American Journal of Epidemiology
Copyright © 2005 by the Johns Hopkins Bloomberg School of Public Health
All rights reserved
Vol. 161, No. 5
Printed in U.S.A.
DOI: 10.1093/aje/kwi062
Alcohol Consumption, Binge Drinking, and Early Coronary Calcification: Findings
from the Coronary Artery Risk Development in Young Adults (CARDIA) Study
Mark J. Pletcher1,2, Paul Varosy3, Catarina I. Kiefe4,5, Cora E. Lewis4, Stephen Sidney6, and
Stephen B. Hulley1
1
Department of Epidemiology and Biostatistics, University of California, San Francisco, San Francisco, CA.
Division of General Internal Medicine, University of California, San Francisco, San Francisco, CA.
3 Division of Cardiology, University of California, San Francisco, San Francisco, CA.
4 Division of Preventive Medicine, Department of Medicine, University of Alabama at Birmingham, Birmingham, AL.
5 Veterans Affairs Medical Center, Birmingham, AL.
6 Division of Research, Kaiser Permanente, Oakland, CA.
2
Received for publication July 16, 2004; accepted for publication September 20, 2004.
It is unclear to what extent the apparently beneficial cardiovascular effects of moderate alcohol consumption
are mediated by protection against atherosclerosis. Alcohol consumption, coronary heart disease risk factors,
and coronary calcification (a marker of atherosclerosis) were measured during 15 years of follow-up in the
Coronary Artery Risk Development in Young Adults (CARDIA) Study (1985–2001). Among 3,037 participants
aged 33–45 years after follow-up (55% women, 45% Black), the prevalence of coronary calcification was 8% for
consumption of 0 drinks/week (n = 1,435), 9% for 1–6 drinks/week (n = 1,023), 13% for 7–13 drinks/week (n =
341), and 19% for ≥14 drinks/week (n = 238) (p < 0.001 for trend). Calcification was also more common among
binge drinkers (odds ratio = 2.1, 95% confidence interval: 1.6, 2.7). These associations persisted after adjustment
for potential confounders (age, gender/ethnicity, income, physical activity, family history, body mass index,
smoking) and intermediary factors (lipids, blood pressure, glucose, C-reactive protein, and fibrinogen).
Stratification showed the dose-response relation most clearly in Black men; only heavier alcohol consumption
(≥14 drinks/week) was associated with coronary calcification in other race/sex subgroups. These surprising
findings suggest the presence of proatherogenic effects of alcohol in young adults, especially Black men, which
may counterbalance high density lipoprotein cholesterol elevation and other possible benefits of alcohol
consumption.
alcohol drinking; calcification, physiologic; calcium; coronary arteriosclerosis; coronary disease; ethanol; heart
diseases
Abbreviations: CARDIA, Coronary Artery Risk Development in Young Adults; CI, confidence interval; HU, Hounsfield units;
NHANES, National Health and Nutrition Examination Survey; OR, odds ratio.
Nearly half of American adults drink alcohol (1), and
about one third will die from heart disease, the leading cause
of death in the United States (2). Despite its public health
importance and a large body of relevant research, the relation
between alcohol consumption and coronary heart disease
remains puzzling and complex. Among drinkers who do not
binge, there appears to be a J- or U-shaped relation between
alcohol consumption and coronary heart disease, with the
lowest mortality rates being seen among light-to-moderate
drinkers and higher rates among abstainers and heavy
drinkers (3–5). “Binge” drinking (consumption of five or
more drinks on any given occasion) appears to negate the
benefits of moderate alcohol consumption, leading to higher
rates of coronary heart disease in most (6–10) (though not all
(11, 12)) studies. Ethnicity (13), gender (3–5), and type of
alcoholic beverage (14) also influence the relation between
alcohol consumption and coronary heart disease in some
studies.
Correspondence to Dr. Mark Pletcher, 500 Parnassus Avenue, MU 420W, Box 0560, San Francisco, CA 94143-0560 (e-mail:
[email protected]).
423
Am J Epidemiol 2005;161:423–433
424 Pletcher et al.
Alcohol may affect coronary heart disease through a
variety of mechanisms. Moderate consumption of alcohol,
for example, has favorable effects on levels of lipids (particularly high density lipoprotein cholesterol) (15), clotting and
platelet aggregation (15), systemic inflammation (16),
endothelial function (17, 18), and resistance of myocytes to
ischemic injury (19), and it has mixed effects on glucose
tolerance (20–22); heavier alcohol consumption also has
detrimental effects on blood pressure (23, 24). Several basic
questions remain unanswered, however: Does some combination of these mechanisms explain the J-shaped relation
between regular alcohol consumption and clinical coronary
heart disease, or are unidentified mechanisms or noncausal
associations in play? Why should moderate drinkers who
occasionally binge on alcohol not also show lower rates of
coronary heart disease than abstainers? If the observed
associations are causal, at what stage in the development of
clinical coronary heart disease—atherogenesis, plaque
destabilization, platelet aggregation and clot formation, or
ischemic injury and cell death—do moderate alcohol
consumption and bingeing exert their effects?
To address these questions, we sought to isolate the relation between alcohol and the development of coronary
atherosclerosis. This early pathophysiologic event in the
process leading to clinical coronary heart disease may be
detected by measuring coronary calcium, a common component of atherosclerotic plaques that occurs only in the
context of atherosclerosis (25) and strongly predicts future
coronary heart disease events (26). Therefore, we analyzed
the relation between usual alcohol consumption, binge
drinking, and coronary calcification among 33- to 45-yearold Black and White women and men participating in the
Coronary Artery Risk Development in Young Adults
(CARDIA) Study.
