607

Journal of Alzheimers Disease 23 (2011) 607615

DOI 10.3233/JAD-2010-101428
IOS Press

Coffee Intake in Midlife and Risk

of Dementia and its Neuropathologic
Correlates
Rebecca P. Gelbera, , Helen Petrovitchb , Kamal H. Masakic , G. Webster Rossb and Lon R. Whitec
a Honolulu-Asia Aging Study at Kuakini Medical Center, the VA Pacic Islands Healthcare System,
Department of Medicine, University of Hawaii, John A. Burns School of Medicine, Honolulu, HI, USA
b Honolulu-Asia Aging Study at Kuakini Medical Center, the VA Pacic Islands Healthcare System,
Department of Geriatric Medicine, University of Hawaii, John A. Burns School of Medicine, Honolulu, HI, USA
c Honolulu-Asia Aging Study at Kuakini Medical Center, Department of Geriatric Medicine, University
of Hawaii, John A. Burns School of Medicine, Honolulu, HI, USA

Accepted 10 October 2010

Abstract. While animal data suggest a protective effect of caffeine on cognition, studies in humans remain inconsistent. We
examined associations of coffee and caffeine intake in midlife with risk of dementia, its neuropathologic correlates, and cognitive
impairment among 3494 men in the Honolulu-Asia Aging Study (mean age 52 at cohort entry, 19651968) examined for dementia
in 19911993, including 418 decedents (19922004) who underwent brain autopsy. Caffeine intake was determined according
to self-reported coffee, tea, and cola consumption at baseline. Logistic regression was used to calculate odds ratios (OR) and
95% confidence intervals (CI) for overall dementia, Alzheimers disease (AD), vascular dementia (VaD), cognitive impairment
(Cognitive Abilities Screening Instrument score <74), and neuropathologic lesions at death (Alzheimer lesions, microvascular
ischemic lesions, cortical Lewy bodies, hippocampal sclerosis, generalized atrophy), according to coffee and caffeine intake.
Dementia was diagnosed in 226 men (including 118 AD, 80 VaD), and cognitive impairment in 347. There were no significant
associations between coffee or caffeine intake and risk of cognitive impairment, overall dementia, AD, VaD, or moderate/high
levels of the individual neuropathologic lesion types. However, men in the highest quartile of caffeine intake (>277.5 mg/d) were
less likely than men in the lowest quartile (115.5 mg) to have any of the lesion types (adjusted-OR, 0.45; 95% CI, 0.230.89;
p, trend = 0.04). Coffee and caffeine intake in midlife were not associated with cognitive impairment, dementia, or individual
neuropathologic lesions, although higher caffeine intake was associated with a lower odds of having any of the lesion types at
autopsy.
Keywords: Caffeine, coffee, cohort studies, dementia

INTRODUCTION
Recent studies have provided conflicting information on the role of caffeine intake in the development of
dementia. Experiments in Alzheimers disease (AD) Correspondence to: Rebecca P. Gelber, MD, DrPH, Kuakini
Physicians Tower, 405 N. Kuakini Street, Ste. 1111, Honolulu, HI
96817, USA. Tel.: +1 808 547 9706; Fax: +1 808 547 9305; E-mail:
[email protected].

transgenic mice have demonstrated beneficial effects,

with improvements in cognition in treated animals
[13]. Results from epidemiologic studies in humans,
however, have been inconclusive, with some suggesting a reduced risk of dementia with higher coffee intake
in midlife [4], and others pointing to a lack of association [5].
We therefore examined the associations between
midlife coffee and caffeine intake and the risk of
dementia, its neuropathologic correlates, and cognitive

ISSN 1387-2877/11/$27.50 2011 IOS Press and the authors. All rights reserved

608

R.P. Gelber et al. / Coffee Intake and Risk of Dementia

impairment in late life in a nested case-control study

of more than 3400 men ages 7193 years, for whom
standardized information reflecting caffeine intake had
been collected at ages 4565 years.
METHODS
Study population
Data were from the Honolulu-Asia Aging Study
(HAAS), a community-based prospective cohort study
of 3734 Japanese American men followed since 1965
as part of the Honolulu Heart Program (HHP). Details
of the study have been described elsewhere [6, 7].
Briefly, the HHP included 8006 men of Japanese ancestry born between 19001919, who were residing on
Oahu, Hawaii, in 1964. Information on baseline characteristics, including demographic, dietary, physical
activity, and medical history, was obtained through
structured interview at the first HHP examination
(19651968). The HAAS was established in 1991 as
an extension of the HHP to study aging-related conditions, with a special focus on brain diseases, and included 3734 men, ages 7193 years (representing approximately 80% of surviving members of the original
cohort) [7]. The Kuakini Medical Center Institutional
Review Board reviewed and approved this study. All
participants or their representatives provided written
informed consent.
Clinical dementia and cognitive impairment
outcomes
A nested case-control design was used with exposure and covariate information collected prospectively
from the HHP cohort in 19651968, and cases and
controls defined in 19911993, the first time at which
dementia case finding and standardized cognitive
assessments were performed. Analyses were limited
to men with available Cognitive Abilities Screening
Instrument (CASI) scores (n = 3734) and excluded
those with scores <74 who did not participate in the
dementia evaluation phase (n = 240), leaving 3494 men
available for analysis.
Case finding was conducted in a multi-step procedure according to a previously described protocol that
included screening and standardized dementia evaluations [7]. Briefly, cognitive function of all participants
was tested with the 100-point CASI, a combination of
the Hasegawa Dementia Screening Scale, the Folstein
Mini-Mental State Examination, and the Modified
Mini-Mental State Test [8]. Participants with poor

