to hearing loss for par- than 60 years in our cohort, more than one-third of the
ticipants older than 60 years in our cohort, assuming that risk of incident all-cause dementia was associated with
hearing loss could be causally associated with demen- hearing loss.
tia. Hearing loss of at least 25 dB in the better-hearing Our findings contribute significantly to the discussion
ear was present in 43% of this subcohort, and the rela- in the literature on whether hearing loss is a risk factor for

(REPRINTED) ARCH NEUROL / VOL 68 (NO. 2), FEB 2011 WWW.ARCHNEUROL.COM

217
Downloaded from www.archneurol.com at Emory University, on February 17, 2011
©2011 American Medical Association. All rights reserved.
dementia. Previous studies suggested that individuals with dementia also cannot be fully excluded. However, risk
hearing loss are more likely to have a diagnosis of demen- factors for vascular disease such as diabetes, smoking, and
tia5,6 and poorer cognitive function.16 Supporting this hy- hypertension were adjusted for in our models, and a pre-
pothesis, smaller prospective studies have observed that liminary study has not found a positive association be-
hearing loss is associated with accelerated cognitive de- tween ApoE status and hearing loss.28 Other variables,
cline in individuals with prevalent dementia.17,18 Al- such as mental and leisure activities, were not included
though a prospective study19 of cognitively normal el- as covariates in our models because these variables would
derly volunteers failed to find any meaningful association not be expected to cause hearing loss and act as mean-
between hearing loss at study entry and later cognitive func- ingful confounders in our models. Our results were also
tion, the results of that study are questionable because of robust to excluding individuals at baseline who had a his-
the short (5-year) follow-up and a 50% dropout rate. In our tory of stroke or transient ischemic attack.
study, hearing loss, a condition that is highly prevalent in Finally, hearing loss may be causally related to de-
older adults and that often remains untreated,20 was strongly mentia, possibly through exhaustion of cognitive re-
and prospectively associated with incident dementia. serve, social isolation, environmental deafferentation, or
A number of mechanisms may be theoretically impli- a combination of these pathways. Cognitive reserve re-
cated in the observed association between hearing loss flects interindividual differences in neurocognitive pro-
and incident dementia. There may be an overdiagnosis cessing that allow some individuals to cope better with
of dementia in individuals affected by hearing loss or, vice neuropathology than others.29 Functional magnetic reso-
versa, an overdiagnosis of hearing loss in individuals with nance imaging studies showing interindividual varia-
cognitive impairment at baseline. An overdiagnosis of de- tion in efficiency of task-related neural processing pro-
mentia in our study is unlikely because the diagnostic vide some evidence of this concept.30,31 Cognitive reserve
protocol for incident dementia relied on a consensus con- has also been used to explain discrepancies between the
ference that examined information from multiple sources. extent of neuropathology seen at autopsy and clinical ex-
We also conducted sensitivity analyses censoring indi- pression of dementia.32 The potential effect of hearing loss
viduals diagnosed as having dementia during a 6-year on cognitive reserve is suggested by studies demonstrat-
washout period from baseline that did not affect our re- ing that, under conditions in which auditory perception
sults. In such an analysis, individuals would already have is difficult (ie, hearing loss), greater cognitive resources