MATERIALS AND METHODS
Study design
The CARDIA Study is a longitudinal study of risk factors
for coronary artery disease in a cohort of Black and White
women and men (n = 5,115) aged 18–30 years who were
healthy at the time of enrollment in 1985–1986. With the
informed consent of participants and the approval of institutional review boards at each site, participants have undergone examination in six cycles to date, including a baseline
examination and follow-up examinations at years 2, 5, 7, 10,
and 15, with 74 percent retention at year 15. Details on the
study design, recruitment, and procedures have been
published elsewhere (27, 28). For the purposes of this investigation, we measured the association between alcohol
consumption patterns and the prevalence of coronary calcification at year 15.
beer, and liquor they usually consumed per week. Assuming
that one drink of beer, wine, or liquor contains 16.7 ml, 17.0
ml, or 19.1 ml of ethanol, respectively (per CARDIA
protocol), we estimated total ethanol consumption per week
in milliliters of ethanol and divided it by 17.24 ml of ethanol
per average drink to estimate the usual number of drinks per
week that each participant reported at year 15. Information
on binge drinking was elicited separately by asking participants how many times during the past 30 days they had
consumed five or more drinks on any one occasion.
Using data from past CARDIA examinations, we identified lifetime abstainers, past drinkers, and occasional
drinkers (persons reporting some consumption in the past
year but zero drinks in a typical week) and measured usual
consumption at the baseline CARDIA examination. We also
estimated lifetime alcohol consumption in “drink-years,”
defining 1 drink-year as the amount of alcohol consumed in
1 year by a person consuming 1 drink/day (365 days/year ×
17.24 ml of alcohol/day = 6,293 ml of alcohol). Finally, we
categorized current alcohol drinkers by beverage preference.
Participants who reported more drinks of wine each week
than beer and liquor combined were classified as preferring
wine; participants preferring beer and preferring liquor were
similarly identified; and participants not meeting these
criteria were labeled as having no preference.
Measurement of coronary calcification
Consenting CARDIA participants reporting for their year
15 examination underwent computed tomography scanning
with either an Imatron C-150 electron beam scanner (GE
Imatron, San Francisco, California), a GE Lightspeed multidetector scanner (General Electric Company, Fairfield,
Connecticut), or a Siemens VZ multidetector scanner
(Siemens AG, Munich, Germany). A committee composed of
cardiologists, radiologists, and a physicist developed a scanning protocol for standardizing scan acquisition across these
slightly different technologies; the protocol included two
electrocardiogram-gated scans 2.5–3 mm thick that were
completed within 100–520 ms on each participant, using a
hydroxyapatite phantom to allow standardization of image
brightness. With the help of specialized image processing
software, cardiovascular radiologists blinded as to participant
characteristics identified the presence of coronary calcification in each scan. An expert investigator reviewed and adjudicated all discordant scan pairs. A total coronary calcium
score was calculated for each scan by multiplying the area of
the focus by a coefficient ranging between 1 and 4 based on
the peak density in the focus (1 = 131–200 Hounsfield units
(HU), 2 = 201–300 HU, 3 = 301–400 HU, and 4 = ≥401 HU),
according to the method described by Agatston et al. (29). All
readers were blinded to participant characteristics, as well as
to image data from the other paired scan. Both between- and
within-reader reproducibility were high (30).
Measurement of alcohol consumption
Alcohol consumption was measured during each of the six
CARDIA examinations. Participants were asked, “Did you
drink any alcoholic beverages in the past year?” and three
follow-up questions regarding how many drinks of wine,
Measurement of other covariates
Gender, ethnicity, and date of birth were recorded at baseline. A positive family history of coronary artery disease was
defined as a heart attack before age 60 years in either of the
Am J Epidemiol 2005;161:423–433
Alcohol Consumption and Early Coronary Calcification 425
subject’s parents. Educational grade attained, family
income, and level of physical activity were measured by selfreport at the year 15 examination. Lifetime exposure to
tobacco in pack-years was estimated on the basis of selfreported number of cigarettes smoked per day at each
CARDIA examination. Glucose intolerance was defined as
the use of diabetes mellitus medication or a fasting blood
glucose level greater than 110 mg/dl. Body mass index
(weight (kg)/height (m)2), systolic and diastolic blood pressure, and plasma levels of low density lipoprotein cholesterol, high density lipoprotein cholesterol, and triglycerides
were measured during the year 15 examination according to
CARDIA protocol (31). Plasma C-reactive protein levels
were measured during the year 15 examination using the
high-sensitivity assay manufactured by Calbiochem (EMD
Biosciences, Inc., Darmstadt, Germany). Plasma fibrinogen
levels were measured during the year 5 examination by
Medlantic Laboratories (Hyattsville, Maryland). Duplicate
serum analyses were carried out on 10 percent of participants
for quality control.
Statistical analysis
We categorized participants by their usual alcohol
consumption (0, 1–6, 7–13, or ≥14 drinks/week) in order not
to obscure a nonlinear (J- or U-shaped) relation. Only 127
participants usually consumed 21 or more drinks per week,
so we did not analyze them separately. We compared the
prevalence of alcohol consumption and the prevalence of
binge drinking according to category of participant characteristics using χ2 tests, testing for trend where appropriate.