CASI scores (<74) and stratified samples from subsets with higher scores were invited to return for
the dementia evaluation, accompanied by a proxy
informant. Follow-up examinations and the dementia
evaluations included a neurologic examination, neuropsychological testing, and informant interview about
changes in cognitive function and behavior. Computed
tomography or magnetic resonance brain imaging was
performed and routine blood tests conducted for men
provisionally classified as demented.
Based on these data, a consensus diagnosis for
dementia was provided by the study neurologist and
two physicians with expertise in dementia, according to criteria of the Diagnostic and Statistical
Manual of Mental Disorders, third edition, revised
(DSM III-R) [9]. Probable or possible AD was diagnosed according to criteria from the National Institute
of Neurological and Communicative Disorders and
Stroke-Alzheimers Disease and Related Disorders
Association [10]. Dementia attributed to AD but with
minor evidence of co-prevalent conditions was classified as possible AD. In a previously examined subset
of the HAAS cohort, 77% of men with the clinical
diagnosis of probable pure AD demonstrated sufficient
neuritic plaques to meet neuropathological criteria for
definite or probable AD, while those classified as possible were less often confirmed as definite [11]. Probable
or possible vascular dementia (VaD) was diagnosed
using criteria of the California Alzheimers Disease
Diagnostic and Treatment Centers [12]. Mixed dementia attributed to VaD as the primary cause and mixed
dementia attributed principally to AD were classified
separately. For the purpose of the present analysis,
dementia cases classified as AD included probable AD,
possible AD, and mixed dementia with major attribution to AD. Dementia cases classified as VaD included
probable VaD, possible VaD, and mixed dementia with
dominant attribution to vascular causes.
We defined cognitive impairment as a CASI score
<74 (16th percentile of scores for all men examined), independent of dementia classification. This
corresponds approximately to a score of 22 on the
Mini-Mental State Examination, a level often used to
indicate cognitive impairment [13]. Substantial overlap existed between cognitive impairment and general
(all cause) dementia classifications.
Neuropathologic outcomes
Of the 3494 men included in the clinical outcome analyses, 418 decedents were autopsied from
19922004. These men were included in analyses of

R.P. Gelber et al. / Coffee Intake and Risk of Dementia

midlife coffee and caffeine intake and the presence of

neuropathologic lesions at death. Details of the HAAS
autopsy protocol and evaluation of the neuropathologic
material have been described in detail elsewhere [14,
15]. Permission for the full or limited autopsy was
obtained from the next of kin.
We examined each of the following categories of
neuropathologic lesions without respect to clinical
condition: Alzheimer lesions (neocortical neurofibrillary tangles and neuritic plaques), microvascular
ischemic lesions (microinfarcts and lacunar infarcts),
neocortical Lewy bodies, hippocampal sclerosis, and
generalized brain atrophy. For each of the five
lesion categories, decedents were classified using a
3-level lesion index (none/negligible/low, moderate,
and high). As previously described, these indices were
developed through a multi-step process to reflect similar strengths of association with cognitive impairment
and dementia [14]. For these analyses, each lesion type
was defined as being present if moderate or high levels
of the lesion were identified.
Assessment of coffee and caffeine intake
Information on coffee, tea, and cola consumption
was obtained from a standardized, validated 24-hour
dietary recall administered on all participants at cohort
entry (19651968) [16, 17]. Participants reported caffeinated coffee intake as the number of 4-oz cups
consumed. For analyses of clinical outcomes, coffee
intake categories were defined as 0, 48, 1216, 2024,
and 28 oz/d. For analyses of pathological outcomes,
four coffee intake categories were used, given fewer
case numbers (0, 48, 1216, 20 oz/d). Tea intake
was reported as the number of 4-oz cups consumed
(without specification of black or green tea), and cola
intake as the number of 6-oz cups consumed.
Caffeine intake was calculated based on the estimated caffeine content for coffee (137 mg/8 oz), tea
(47 mg/8 oz), and cola (46 mg/12 oz) consumption
reported from the 24-hour dietary recall [18]. Caffeine
was examined in quintiles for analyses of clinical outcomes and in quartiles for analyses of pathological
outcomes, given fewer case numbers, with quartiles
calculated based on the autopsied subgroup. Caffeine
consumption was also assessed as a continuous variable, per standard deviation (SD) unit (223 mg).
Covariates
Covariates assessed at baseline included age (years,
continuous), body mass index (BMI, <25.0, 25.029.9,