had normal findings on several cognitive examinations are dedicated to auditory perceptual processing to the det-
with hearing loss before being diagnosed as having de- riment of other cognitive processes such as working
mentia, likely indicating that the dementia diagnosis was memory.33,34 This reallocation of neural resources to au-
not confounded by poor communication. Hearing loss ditory processing could deplete the cognitive reserve avail-
(short of profound deafness) also minimally impairs face- able to other cognitive processes and possibly lead to the
to-face communication in quiet environments (ie, dur- earlier clinical expression of dementia.35
ing cognitive testing), particularly in the setting of test- Communication impairments caused by hearing loss
ing by experienced examiners who are accustomed to can also lead to social isolation in older adults,36,37 and epi-
working with older adults.21 demiologic38,39 and neuroanatomic studies40 have demon-
An overdiagnosis of hearing loss is also unlikely be- strated associations between poor social networks and de-
cause no evidence suggests that mild cognitive impair- mentia. Our results also seem to support this possible
ment would affect the reliability of audiometric testing. pathway because the risk of dementia associated with hear-
Pure-tone audiometry has been performed in children as ing loss appeared to only increase at hearing thresholds
young as 5 years. We also excluded any individuals with of greater than 25 dB, which is considered the threshold
recognized cognitive impairment at baseline (mild cogni- at which hearing loss begins to impair verbal communi-
tive impairment or Blessed score ⬎3), and our results were cation.41 Finally, a hypothetical mechanism by which hear-
robust to models controlling for baseline Blessed scores. ing loss could directly affect AD neuropathology is sug-
Another possibility is that hearing loss and progres- gested by animal studies demonstrating that environmental
sive cognitive impairment are caused by a common neu- enrichment (possibly analogous in humans to having ac-
ropathologic process, possibly the same that leads to AD. cess to auditory and environmental stimuli) can reduce
However, pure-tone audiometry is typically considered ␤-amyloid levels in transgenic mouse models.42 This hy-
a measure of the auditory periphery because detection pothesis is also supported by studies showing that indi-
of pure tones relies solely on cochlear transduction and viduals who remain engaged in leisure activities have a
neuronal afferents to brainstem nuclei and the primary lower risk of dementia.43
auditory cortex. Perception of pure tones does not re- In the present study, self-reported hearing aid use was
quire higher levels of auditory cortical processing,22 and not associated with a significant reduction in dementia risk,
results of auditory brainstem response testing of these but data on other key variables (eg, type of hearing aid used,
pathways are usually normal in patients with AD.23 In con- hours worn per day, number of years used, characteris-
trast, central auditory nuclei required for higher-order tics of participants choosing to use hearing aids, use of other
auditory processing can be affected by AD neuropathol- communicative strategies, and adequacy of rehabilita-
ogy,24-26 and tests of central auditory function have been tion) that would affect the success of aural rehabilitation
found to be associated with AD.27 and affect any observed association were not gathered. Con-
The likelihood of another neurobiological process such sequently, whether hearing devices and aural rehabilita-
as vascular disease or factors related to family history (eg, tive strategies could affect cognitive decline and demen-
apolipoprotein E [ApoE] status) causing hearing loss and tia remains unknown and will require further study.