We used nonparametric tests (Kruskal-Wallis test, or
Cuzick’s extension to the Wilcoxon rank-sum test to test for
trend where appropriate) to compare median alcohol
consumption by participant characteristics among drinkers.
To assess the association between alcohol consumption and
potential mediators of coronary heart disease, we used linear
regression (and logistic regression for proportion with
glucose intolerance), log-transforming data on triglycerides
and C-reactive protein; we tested for trend with a linear
contrast. When assessing the age-adjusted relation between
different measures of alcohol consumption and the prevalence of coronary calcification, we used logistic regression,
testing for trend where appropriate with a linear contrast. We
tested interactions between usual alcohol consumption and
ethnicity, beverage preference, and binge drinking by simultaneously testing interaction terms between alcohol indicator
variables and the characteristic of interest.
We used logistic regression to model the association of
alcohol consumption with coronary calcification. We examined this association in stages: first without adjustment, then
adjusting for age and other potential confounders, and then
additionally adjusting for potential mediators of coronary
disease. We considered age, gender, ethnicity, income,
education, self-reported physical activity, body mass index,
family history of coronary artery disease, and smoking to be
potential confounders of the relation between alcohol
consumption and coronary calcification, thinking them
unlikely to mediate etiologic pathways of interest. We
considered lipids, blood pressure, C-reactive protein, fibrinAm J Epidemiol 2005;161:423–433
ogen, and the presence of glucose intolerance to be potential
intermediary factors. For all multivariable regression analyses, we used a backwards stepwise variable selection
process, eliminating the variable with the highest p value
until all variables in the model were associated with the
outcome at p < 0.1. Participants missing any covariate data
were excluded from all multivariable analyses.
We performed several sensitivity analyses to assess the
impact of four alternate modeling decisions: 1) excluding
past drinkers, 2) keeping all variables in the model (instead
of using a backwards stepwise elimination process), 3) limiting the analysis to nonsmokers, and 4) further categorizing
nonzero coronary calcium scores (instead of dichotomizing
as any vs. none) and using ordinal logistic regression. All
analyses were performed using Stata 7.0 (Stata Corporation,
College Station, Texas) and verified by the CARDIA Data
Coordinating Center. All p values are two-sided.
RESULTS
Study participants and their alcohol consumption
patterns
Of 3,042 CARDIA participants who underwent computed
tomography scanning at the year 15 follow-up examination,
conducted in 2000–2001, 3,037 also reported their usual
alcohol consumption at this examination. Participants were
aged 33–45 years; 55 percent were women, and 45 percent
were Black. In comparison with the original 5,115 persons in
the baseline CARDIA examination (1985–1986) who did
not participate in follow-up at year 15, our sample was
slightly older (mean age of 25.2 years vs. 24.3 years), was
more often Caucasian (55 percent vs. 39 percent), and had
the same proportion of females (55 percent vs. 54 percent).
Among the half of participants who consumed alcohol on
a regular basis at year 15, the median level of consumption
was 5 drinks/week (interquartile range: 2, 10). Nearly one
fourth of participants (23 percent) reported having engaged
in binge drinking on at least one occasion during the past 30
days. Consumption patterns were strongly related to many
participant characteristics (table 1), and frequency of binge
drinking was highly correlated with usual alcohol consumption (figure 1; Spearman’s rank correlation coefficient = 0.57,
p < 0.001).
Alcohol consumption and potential mediators of
coronary artery disease
In unadjusted analyses, increasing alcohol consumption
was associated with increasing levels of high density lipoprotein cholesterol and systolic and diastolic blood pressure
and decreasing levels of C-reactive protein and fibrinogen.
Alcohol consumption was not significantly associated with
low density lipoprotein cholesterol, triglycerides, or glucose
intolerance (table 2).
Alcohol consumption and coronary calcification
Participants who consumed more alcohol were more likely
to have coronary calcification. The prevalence of coronary
426 Pletcher et al.
TABLE 1. Characteristics of participants according to pattern of alcohol consumption, Coronary Artery
Risk Development in Young Adults (CARDIA) Study, 2000–2001†
Characteristic
No. of
subjects
%
Median consumption
among drinkers
Any current
alcohol
consumption
(%)
No. of
drinks/week
Interquartile
range
(25%, 75%)
Any binge
drinking (%)
Age (years)
32–38
959
32
47*
4**
2, 9
22
39–42
1,037
34
58
5
2, 10
25
43–47
1,041
34
53
6
2, 11
21
Black women
799
26
34***
4***
2, 7
13***
Black men
575
19
54
7
3, 15
33
White women
859
28
55
4
2, 8
15
White men
804
26
68
6
3, 11
34
Gender/ethnic group
Educational level
High school or less
653
22
48***
7**
3, 15
31***
1,738
57
51
5
2, 10
21
641
21
62
5
2, 9
18
<$25,000
439
15
46***
7**
3, 15
26
$25,000–$49,999
743
25
47
5
2, 11
24
$50,000–$99,999
1,126
38
50
4
2, 8
19
693
23
67
5
3, 10
25
Some college education
Some graduate-level education
Annual income
≥$100,000
Cigarette smoking (pack-years)
0
1,675
55
46***
4**
2, 8
15***
1–10
891
29
60
6
3, 11
28
11–20
276
9
64
8
3, 17
40
≥21
195
6
57
10
4, 19
38
185
6
41***
5
2, 10
18***
Self-rated physical activity level
1 (physically inactive)
2
506
17
45
4
2, 9
19
1,365
45
52
5
2, 10
21
4
523
17
62
5
2, 10
27
5 (very active)
453
15
59
6
3, 11
30
3 (moderately active)
Body mass index‡
<25
25–30
≥31
960
32
60***
5
2, 10
21
1,122
37
56
6
3, 11
28
944
31
41
4
2, 10
19
360
12
53
5
2, 10
23
2,677
88
51
5
2, 10
23
Family history of premature coronary
heart disease
Yes
No
* p < 0.05; **p < 0.01; ***p < 0.001.