609

30.0 kg/m2 ), physical activity index (tertiles) [19],

cigarette smoking (never, past, current), alcohol
consumption (0, >0<0.5, 0.5<1.0, 1.0<2.0,
2.0 drinks/d), education (years of completed schooling), number of childhood years spent in Japan
(<5/5), occupational complexity (a 6-level ordinal scale reflecting average educational attainment),
marital status, elevated cholesterol (serum level
240 mg/dL), hypertension (systolic blood pressure
[SBP] 140 mmHg, diastolic blood pressure [DBP]
90 mmHg, or history of antihypertensive medication
use), diabetes (history of diabetes or use of insulin or
oral medications for diabetes), and cardiovascular disease (history of stroke, angina, myocardial infarction,
or other coronary heart disease). Genotyping for the
presence of apolipoprotein E (APOE) 4 (1 or 2 alleles
vs. none) was performed at Duke University, Durham,
NC, using conventional methods [20].
Analysis
Age-adjusted participant characteristics in midlife
were compared according to categories of coffee intake
using analysis of variance for continuous variables and
logistic regression for categorical variables. Logistic
regression was used to calculate odds ratios (ORs)
and corresponding 95% confidence intervals (CIs) for
overall dementia, AD, VaD, their neuropathologic correlates, and cognitive impairment, according to midlife
coffee and caffeine intake. The main multivariableadjusted model included age (age at cohort entry for
clinical outcomes and age at death for pathological
outcomes), physical activity index, smoking, years
of education, hypertension, elevated cholesterol (all
assessed at HHP baseline), plus APOE 4 status determined at the HAAS examination. A second model
further adjusted for baseline BMI, alcohol consumption, number of childhood years spent in Japan, history
of diabetes mellitus, cardiovascular disease, occupational complexity, and marital status, in addition to the
variables included in the main model.
Separate models included joint terms to evaluate
potential interactions between coffee intake and smoking status (never, past, current), physical activity index
(tertile 3 vs. 1/2), and BMI (</25.0 kg/m2 ). The
likelihood ratio test (LRT) was used to assess for
statistically significant effect modification, contrasting age-adjusted models with and without interaction
terms of interest. Tests for linear trends in the OR
across categories of exposure were performed using the
LRT, assigning median values to each category where
appropriate.

610

R.P. Gelber et al. / Coffee Intake and Risk of Dementia

similar results after adjustment for the main potential confounders. Further adjustment for BMI, alcohol
consumption, diabetes, cardiovascular disease, occupational complexity, marital status, and childhood
years spent in Japan did not materially alter the results.
Coffee and caffeine consumption were also not associated with CASI score as a continuous variable (data
not shown).
Coffee intake was not significantly associated with
overall dementia in analyses stratified by smoking status using joint terms (LRT, p = 0.66), nor was there
evidence for effect modification by midlife physical
activity or BMI (LRT, p = 0.69 and 0.78, respectively).
Similarly, there were no significant associations with
cognitive impairment examining the joint effects of
coffee intake and smoking status, physical activity, or
BMI (data not shown).
Among 418 decedents, there were no significant
associations between midlife coffee (Table 4 ) or caffeine (Table 5) intake and the presence of Alzheimer
lesion, microvascular ischemic lesions, generalized
atrophy, cortical Lewy bodies, or hippocampal sclerosis at autopsy. However, the risk of having any of
the lesions was decreased among those in the highest quartile of caffeine intake (multivariable-adjusted
OR, 0.45; 95% CI, 0.230.89; LRT p, trend = 0.04;

In secondary analyses, linear regression was used

to examine associations between coffee and caffeine
intake and CASI score as a continuous variable.
Mean caffeine intake was compared between groups
of decedents defined by the presence of neuropathologic lesions using general linear models. Two-sided
p-values were reported in all analyses. p-values <0.05
were considered statistically significant. All data analyses were performed using SAS Software Version 9.2
(SAS Institute Inc, Cary, NC).