(REPRINTED) ARCH NEUROL / VOL 68 (NO. 2), FEB 2011 WWW.ARCHNEUROL.COM

218
Downloaded from www.archneurol.com at Emory University, on February 17, 2011
©2011 American Medical Association. All rights reserved.
 Our study has limitations. First, only the severity of hear- Funding/Support: This work was supported by the In-
ing loss at baseline was considered in the analysis, and in- tramural Research Program of the National Institute on
formation was not available on the trajectory of hearing loss Aging and grant 1K23DC011279-01 from the National
after baseline assessment or on the possible etiology of the Institute on Deafness and Other Communication Disor-
hearing loss. However, it is unlikely that this limitation sub- ders (Dr Lin).
stantially biased our findings given that reversible hearing Role of the Sponsor: The National Institute on Aging
loss is rare, and hearing loss tends to only worsen with time. funded the design and conduct of the study; collection,
Residual confounding by other environmental, genetic, or management, analysis, and interpretation of the data; and
neuropathologic processes is also plausible but specula- preparation and review of the manuscript. The sponsor
tive based on our current knowledge of established risk fac- was not involved in manuscript approval.
tors for hearing loss and dementia. Given the very close
association between age and both hearing loss and demen-
REFERENCES
tia, there is a possibility of unaccounted residual confound-
ing. However, this is unlikely because we also confirmed
1. Ferri CP, Prince M, Brayne C, et al; Alzheimer’s Disease International. Global preva-
our findings in a statistical model using age rather than time- lence of dementia: a Delphi consensus study. Lancet. 2005;366(9503):2112-
on-study as the time scale to account for nonlinear effects 2117.
of age on hearing and cognition.44 Our findings were also 2. Alzheimer’s Disease International. World Alzheimer Report 2009. In: Prince M,
unchanged after restricting our cohort to participants 65 Jackson J, eds. London, England: Alzheimer’s Disease International; 2009.
3. Brookmeyer R, Johnson E, Ziegler-Graham K, Arrighi HM. Forecasting the global
years or older at baseline. burden of Alzheimer’s disease. Alzheimers Dement. 2007;3(3):186-191.
Finally, caution must be applied when generalizing 4. Coley N, Andrieu S, Gardette V, et al. Dementia prevention: methodological ex-
the results of our current study because the BLSA is a vol- planations for inconsistent results. Epidemiol Rev. 2008;30:35-66.
unteer cohort of individuals of high socioceconomic sta- 5. Uhlmann RF, Larson EB, Rees TS, Koepsell TD, Duckert LG. Relationship of hear-
tus. Further confirmation of our results will need to be ing impairment to dementia and cognitive dysfunction in older adults. JAMA. 1989;
261(13):1916-1919.
performed in larger studies using more representative, 6. Ives DG, Bonino P, Traven ND, Kuller LH. Characteristics and comorbidities of rural
community-based samples. However, this potential limi- older adults with hearing impairment. J Am Geriatr Soc. 1995;43(7):803-806.
tation to broad generalizability could strengthen the in- 7. Agrawal Y, Platz EA, Niparko JK. Prevalence of hearing loss and differences by
ternal validity of our findings given the relative homo- demographic characteristics among US adults: data from the National Health and
Nutrition Examination Survey, 1999-2004. Arch Intern Med. 2008;168(14):
geneity of the study cohort in observed and likely 1522-1530.
unobservable characteristics. 8. Shock N, Greulich R, Andres R, et al. Normal Human Aging: The Baltimore Lon-
If confirmed in other independent cohorts, the find- gitudinal Study of Aging . Washington, DC: National Institutes of Health; 1984.
ings of our study could have substantial implications for NIH publication 84-2450.
individuals and public health. Hearing loss in older adults 9. Kawas C, Gray S, Brookmeyer R, Fozard J, Zonderman A. Age-specific inci-