† p values refer to the comparison of proportions (any alcohol consumption and any binge drinking) or medians
(median consumption among drinkers) between groups of participants defined by each characteristic. Tests of
trend (χ2 or nonparametric) were used for age, education, income, cigarette smoking, physical activity, and body
mass index.
‡ Weight (kg)/height (m)2.
Am J Epidemiol 2005;161:423–433
Alcohol Consumption and Early Coronary Calcification 427
FIGURE 1. Relation of frequency of binge drinking to pattern of
usual alcohol consumption, Coronary Artery Risk Development in
Young Adults (CARDIA) Study, 2000–2001. Frequency of binge
drinking, defined as consumption of five or more alcoholic beverages
on one occasion in the past 30 days, was strongly associated with
the number of drinks consumed weekly (Spearman’s rank correlation
coefficient = 0.57, p < 0.001). Information on usual alcohol consumption was elicited by asking, “How many drinks of (wine, beer, liquor)
per week do you usually drink?” Therefore, a participant claiming to
consume 0 drinks/week in a typical week might still report an episode
of binge drinking.
calcification was 8 percent for participants who consumed 0
drinks/week, 9 percent for 1–6 drinks/week, 13 percent for
7–13 drinks/week, and 19 percent for ≥14 drinks/week (p <
0.001 for trend). Age-adjustment had little effect on this
trend (table 3). Coronary calcification was also more
common among binge drinkers (odds ratio (OR) = 2.1, 95
percent confidence interval (CI): 1.6, 2.7; age-adjusted OR =
2.2, 95 percent CI: 1.7, 2.9). Each measure of alcohol exposure—including usual consumption at year 15 and year 0,
consumption pattern, total lifetime exposure to alcohol, and
frequency of binge drinking—was strongly associated with
the age-adjusted prevalence of coronary calcification (table
3).
The association between usual alcohol consumption and
coronary calcification appeared to be strongest among Black
men (p < 0.001 for trend), though statistical evidence that the
relation differed among the different gender/ethnic groups
was weak (p = 0.36 for the interaction). This association
remained strong among Black men who denied binge
drinking (n = 383; for 1–6 drinks/week, OR = 1.7; for 7–13
drinks/week, OR = 1.9; and for ≥14 drinks/week, OR = 2.7
(compared with 0 drinks/week)), though it was no longer
statistically significant (p = 0.11) (figure 2).
There was a faint suggestion of a J-shaped relation
between usual alcohol consumption and coronary calcification among all gender/ethnic groups besides Black men, but
this effect was not statistically significant, and the nadir of
the “J” occurred at relatively low levels of consumption (1–
6 drinks/week) (figure 2).
The association between alcohol consumption and coronary calcification did not differ by alcoholic beverage preference (figure 2).
Multivariable analysis
Usual alcohol consumption and binge drinking remained
strong predictors of coronary calcification in multivariable
analyses (table 4). While many potentially strong confounders
were included in the final logistic models—including age,
gender/ethnicity, tobacco exposure, body mass index, family
history of coronary heart disease, and income—the associa-
TABLE 2. Association of alcohol intake with potential mediators of coronary artery disease, Coronary Artery Risk Development in
Young Adults (CARDIA) Study, 2000–2001
Potential mediator
Alcohol consumption
Mean LDL*
Mean HDL*
(drinks/week)
cholesterol level cholesterol level
(mmol/liter†)
(mmol/liter†)
Mean triglyceride level
(mmol/liter†)
Mean systolic
blood pressure
(mmHg)
Mean diastolic
Prevalence
Mean C-reactive
blood pressure
of glucose
protein level
(mmHg)
intolerance‡ (%)
(µg/ml)
Mean fibrinogen level
(mmol/liter†)
0
3.0 (0.02)§
1.26 (0.01)
1.2 (0.02)
113 (0.4)
75 (0.3)
5.9 (0.6)
2.2 (0.07)
8.0 (0.05)
1–6
2.9 (0.02)
1.31 (0.01)
1.2 (0.04)
112 (0.4)
73 (0.3)
3.0 (0.5)
1.7 (0.07)
7.5 (0.05)
7–13
3.0 (0.05)
1.39 (0.02)
1.3 (0.07)
115 (0.8)
75 (0.6)
2.3 (0.8)
1.9 (0.2)
7.3 (0.09)
≥14
2.9 (0.06)
1.46 (0.03)
1.2 (0.06)
118 (1.0)
77 (0.8)
4.2 (1.3)
1.8 (0.1)
7.4 (0.11)
0.32
<0.001
0.10
<0.001
<0.001
0.22
0.019
<0.001
p for trend¶
* LDL, low density lipoprotein; HDL, high density lipoprotein.