RESULTS
Participant characteristics are shown according to
midlife coffee intake in Table 1. Men who consumed
more coffee were slightly younger, less likely to
have hypertension, more likely to have high cholesterol and to smoke, and tended to be less physically
active.
A total of 226 cases of dementia (including 118
AD, 80 VaD) and 347 cases of cognitive impairment
were diagnosed 25 years following cohort entry. Coffee (Table 2) and caffeine (Table 3) intake were not
significantly associated with overall dementia, AD,
VaD, or cognitive impairment adjusting for age, with

Table 1
Age-adjusted midlife characteristics according to coffee intake for the 3494 men in the HAAS
Coffee intake in midlife (oz/d)
Number of men
Mean age (SD)
Mean caffeine intake, mg (SD)
BMI 25.0 kg/m2 (%)
Hypertension (%)
High cholesterol (%)
Diabetes mellitus (%)
Cardiovascular disease|| (%)
APOE 4 positive (%)
Smoking status (%)
Never
Past
Current
Highest physical activity (%)
Alcohol intake 1 drink/d (%)
Education, yrs (se)
Married (%)
5 childhood yrs in Japan (%)

p*

48

1216

2024

28

588
52.7 (4.7)
56.3 (65.0)
34.5
36.3
23.2
9.0
2.5
19.1

1074
53.0 (4.9)
168.1 (67.5)
33.8
37.2
22.9
6.5
2.2
17.9

914
52.4 (4.4)
282.8 (63.3)
33.6
33.6
24.9
5.8
2.5
16.9

463
52.2 (4.2)
429.0 (60.1)
33.2
29.9
28.2
6.1
1.8
21.5

455
51.3 (3.8)
708.5 (233.4)
36.4
27.3
28.9
8.4
3.3
20.0

<0.001
<0.001
0.68
<0.001
0.005
0.58
0.60
0.34

45.3
30.6
24.0
35.8
20.1
10.5 (0.13)
89.2
17.7

41.6
27.2
31.2
36.4
27.1
10.7 (0.10)
91.8
17.7

31.8
30.0
38.2
29.2
26.1
10.6 (0.10)
92.1
17.1

27.9
31.8
40.2
31.1
21.0
10.6 (0.14)
91.5
17.9

19.2
22.6
57.6
34.2
19.2
10.8 (0.15)
91.6
14.6

<0.001
0.15
<0.001
0.07
0.14
0.60
0.24
0.27

* p-values are from Wald chi square tests for trend for categorical variables and analysis of variance for continuous
variables;
Hypertension defined as systolic blood pressure 140 mmHg, or diastolic blood pressure 90 mmHg, or antihypertensive medication use;
High cholesterol defined as level 240 mg/dl;
Diabetes defined as history of diabetes or use of insulin or oral medications for diabetes;
|| Cardiovascular disease defined as history of myocardial infarction, angina, other coronary heart disease, or stroke;
High physical activity defined as the highest tertile of physical activity index.

R.P. Gelber et al. / Coffee Intake and Risk of Dementia

611

Table 2
Odds ratios (95% confidence intervals) for dementia and cognitive impairment after 25 years according to midlife coffee
intake
Coffee intake in midlife (oz/d)
48

1216

2024

28

1074

914

463

455

69
0.83 (0.541.26)
0.93 (0.591.46)

63
1.13 (0.731.74)
1.24 (0.781.97)

30
1.15 (0.691.93)
1.14 (0.661.98)

23
1.11 (0.641.93)
1.09 (0.592.00)

39
0.76 (0.441.30)
0.89 (0.501.59)

32
0.93 (0.531.62)
1.09 (0.602.00)

14
0.89 (0.451.78)
0.95 (0.452.00)

8
0.64 (0.281.48)
0.59 (0.231.54)

26
1.34 (0.632.83)
1.37 (0.642.94)

21
1.55 (0.713.35)
1.54 (0.693.41)

12
1.93 (0.814.58)
1.57 (0.623.95)

11
2.21 (0.915.37)
1.96 (0.765.03)

120
1.04 (0.731.47)
1.08 (0.741.57)

86
1.02 (0.701.47)
1.07 (0.721.59)

47
1.20 (0.781.85)
1.18 (0.751.87)

34
1.05 (0.661.67)
0.99 (0.601.65)

0
No. of men
588
All dementia
No. of cases
41
Age-adjusted
1.00
Multivariable-adjusted*
1.00
Alzheimers disease
No. of cases
25
Age-adjusted
1.00
Multivariable-adjusted*
1.00
Vascular dementia
No. of cases
10
Age-adjusted
1.00
Multivariable-adjusted*
1.00
Cognitive impairment (CASI <74)
No. of cases
60
Age-adjusted
1.00
Multivariable-adjusted*
1.00

* Adjusts for age, physical activity index, smoking, years of education, APOE 4 status, elevated cholesterol, and
hypertension.
Table 3
Odds ratios (95% confidence intervals) for dementia and cognitive impairment after 25 years according to midlife caffeine
intake
Caffeine intake (mg/d)
No. of men
All dementia
No. of cases
Age-adjusted
Multivariable-adjusted*
Alzheimers disease
No. of cases
Age-adjusted
Multivariable-adjusted*
Vascular dementia
No. of cases
Age-adjusted
Multivariable-adjusted*
Cognitive impairment (CASI <74)
No. of cases
Age-adjusted
Multivariable-adjusted*