dence rates of Alzheimer’s disease: the Baltimore Longitudinal Study of Aging.
may be preventable45 and can be practically addressed with Neurology. 2000;54(11):2072-2077.
current technology (eg, digital hearing aids and coch- 10. American Psychiatric Association. Diagnostic and Statistical Manual of Mental
lear implants) and with other rehabilitative interven- Disorders, Third Edition, Revised. Washington, DC: American Psychiatric Asso-
tions focused on optimizing social and environmental con- ciation; 1987.
11. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical
ditions for hearing. With the increasing number of people diagnosis of Alzheimer’s disease: report of the NINCDS-ADRDA Work Group un-
with hearing loss, research into the mechanistic path- der the auspices of Department of Health and Human Services Task Force on
ways linking hearing loss with dementia and the poten- Alzheimer’s Disease. Neurology. 1984;34(7):939-944.
tial of rehabilitative strategies to moderate this associa- 12. Petersen RC, Smith GE, Waring SC, Ivnik RJ, Tangalos EG, Kokmen E. Mild cog-
tion are critically needed. nitive impairment: clinical characterization and outcome. Arch Neurol. 1999;
56(3):303-308.
13. Schoenfeld D. Partial residuals for the proportional hazards regression model.
Accepted for Publication: April 29, 2010. Biometrika. 1982;69(1):239-241.
Correspondence: Frank R. Lin, MD, PhD, Department 14. Therneau T, Grambsch P. Modeling Survival Data: Extending the Cox Model. New
of Otolaryngology–Head and Neck Surgery, The Johns York, NY: Springer Publishing Co Inc; 2000.
15. Northridge ME. Public health methods: attributable risk as a link between cau-
Hopkins School of Medicine, JHOC 6120, 601 N Caro- sality and public health action. Am J Public Health. 1995;85(9):1202-1204.
line St, Baltimore, MD 21287 ([email protected]). 16. Cacciatore F, Napoli C, Abete P, Marciano E, Triassi M, Rengo F. Quality of life
Author Contributions: Drs Metter and Ferrucci had full determinants and hearing function in an elderly population: Osservatorio Geri-
access to all the data in the study and take responsibility atrico Campano Study Group. Gerontology. 1999;45(6):323-328.
for the integrity of the data and the accuracy of the data 17. Peters CA, Potter JF, Scholer SG. Hearing impairment as a predictor of cognitive
decline in dementia. J Am Geriatr Soc. 1988;36(11):981-986.
analysis. Study concept and design: Lin, Metter, and Fer- 18. Uhlmann RF, Larson EB, Koepsell TD. Hearing impairment and cognitive decline in
rucci. Acquisition of data: Metter, O’Brien, Resnick, Zon- senile dementia of the Alzheimer’s type. J Am Geriatr Soc. 1986;34(3):207-210.
derman, and Ferrucci. Analysis and interpretation of data: 19. Gennis V, Garry PJ, Haaland KY, Yeo RA, Goodwin JS. Hearing and cognition in
Lin, Metter, O’Brien, Resnick, Zonderman, and Ferrucci. the elderly: new findings and a review of the literature. Arch Intern Med. 1991;
151(11):2259-2264.
Drafting of the manuscript: Lin and Metter. Critical revi- 20. Popelka MM, Cruickshanks KJ, Wiley TL, Tweed TS, Klein BE, Klein R. Low preva-
sion of the manuscript for important intellectual content: Lin, lence of hearing aid use among older adults with hearing loss: the Epidemiology
O’Brien, Resnick, Zonderman, and Ferrucci. Statistical of Hearing Loss Study. J Am Geriatr Soc. 1998;46(9):1075-1078.
analysis: Lin, Metter, Zonderman, and Ferrucci. Obtained 21. Gordon-Salant S. Hearing loss and aging: new research findings and clinical
funding: O’Brien and Resnick. Administrative, technical, or implications. J Rehabil Res Dev. 2005;42(4)(suppl 2):9-24.
22. Pickles JO. An Introduction to the Physiology of Hearing. Bingley, England: Em-
material support: O’Brien and Zonderman. Study supervi- erald Group Publishing; 2008.
sion: Resnick and Ferrucci. 23. Grimes AM, Grady CL, Pikus A. Auditory evoked potentials in patients with de-
Financial Disclosure: None reported. mentia of the Alzheimer type. Ear Hear. 1987;8(3):157-161.