† Conversion factors: LDL cholesterol and HDL cholesterol, 1 mg/dl = 0.0259 mmol/liter; triglycerides, 1 mg/dl = 0.0113 mmol/liter; fibrinogen,
1 mg/dl = 0.0294 mmol/liter.
‡ Glucose intolerance was defined as self-reported use of diabetes medication or a fasting blood glucose level greater than 110 mg/dl (see
Materials and Methods).
§ Numbers in parentheses, standard error.
¶ We used linear regression (logistic regression for glucose intolerance) to model the unadjusted association between alcohol consumption
and each outcome. Data on triglyceride and C-reactive protein levels were log-transformed. A linear contrast was used to test for trend.
Specifically, we tested the null hypothesis that –β(1–6 drinks/week) + β(7–13 drinks/week) + [3 × β(≥14 drinks/week)] = 0.
Am J Epidemiol 2005;161:423–433
428 Pletcher et al.
TABLE 3. Age-adjusted prevalence of coronary calcification according to different measures of alcohol
consumption, Coronary Artery Risk Development in Young Adults (CARDIA) Study, 2000–2001
Coronary calcification
No. of
subjects
%
0
1,435
1–6
1,023
7–13
≥14
Measure of alcohol consumption
Prevalence
(%)
p value*
47
7
<0.001
34
8
341
11
11
238
8
17
Usual alcohol consumption (drinks/week†)
Alcohol consumption pattern‡
Never (lifetime abstainer)
109
4
4
Past drinker, no drinking currently
518
17
8
Occasional (<1 drink/week, no bingeing)
766
25
6
Binge drinker (≥5 drinks on one occasion)
691
23
14
Any regular, nonbinge drinking
953
31
7
543
18
6
<0.001
Lifetime alcohol consumption (drink-years§)
0
0.1–10
1,300
43
7
10–19.9
522
17
9
20–29.9
283
9
9
30–39.9
135
4
8
40–49.9
93
3
24
161
5
19
0
2,346
77
7
1
243
8
11
2–4
284
9
15
≥5
164
5
15
0
1,163
38
6
1–6
1,211
40
8
7–13
389
13
11
≥2
260
9
16
≥50
<0.001
No. of episodes of binge drinking in past 30 days
<0.001
Alcohol consumption at year 0¶ (drinks/week)
<0.001
* A linear contrast was used to test for trend. For example, for drinks/week, we tested the null hypothesis that
–β(1–6 drinks/week) + β(7–13 drinks/week) + [3 × β(≥14 drinks/week)] = 0. For “Alcohol consumption pattern,” we
used a Wald test to assess logistic regression coefficients (not a test of trend).
† On average, one drink contains 17.24 ml of ethanol.
‡ Persons who answered “no” to the questions “Have you had any alcoholic beverage in the past year?” and “Did
you drink alcohol more than once or twice in the past?” at each CARDIA examination were considered lifetime
abstainers. Persons who answered “no” to the question “Have you had any alcoholic beverages in the past year?”
but were not lifetime abstainers were considered “past” drinkers. Persons who had had an alcoholic drink in the
past year but did not drink any alcohol in a typical week and did not binge were considered “occasional” drinkers.
Persons reporting ≥5 drinks on one or more occasions during the previous month were considered “binge
drinkers.”
§ A drink-year was defined as the total amount of ethanol consumed by a person who had one alcoholic drink
per day for 1 year (1 drink-year = 17.24 ml of ethanol/drink × 1 drink/day × 365 days/year = 6,292.6 ml of ethanol).
¶ Based on the 3,023 participants who answered alcohol-related questions during the baseline CARDIA
examination in 1985–1986 and subsequently had coronary calcification measured at year 15.
tion between alcohol variables and coronary calcification
persisted (table 4). Higher systolic blood pressure, higher
low density lipoprotein cholesterol, and lower high density
lipoprotein cholesterol were also strong independent predictors of coronary calcification; inclusion of these potential
mediators had no apparent effect on associations between
Am J Epidemiol 2005;161:423–433
Alcohol Consumption and Early Coronary Calcification 429
FIGURE 2. Prevalence of coronary calcification according to usual alcohol consumption in different subgroups of participants, Coronary Artery
Risk Development in Young Adults (CARDIA) Study, 2000–2001. Prevalence estimates were adjusted for age. Participants were categorized as
preferring a specific beverage type if they reported consuming more of that type of beverage than of the other two types combined. Binge drinking
was defined as consumption of five or more alcoholic beverages on one occasion. The p values for trends within each subgroup are shown in the
figure. The overall p values from evaluation of interactions were p = 0.36 for the interaction with gender/ethnicity, p = 0.64 for the interaction with
beverage preference, and p = 0.73 for the interaction with bingeing pattern.
alcohol and coronary calcification (table 4). Simultaneous
adjustment for both usual consumption and binge drinking
weakened each individual alcohol-calcification association,
but the overall contribution of alcohol variables to the model
remained significant (table 4, second section). CategorizaAm J Epidemiol 2005;161:423–433
tion of all participants according to both binge frequency and
usual consumption suggested that increases in either binge
frequency or usual alcohol consumption were sufficient to
induce an association with coronary calcification (table 4,
third section).