0115.5

>115.5188.0

>188.0277.5

>277.5415.0

>415.02673.0

707

604

784

704

695

43
1.00
1.00

42
51
54
36
1.13 (0.711.80) 1.26 (0.811.96) 1.46 (0.942.26) 1.25 (0.772.03)
1.21 (0.741.96) 1.31 (0.822.09) 1.47 (0.922.35) 1.12 (0.661.91)

26
1.00
1.00

24
26
24
18
1.05 (0.581.90) 1.05 (0.591.87) 1.03 (0.571.86) 1.08 (0.572.03)
1.20 (0.652.23) 1.15 (0.622.11) 1.07 (0.572.00) 0.95 (0.461.95)

11
1.00
1.00

15
19
22
13
1.59 (0.713.52) 1.82 (0.853.89) 2.28 (1.084.81) 1.74 (0.763.98)
1.52 (0.683.43) 1.83 (0.843.99) 2.09 (0.974.52) 1.45 (0.593.53)

67
1.00
1.00

69
77
80
54
1.22 (0.831.78) 1.19 (0.821.72) 1.37 (0.951.97) 1.13 (0.761.69)
1.20 (0.801.80) 1.21 (0.821.79) 1.33 (0.901.98) 1.02 (0.661.59)

* Adjusts for age, physical activity index, smoking, years of education, APOE 4 status, elevated cholesterol, and
hypertension.

Table 5). Examining caffeine as a continuous variable, per SD-unit (223 mg), there were no significant
associations with any of the clinical outcomes or with
specific neuropathologic lesions at death. However,
caffeine intake was again associated with the presence
of any of the lesions. Adjusting for age at death, the
OR for having any lesions, as compared to no lesions,
was 0.78 (95% CI, 0.640.96). Additionally adjusting for education, APOE 4 status, smoking, physical

activity, elevated cholesterol, and history of hypertension, the OR for any lesions was 0.77 (95% CI,
0.610.97).
Mean caffeine intake in midlife was 276 mg among
the 154 decedents with microvascular ischemic
lesions, as compared to 333 mg among the 121 decedents without any lesions, adjusting for age at death,
years of education, and APOE 4 status (p = 0.06).
Adjusted mean caffeine intake among decedents with

612

R.P. Gelber et al. / Coffee Intake and Risk of Dementia

Table 4
Odds ratios (95% confidence intervals) for brain pathology according to midlife coffee intake among 418 decedents,
19922004
Coffee intake in midlife (oz/d)
No. of men
Alzheimer lesions
No. of cases
Age-adjusted
Multivariable-adjusted*
Microvascular ischemic lesions
No. of cases
Age-adjusted
Multivariable-adjusted*
Generalized atrophy
No. of cases
Age-adjusted
Multivariable-adjusted*
Cortical Lewy bodies
No. of cases
Age-adjusted
Multivariable-adjusted*
Hippocampal sclerosis
No. of cases
Age-adjusted
Multivariable-adjusted*
Any lesions
No. of cases
Age-adjusted
Multivariable-adjusted*

48

1216

>20

64

137

103

114

14
1.00
1.00

31
1.02 (0.492.11)
1.16 (0.532.54)

30
1.56 (0.743.29)
1.82 (0.814.08)

28
1.20 (0.572.52)
1.02 (0.442.37)

28
1.00
1.00

53
0.81 (0.441.48)
0.80 (0.421.49)

39
0.79 (0.421.49)
0.84 (0.431.62)

34
0.55 (0.291.04)
0.58 (0.291.15)

28
1.00
1.00

63
1.06 (0.571.96)
1.20 (0.632.27)

47
1.12 (0.592.13)
1.30 (0.662.55)

50
1.02 (0.541.92)
1.24 (0.622.48)

14
1.00
1.00

20
0.59 (0.271.27)
0.63 (0.281.40)

16
0.67 (0.301.50)
0.76 (0.331.76)

13
0.46 (0.201.07)
0.52 (0.211.31)

9
1.00
1.00

12
0.56 (0.221.44)
0.50 (0.181.40)

8
0.53 (0.191.48)
0.65 (0.221.93)

9
0.54 (0.201.45)
0.64 (0.211.97)

49
1.00
1.00

100
0.78 (0.391.59)
0.84 (0.401.74)

75
0.83 (0.401.72)
1.00 (0.462.17)

72
0.51 (0.251.03)
0.52 (0.241.13)

* Adjusts for age at death, physical activity index, smoking, years of education, APOE 4 status, elevated cholesterol,
and hypertension.