(REPRINTED) ARCH NEUROL / VOL 68 (NO. 2), FEB 2011 WWW.ARCHNEUROL.COM

219
Downloaded from www.archneurol.com at Emory University, on February 17, 2011
©2011 American Medical Association. All rights reserved.
24. Sinha UK, Hollen KM, Rodriguez R, Miller CA. Auditory system degeneration in sources reduce the effect of Alzheimer pathology on other cognitive systems.
Alzheimer’s disease. Neurology. 1993;43(4):779-785. Neurology. 2008;70(17):1534-1542.
25. Parvizi J, Van Hoesen GW, Damasio A. The selective vulnerability of brainstem 36. Strawbridge WJ, Wallhagen MI, Shema SJ, Kaplan GA. Negative consequences
nuclei to Alzheimer’s disease. Ann Neurol. 2001;49(1):53-66. of hearing impairment in old age: a longitudinal analysis. Gerontologist. 2000;
26. Baloyannis SJ, Mauroudis I, Manolides SL, Manolides LS. Synaptic alterations in 40(3):320-326.
the medial geniculate bodies and the inferior colliculi in Alzheimer’s disease: a Golgi 37. Weinstein BE, Ventry IM. Hearing impairment and social isolation in the elderly.
and electron microscope study. Acta Otolaryngol. 2009;129(4):416-418. J Speech Hear Res. 1982;25(4):593-599.
27. Gates GA, Beiser A, Rees TS, D’Agostino RB, Wolf PA. Central auditory dysfunc- 38. Fratiglioni L, Wang HX, Ericsson K, Maytan M, Winblad B. Influence of social
tion may precede the onset of clinical dementia in people with probable Alzhei- network on occurrence of dementia: a community-based longitudinal study. Lancet.
mer’s disease. J Am Geriatr Soc. 2002;50(3):482-488. 2000;355(9212):1315-1319.
28. O’Grady G, Boyles AL, Speer M, DeRuyter F, Strittmatter W, Worley G. Apolipo- 39. Barnes LL, Mendes de Leon CF, Wilson RS, Bienias JL, Evans DA. Social re-
protein E alleles and sensorineural hearing loss. Int J Audiol. 2007;46(4):183- sources and cognitive decline in a population of older African Americans and whites.
186. Neurology. 2004;63(12):2322-2326.
29. Stern Y. Cognitive reserve. Neuropsychologia. 2009;47(10):2015-2028. 40. Bennett DA, Schneider JA, Tang Y, Arnold SE, Wilson RS. The effect of social
30. Holtzer R, Rakitin BC, Steffener J, Flynn J, Kumar A, Stern Y. Age effects on load- networks on the relation between Alzheimer’s disease pathology and level of cog-
dependent brain activations in working memory for novel material. Brain Res. nitive function in old people: a longitudinal cohort study. Lancet Neurol. 2006;
2009;1249:148-161. 5(5):406-412.
31. Zarahn E, Rakitin B, Abela D, Flynn J, Stern Y. Age-related changes in brain acti- 41. Dalton DS, Cruickshanks KJ, Klein BE, Klein R, Wiley TL, Nondahl DM. The im-
vation during a delayed item recognition task. Neurobiol Aging. 2007;28(5):784- pact of hearing loss on quality of life in older adults. Gerontologist. 2003;43
798. (5):661-668.
32. Savva GM, Wharton SB, Ince PG, Forster G, Matthews FE, Brayne C; Medical Re- 42. Lazarov O, Robinson J, Tang YP, et al. Environmental enrichment reduces A␤
search Council Cognitive Function and Ageing Study. Age, neuropathology, and levels and amyloid deposition in transgenic mice. Cell. 2005;120(5):701-713.
dementia. N Engl J Med. 2009;360(22):2302-2309. 43. Verghese J, Lipton RB, Katz MJ, et al. Leisure activities and the risk of dementia
33. Tun PA, McCoy S, Wingfield A. Aging, hearing acuity, and the attentional costs in the elderly. N Engl J Med. 2003;348(25):2508-2516.
of effortful listening. Psychol Aging. 2009;24(3):761-766. 44. Korn EL, Graubard BI, Midthune D. Time-to-event analysis of longitudinal follow-
34. Pichora-Fuller MK, Schneider BA, Daneman M. How young and old adults listen up of a survey: choice of the time-scale. Am J Epidemiol. 1997;145(1):72-80.
to and remember speech in noise. J Acoust Soc Am. 1995;97(1):593-608. 45. Zhan W, Cruickshanks KJ, Klein BE, et al. Generational differences in the prevalence
35. Boyle PA, Wilson RS, Schneider JA, Bienias JL, Bennett DA. Processing re- of hearing impairment in older adults. Am J Epidemiol. 2010;171(2):260-266.

Announcement

“What is Your Diagnosis?” is a new quarterly online fea-

ture of the Archives of Neurology edited by Lawrence S.
Honig, MD, PhD, of Columbia University. A case his-
tory including an image will be presented, followed by
the request for your diagnosis from a list of 4 possible
choices. The correct diagnosis will then be presented with
a commentary as to why it is correct. We believe it will
become a popular and anticipated new feature and wel-
come your comments and suggestions.

(REPRINTED) ARCH NEUROL / VOL 68 (NO. 2), FEB 2011 WWW.ARCHNEUROL.COM

220
Downloaded from www.archneurol.com at Emory University, on February 17, 2011
©2011 American Medical Association. All rights reserved.