430 Pletcher et al.
TABLE 4. Adjusted odds of coronary calcification among persons with different alcohol consumption patterns, Coronary Artery Risk
Development in Young Adults (CARDIA) Study, 2000–2001
Prevalence of coronary calcification
Alcohol consumption pattern
(among all participants)
No. of
subjects
Adjusted for potential
confounders*
Unadjusted
%
95% CI‡
OR‡
95% CI
OR
95% CI
Adjusted for potential
confounders* and
potential mediators†
OR
95% CI
Separate analyses of overall consumption
and binge drinking
Usual alcohol consumption (drinks/week)
2,723§
0
1,257
7
6, 9
1¶
1–6
933
9
7, 11
1.2
0.9, 1.7
1.1
0.8, 1.6
1.1
0.8, 1.6
7–13
315
13
10, 18
1.9
1.3, 2.9
1.6
1.0, 2.4
1.5
1.0, 2.3
≥14
218
19
14, 25
2.9
2.0, 4.4
1.9
1.2, 2.9
2.0
1.3, 3.2
p value#
Bingeing
<0.001
1¶
<0.001
1¶
0.002
0.002
2,723
No bingeing
2106
8
7, 9
1¶
Bingeing
617
16
13, 19
2.2
p value
<0.001
<0.001
—**
1¶
1¶
1.7, 2.9
1.6
1¶
1.2, 2.2
0.001
1.7
1.2, 2.3
0.001
Simultaneous adjustment for overall
consumption and binge
drinking**
Usual alcohol consumption (drinks/week)
0
2,723
—**
1¶
1¶
1–6
1.1
0.8, 1.5
1.1
0.7, 1.5
1.1
0.8, 1.5
7–13
1.5
1.0, 2.3
1.3
0.8, 2.1
1.3
0.8, 2.1
≥14
1.9
1.1, 3.1
1.4
0.8, 2.4
1.5
0.9, 2.7
p value#
0.007
0.146
0.092
Bingeing
No bingeing
1¶
Bingeing
1.7
1¶
1.2, 2.4
1.4
1¶
1.0, 2.0
1.4
p value
0.003
0.078
0.059
Overall p value††
0.003
0.011
0.006
1.0, 2.1
Table continues
Sensitivity analyses
Exclusion of past drinkers (n = 447) from the reference
group and backwards stepwise elimination of variables from
each multivariable model (or lack thereof) had virtually no
effect on the magnitude or statistical significance of our findings. Limiting the analysis to lifetime nonsmokers (n =
1,675) strengthened the linear association between alcohol
consumption and coronary calcification (fully adjusted
ORs = 1.2, 1.8, and 3.3 (compared with nondrinkers); p =
0.002 for trend). When we treated the coronary calcium
score as ordinal instead of dichotomous by categorizing
nonzero coronary calcium scores (0, 1–9, 10–99, and ≥100),
the graded association with usual consumption and binge
consumption remained essentially unchanged. After adjustment for potential confounders and potential mediators using
ordinal logistic regression, odds ratios for having a coronary
calcium score in a higher category were 1.2 (95 percent CI:
0.9, 1.7) for 1–6 drinks/week, 1.6 (95 percent CI: 1.1, 2.5)
for 7–13 drinks/week, and 2.1 (95 percent CI: 1.4, 3.4) for
≥14 drinks/week (in comparison with 0 drinks/week) (p <
0.001 for trend).
DISCUSSION
In this article, we report a direct association between higher
levels of alcohol consumption and coronary calcification.
Our analyses suggested that both usual alcohol consumption
and binge drinking contribute to this association and that
Black men may differ from others in their response to
alcohol. The association remained present after adjustment
for many potential confounders and did not appear to be
mediated by effects of alcohol on lipids, blood pressure, Creactive protein, fibrinogen, or glucose intolerance.
To our knowledge, our study is the first to demonstrate an
association between binge drinking and atherosclerosis of
the coronary arteries (as measured by coronary artery calcification). This association is consistent with previous studies
of binge drinking and coronary heart disease events, most of
which have found higher rates of events among persons who
Am J Epidemiol 2005;161:423–433
Alcohol Consumption and Early Coronary Calcification 431
TABLE 4. Continued
Prevalence of coronary calcification
Alcohol consumption pattern
(among all participants)
No. of
subjects
%
Joint categorization of overall consumption
and binge drinking‡‡
Adjusted for potential
confounders*
Unadjusted
95% CI
OR
95% CI
OR
95% CI
Adjusted for potential
confounders* and
potential mediators†
OR
95% CI
2,723
0–6 drinks/week
0 binges/month
1,923
7
6, 9
1¶
1 binge/month
139
11
6, 17
1.5
0.9, 2.7
1¶
1.3
0.7, 2.3
1¶
1.2
0.7, 2.2
≥2 binges/month
128
14
9, 21
2.1
1.2, 3.5
1.6
0.9, 2.7
1.7
1.0, 2.9
0 binges/month
156
11
6, 17
1.5
0.9, 2.6
1.3
0.8, 2.4
1.3
0.7, 2.3
1 binge/month
57
9
3, 19
1.2
0.5, 3.1
0.9
0.3, 2.3
0.8
0.3, 2.2
≥2 binges/month
102
20
12, 29
3.1
1.8, 5.2
2.4
1.4, 4.2
2.4
1.4, 4.3
0–1 binge/month
50
16
7, 29
2.4
1.1, 5.2
1.8
0.8, 4.0
1.7
0.7, 3.8
≥2 binges/month
168
20
14, 26
3.1
2.0, 4.7
1.9
1.2, 3.0
2.1
1.3, 3.5
7–13 drinks/week
≥14 drinks/week
p value††
<0.001
<0.001
0.019
0.012
* We considered the following to be potential confounders: age, gender/ethnicity, education, income, self-reported physical activity level, family history of
premature coronary heart disease, body mass index, and smoking exposure (in pack-years). In the full model containing both usual alcohol consumption and binge
drinking, the predictors remaining in the model after backwards stepwise elimination were age, gender/ethnic group, tobacco exposure, body mass index, family
history of coronary heart disease, and income. Retaining all potential confounders without stepwise elimination resulted in essentially identical results (see
“Sensitivity analyses” in text).