AD lesions was 279 mg (p = 0.10, as compared to

those without any lesions).
DISCUSSION
In this nested case-control study, coffee and caffeine intake in midlife were not associated with overall
dementia, AD, VaD, or cognitive impairment after 25
years. Among decedents, the highest level of caffeine
consumption was associated with a lower risk of having any of the neuropathologic lesions, however, there
were no significant associations according to lesion
type.
The association between the highest level of caffeine
intake and a lower OR for having moderate/high levels
of any of the lesion types may reflect a causal relationship. Analyses of some of the specific lesion types
may not have reached statistical significance due to
limited case numbers, and it is plausible that caffeine
serves a protective role common to the development
of co-prevalent lesion types. Caffeine, an adenosine
receptor antagonist, acts as a brain stimulant acutely
and enhances alertness [21]. Recent studies in mice
have supported a protective role of caffeine exposure in
the development of cognitive impairment, the restora-

tion of cognition, and the reversal of AD pathology,

suggesting a potentially therapeutic and preventive role
of caffeine in AD [13]. Furthermore, higher caffeine
intake has been associated with fewer white matter
lesions on magnetic resonance brain imaging in women
[22]. Caffeine has also been shown in multiple studies to have a protective role in the development of
Parkinsons disease [23, 24].
However, previous epidemiologic studies of caffeine intake and cognition in humans have provided
inconsistent results [25]. In a case-control study from
Portugal, caffeine intake was associated with a lower
risk of AD [26]. However, caffeine consumption was
assessed retrospectively as the average intake over the
prior 20 years. A subsequent study from Finland suggested that midlife coffee intake decreased risk of late
life AD and any dementia [4], whereas another study
among Finnish twins did not show any associations
[5]. In prospective cohort studies from France [27] and
Portugal [28] among men and women 65 years of age,
higher coffee intake was associated with reduced cognitive decline among women over the ensuing 4 years,
however, midlife coffee intake and long-term effects
were not examined.

R.P. Gelber et al. / Coffee Intake and Risk of Dementia

613

Table 5
Odds ratios (95% confidence intervals) for brain pathology according to midlife caffeine intake among 418 decedents,
19922004
Caffeine intake, quartiles (mg/d)
No. of men
Alzheimer lesions
No. of cases
Age-adjusted
Multivariable-adjusted*
Microvascular ischemic lesions
No. of cases
Age-adjusted
Multivariable-adjusted*
Generalized atrophy
No. of cases
Age-adjusted
Multivariable-adjusted*
Cortical Lewy bodies
No. of cases
Age-adjusted
Multivariable-adjusted*
Hippocampal sclerosis
No. of cases
Age-adjusted
Multivariable-adjusted*
Any lesions
No. of cases
Age-adjusted
Multivariable-adjusted*

0137.0

138.5231.0

251.5391.5

411.01872.5

109

99

105

105

27
1.00
1.00

25
1.00 (0.531.90)
1.10 (0.552.18)

25
0.98 (0.521.84)
0.96 (0.481.90)

26
1.03 (0.551.93)
0.85 (0.421.73)

43
1.00
1.00

46
1.31 (0.752.27)
1.27 (0.722.26)

31
0.64 (0.361.12)
0.66 (0.371.20)

34
0.73 (0.421.28)
0.79 (0.431.45)

54
1.00
1.00

40
0.66 (0.371.16)
0.68 (0.381.22)

46
0.80 (0.461.40)
0.87 (0.491.55)

48
0.87 (0.501.52)
1.01 (0.561.84)

19
1.00
1.00

14
0.76 (0.361.62)
0.82 (0.381.80)

18
0.99 (0.492.03)
1.10 (0.522.31)

12
0.62 (0.281.36)
0.71 (0.301.66)

13
1.00
1.00

9
0.71 (0.291.77)
0.67 (0.251.78)

8
0.61 (0.241.56)
0.62 (0.231.71)

8
0.63 (0.251.60)
0.77 (0.272.20)

87
1.00
1.00

70
0.59 (0.311.12)
0.61 (0.311.20)

71
0.53 (0.281.00)
0.57 (0.301.10)

68
0.46 (0.250.86)
0.45 (0.230.89)

* Adjusts for age at death, physical activity index, smoking, years of education, APOE 4 status, elevated cholesterol,
and hypertension. LRT p, trend=0.04 for any lesions.