† We considered the following physiologic parameters to be potential mediators of coronary artery disease: low density lipoprotein cholesterol, high density
lipoprotein cholesterol, triglycerides, systolic and diastolic blood pressure, glucose intolerance, C-reactive protein, and fibrinogen. In the full model containing both
usual alcohol consumption and binge drinking, potential confounders and mediators remaining in the model after backwards stepwise elimination were age, gender/
ethnic group, tobacco exposure, family history of coronary heart disease, income, low density lipoprotein cholesterol, high density lipoprotein cholesterol, and
systolic blood pressure. Retaining all potential confounders and mediators without stepwise elimination resulted in essentially identical results (see “Sensitivity
analyses” in text).
‡ CI, confidence interval; OR, odds ratio.
§ Of the 3,037 participants meeting the entrance criteria, data on all potential confounders and mediators were available for 2,723 (90%).
¶ Reference category.
# A linear contrast was used to test for trend. Specifically, we tested the null hypothesis that –β(1–6 drinks/week) + β(7–13 drinks/week) + [3 × β(≥14 drinks/
week)] = 0.
** In the simultaneous adjustment strategy, both usual alcohol consumption and binge drinking were included in the same model. For this reason, no prevalence
estimates or numbers of subjects are given.
†† p value computed using a simultaneous Wald test of all alcohol-related regression coefficients (not a test of trend).
‡‡ In the joint exposure categorization strategy, we categorized all participants according to both usual alcohol consumption and frequency of binge drinking.
Some categories were collapsed because of small cell sizes, such as persons drinking ≥14 drinks/week who reported no bingeing (n = 27) and persons who reported
having one binge episode in the past month (n = 23).
binge (9, 10). Our findings also support previous research
suggesting that binge drinking is associated with coronary
heart disease at least partly through its association with
atherosclerosis (32) (demonstrated by ultrasonography of
carotid arteries) rather than merely through its association
with sudden death from arrhythmia or other nonatherosclerotic mechanisms. Dysregulation of inflammatory cytokines
associated with the “hangover” after a binge (33) is a potential explanation for this relation.
In contrast with much previous research on alcohol and
clinical coronary heart disease (5), we did not find evidence
of protection against atherosclerosis from light-to-moderate
regular alcohol consumption. At higher levels of alcohol
consumption (≥14 drinks/week), the risk of coronary calcification was elevated; among Black men, even moderate
drinking was associated with increased risk. While binge
drinking among otherwise-moderate drinkers offers a partial
Am J Epidemiol 2005;161:423–433
explanation, we detected no beneficial association with
alcohol consumption even after excluding binge drinkers.
There are several potential explanations for these findings.
One possibility is that the beneficial associations between
alcohol and clinical coronary heart disease events seen in
previous studies are not generally mediated through an
atherosclerotic pathway, and therefore would not be detected
by measuring coronary calcification. This explanation is
consistent with previous research showing beneficial effects
of alcohol on platelet aggregation (15), endothelial function
(17, 18), inflammation (16), and myocyte resistance to
ischemic injury (19). It is also consistent with three previous
studies that failed to show a protective association between
moderate alcohol consumption and carotid atherosclerosis
(34–36), although two other studies did show such an association (37, 38). Investigators in one prior study who
commented only briefly on alcohol use and coronary calcifi-
432 Pletcher et al.
cation in their much older population also found no beneficial association with alcohol consumption (39).
Perhaps more likely is that the increase in high density
lipoprotein cholesterol caused by alcohol, which accounts
for approximately 50 percent of alcohol-mediated coronary
heart disease protection (40), and other potential benefits
derived from alcohol in terms of atherosclerosis protection
are counterbalanced by other harmful effects from alcohol in
young persons. Even among older adults, the benefits of
alcohol appear to be outweighed by harmful effects at high
levels of consumption (the upswing of the J-shaped curve
(5)). These harmful effects may be more important in young
adults as atherosclerosis develops and perhaps most important among Black men, in whom the dose-response association between alcohol and coronary calcification appeared to
be strongest. The mechanisms by which alcohol may induce
atherosclerosis are unclear, but they may involve inflammation, low density lipoprotein cholesterol oxidation, acetaldehyde produced as a metabolic byproduct, or increased shear
stress and flow turbulence associated with higher blood pressure or adrenergic stress (32).
Finally, it is possible that the coronary calcium we found
in atherosclerotic plaques among these young adults may in
fact represent plaque stabilization. Such a possibility might
imply that lower coronary heart disease rates among
moderate drinkers seen in other studies could be caused by a
plaque-stabilizing effect of alcohol. However, while a given
calcified plaque may be more stable on average than a
noncalcified plaque, the presence of calcified plaques generally indicates a larger overall atherosclerotic plaque burden
(41) and is associated with higher, not lower, rates of coronary heart disease events (26).