Our study has several strengths, including the standardized, prospective collection of information on
risk factors and multiple potential confounders, the
community-based and longitudinal study design, a
large number of participants with high response rates
at the baseline and follow-up examinations, and highly
standardized clinical and neuropathologic endpoints.
Furthermore, previous analyses have provided robust
demonstrations of the relationship of midlife coffee
and caffeine intake with another neurologic condition,
Parkinsons disease [23].
Several limitations should be considered in the interpretation of our results. The cohort included only
Japanese American men residing on Oahu, which
thereby limits generalizability of our results. However,
limiting the study demographic also reduces potential
confounding due to ethnicity and geography. Secondly,
calculation of caffeine intake was likely imprecise,
based on 24-hour recall of coffee, tea, and cola intake,
without information on caffeine from food, such as
chocolate. However, we expect that any misclassification of caffeine intake would have been random
with respect to outcome assessment. Lastly, we did
not analyze incident cases of dementia following expo-

sure assessment, but rather identified prevalent cases

after 25 years. However, censoring and death prior to
case ascertainment may be expected to lead to underestimates of associations between caffeine intake and
risk of dementia. Furthermore, although dementia may
have initially developed years earlier, it is unlikely that
participants were demented at the time of risk factor
assessment in midlife.
In conclusion, our study does not support an association between midlife coffee or caffeine intake and the
risk of dementia or cognitive impairment. The highest levels of caffeine consumption was associated with
a decreased risk of having any of the neuropathologic correlates of dementia at autopsy, however, there
were no associations with specific lesion types. Further
prospective studies and future clinical trials of caffeine
intake, while challenging [29], are needed to confirm
our findings.
ACKNOWLEDGMENT
This work was supported by grants 5U01
AG017155-09, 3U01 AG017155-09S1, and 5U01
AG19349-09 from the National Institutes of Health

614

R.P. Gelber et al. / Coffee Intake and Risk of Dementia

and Aging and resources from the VA Pacific Islands

Health Care System. The information in this paper
does not necessarily reflect the position or the policy
of the government and no official endorsement should
be inferred.
We are indebted to the participants in the HonoluluAsia Aging Study for their outstanding commitment
and to the entire staff of the Honolulu-Asia Aging
Study for their expert assistance.
Authors disclosures available online (http://www.jalz.com/disclosures/view.php?id=652).

[11]

[12]

[13]

[14]

REFERENCES
[15]
[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

Cao C, Cirrito JR, Lin X, Wang L, Verges DK, Dickson A,

Mamcarz M, Zhang C, Mori T, Arendash GW, Holtzman DM,
Potter H (2009) Caffeine suppresses amyloid-beta levels in
plasma and brain of Alzheimers disease transgenic mice. J
Alzheimers Dis 17, 681-697.
Arendash GW, Mori T, Cao C, Mamcarz M, Runfeldt M,
Dickson A, Rezai-Zadeh K, Tane J, Citron BA, Lin X,
Echeverria V, Potter H (2009) Caffeine reverses cognitive
impairment and decreases brain amyloid-beta levels in aged
Alzheimers disease mice. J Alzheimers Dis 17, 661-680.
Arendash GW, Schleif W, Rezai-Zadeh K, Jackson EK,
Zacharia LC, Cracchiolo JR, Shippy D, Tan J (2006) Caffeine
protects Alzheimers mice against cognitive impairment and
reduces brain beta-amyloid production. Neuroscience 142,
941-952.
Eskelinen MH, Ngandu T, Tuomilehto J, Soininen H,
Kivipelto M (2009) Midlife coffee and tea drinking and the
risk of late-life dementia: a population-based CAIDE study.
J Alzheimers Dis 16, 85-91.
Laitala VS, Kaprio J, Koskenvuo M, Raiha I, Rinne JO,
Silventoinen K (2009) Coffee drinking in middle age is not
associated with cognitive performance in old age. Am J Clin
Nutr 90, 640-646.
Syme SL, Marmot MG, Kagan A, Kato H, Rhoads G (1975)
Epidemiologic studies of coronary heart disease and stroke in
Japanese men living in Japan, Hawaii and California: introduction. Am J Epidemiol 102, 477-480.
White L, Petrovitch H, Ross GW, Masaki KH, Abbott
RD, Teng EL, Rodriguez BL, Blanchette PL, Havlik RJ,
Wergowske G, Chiu D, Foley DJ, Murdaugh C, Curb JD
(1996) Prevalence of dementia in older Japanese-American
men in Hawaii: The Honolulu-Asia Aging Study. JAMA 276,
955-960.
Teng EL, Hasegawa K, Homma A, Imai Y, Larson E, Graves
A, Sugimoto K, Yamaguchi T, Sasaki H, Chiu D, et al.
(1994) The Cognitive Abilities Screening Instrument (CASI):
a practical test for cross-cultural epidemiological studies of
dementia. Int Psychogeriatr 6, 45-58; discussion 62.
American Psychiatric Association (1987) Diagnostic and Statistical Manual of Mental Disorders, third edition, revised,
American Psychiatric Association, Washington, DC.
McKhann G, Drachman D, Folstein M, Katzman R, Price
D, Stadlan EM (1984) Clinical diagnosis of Alzheimers disease: report of the NINCDS-ADRDA Work Group under the
auspices of Department of Health and Human Services Task
Force on Alzheimers Disease. Neurology 34, 939-944.

[16]

[17]

[18]

[19]

[20]

[21]

[22]

[23]

[24]

[25]

[26]
[27]

Petrovitch H, White LR, Ross GW, Steinhorn SC, Li CY,

Masaki KH, Davis DG, Nelson J, Hardman J, Curb JD,
Blanchette PL, Launer LJ, Yano K, Markesbery WR (2001)
Accuracy of clinical criteria for AD in the Honolulu-Asia
Aging Study, a population-based study. Neurology 57, 226234.
Chui HC, Victoroff JI, Margolin D, Jagust W, Shankle R,
Katzman R (1992) Criteria for the diagnosis of ischemic
vascular dementia proposed by the State of California
Alzheimers Disease Diagnostic and Treatment Centers. Neurology 42, 473-480.
Abbott RD, White LR, Ross GW, Masaki KH, Curb JD,
Petrovitch H (2004) Walking and dementia in physically capable elderly men. JAMA 292, 1447-1453.
White L (2009) Brain lesions at autopsy in older JapaneseAmerican men as related to cognitive impairment and
dementia in the final years of life: a summary report from the
Honolulu-Asia aging study. J Alzheimers Dis 18, 713-725.
Petrovitch H, Ross GW, Steinhorn SC, Abbott RD,
Markesbery W, Davis D, Nelson J, Hardman J, Masaki K,
Vogt MR, Launer L, White LR (2005) AD lesions and infarcts
in demented and non-demented Japanese-American men. Ann
Neurol 57, 98-103.
Hankin JH, Nomura A, Rhoads GG (1975) Dietary patterns
among men of Japanese ancestry in Hawaii. Cancer Res 35,
3259-3264.
McGee D, Rhoads G, Hankin J, Yano K, Tillotson J (1982)
Within-person variability of nutrient intake in a group of
Hawaiian men of Japanese ancestry. Am J Clin Nutr 36, 657663.
Lopez-Garcia E, Rodriguez-Artalejo F, Rexrode KM,
Logroscino G, Hu FB, van Dam RM (2009) Coffee consumption and risk of stroke in women. Circulation 119,
1116-1123.
Taaffe DR, Irie F, Masaki KH, Abbott RD, Petrovitch H,
Ross GW, White LR (2008) Physical activity, physical function, and incident dementia in elderly men: the Honolulu-Asia
Aging Study. J Gerontol A Biol Sci Med Sci 63, 529-535.
Hixson JE, Vernier DT (1990) Restriction isotyping of human
apolipoprotein E by gene amplification and cleavage with
HhaI. J Lipid Res 31, 545-548.
Fredholm BB, Battig K, Holmen J, Nehlig A, Zvartau EE
(1999) Actions of caffeine in the brain with special reference
to factors that contribute to its widespread use. Pharmacol
Rev 51, 83-133.
Ritchie K, Artero S, Portet F, Brickman A, Muraskin J,
Beanino E, Ancelin ML, Carriere I (2010) Caffeine, cognitive
functioning, and white matter lesions in the elderly: establishing causality from epidemiological evidence. J Alzheimers
Dis 20 (Suppl 1), S161-S166.
Ross GW, Abbott RD, Petrovitch H, Morens DM, Grandinetti
A, Tung KH, Tanner CM, Masaki KH, Blanchette PL, Curb
JD, Popper JS, White LR (2000) Association of coffee and
caffeine intake with the risk of Parkinson disease. JAMA 283,
2674-2679.
Prediger RD (2010) Effects of caffeine in Parkinsons disease:
from neuroprotection to the management of motor and nonmotor symptoms. J Alzheimers Dis 20 (Suppl 1), S205-S220.
Santos C, Costa J, Santos J, Vaz-Carneiro A, Lunet N (2010)
Caffeine intake and dementia: systematic review and metaanalysis. J Alzheimers Dis 20 (Suppl 1), S187-S204.
Maia L, de Mendonca A (2002) Does caffeine intake protect
from Alzheimers disease? Eur J Neurol 9, 377-382.
Ritchie K, Carriere I, de Mendonca A, Portet F, Dartigues
JF, Rouaud O, Barberger-Gateau P, Ancelin ML (2007)

R.P. Gelber et al. / Coffee Intake and Risk of Dementia

[28]

The neuroprotective effects of caffeine: a prospective population study (the Three City Study). Neurology 69, 536545.
Santos C, Lunet N, Azevedo A, de Mendonca A, Ritchie K,
Barros H (2010) Caffeine intake is associated with a lower

[29]

615

risk of cognitive decline: a cohort study from Portugal. J

Alzheimers Dtis 20 (Suppl 1), S175-S185.
de Mendonca A, Cunha RA (2010) Putative neuroprotective
effects of caffeine in clinical trials. Concluding remarks. J
Alzheimers Dis 20 (Suppl 1), S249-S252.