Our finding that the linear trend of increasing coronary
calcification with increasing alcohol consumption was strongest among Black men is notable. Blacks have historically
been underrepresented in large studies of cardiovascular
disease, including those examining the link between alcohol
and coronary heart disease. One previous analysis using data
from the First National Health and Nutrition Examination
Survey (NHANES I) Epidemiologic Follow-up Study did
focus on Blacks and found higher rates of all-cause mortality
with higher alcohol intake (13) and no apparent beneficial
association with moderate consumption. The authors
hypothesized that the lack of benefit might be due to the
binge pattern of alcohol consumption noted previously (but
not in their study) to be more common among Blacks (42). In
our study, there were no ethnic differences in binge drinking,
and the strong dose-response relation between alcohol
consumption and coronary calcification among Black men
appeared to persist even after exclusion of binge drinkers.
The associations we observed persisted despite adjustment
for many correlates of alcohol consumption and binge
drinking. While this does not exclude confounding as an
explanation for our findings, the extensive data on potential
confounders are a strength of our analysis in comparison with
many other studies of alcohol and coronary heart disease.
Limitations of our study include uncertainty about the degree
to which findings in our young-to-middle-aged population
will apply to older men and women and the suboptimal
sample size for subgroup and interaction analysis. We also
lacked detailed data on drinking frequency, a potentially
important factor in characterizing alcohol intake (12). A
strength of our study is the longitudinal assessment of
alcohol consumption patterns available in CARDIA. The
consistency with which different assessments of exposure
over 15 years are associated with coronary calcification
strengthens our inferences, and the estimate of lifetime
alcohol exposure in “drink-years” (a new measure analogous
to “pack-years” in the smoking literature) may prove useful
for examining other cumulative effects of alcohol exposure.
In conclusion, our findings add to the emerging evidence
that binge drinking is associated with higher rates of cardiovascular disease and strengthen the rationale for recommending that those who drink alcohol do so in moderation.
Surprisingly, our findings also raise the possibility that even
moderate alcohol consumption could have proatherogenic
effects among young adults. This result appears to be driven
by the strong dose-response association between alcohol
consumption and coronary calcification that we observed
among Black men. The recent NHANES I Epidemiologic
Follow-up Study report (13) showing higher mortality for
Blacks at all levels of alcohol consumption in comparison
with abstinence supports this potentially important finding
and emphasizes the importance of studying the effects of
alcohol in a variety of ethnic and cultural groups.
ACKNOWLEDGMENTS
The CARDIA Study is supported by contracts N01-HC48047, N01-HC-48048, N01-HC-48049, N01-HC-48050,
and N01-HC-95095 from the National Heart, Lung, and
Blood Institute. Dr. Mark Pletcher was supported by funds
from the Health Resources and Services Administration, US
Public Health Service (grant D14 HP00178).
The authors acknowledge Dr. Mark Pereira, Dr. David
Jacobs, Feng Lin, and Heather McCreath for their generous
assistance.
REFERENCES
1. National Institute on Alcohol Abuse and Alcoholism. Tenth
special report to the U.S. Congress on alcohol and health: highlights from current research. Rockville, MD: National Institute
on Alcohol Abuse and Alcoholism, 2000. (World Wide Web
URL: http://www.niaaa.nih.gov/publications/10report/
intro.pdf).
2. Anderson RN. Deaths: leading causes for 2000. Natl Vital Stat
Rep 2002;50:1–85.
3. Holman CD, English DR, Milne E, et al. Meta-analysis of alcohol and all-cause mortality: a validation of NHMRC recommendations. Med J Aust 1996;164:141–5.
4. White IR. The level of alcohol consumption at which all-cause
mortality is least. J Clin Epidemiol 1999;52:967–75.
5. Corrao G, Rubbiati L, Bagnardi V, et al. Alcohol and coronary
heart disease: a meta-analysis. Addiction 2000;95:1505–23.
6. Kauhanen J, Kaplan GA, Goldberg DE, et al. Beer binging and
mortality: results from the Kuopio Ischaemic Heart Disease
Risk Factor Study, a prospective population based study. BMJ
1997;315:846–51.
Am J Epidemiol 2005;161:423–433
Alcohol Consumption and Early Coronary Calcification 433
7. Kauhanen J, Kaplan GA, Goldberg DD, et al. Frequent hangovers and cardiovascular mortality in middle-aged men.
Epidemiology 1997;8:310–14.
8. McElduff P, Dobson AJ. How much alcohol and how often?
Population based case-control study of alcohol consumption
and risk of a major coronary event. BMJ 1997;314:1159–64.
9. Rehm J, Greenfield TK, Rogers JD. Average volume of alcohol
consumption, patterns of drinking, and all-cause mortality:
results from the US National Alcohol Survey. Am J Epidemiol
2001;153:64–71.
10. Murray RP, Connett JE, Tyas SL, et al. Alcohol volume, drinking pattern, and cardiovascular disease morbidity and mortality:
is there a U-shaped function? Am J Epidemiol 2002;155:242–8.
11. Malyutina S, Bobak M, Kurilovitch S, et al. Relation between
heavy and binge drinking and all-cause and cardiovascular
mortality in Novosibirsk, Russia: a prospective cohort study.
Lancet 2002;360:1448–54.
12. Mukamal KJ, Conigrave KM, Mittleman MA, et al. Roles of
drinking pattern and type of alcohol consumed in coronary
heart disease in men. N Engl J Med 2003;348:109–18.
13. Sempos CT, Rehm J, Wu T, et al. Average volume of alcohol
consumption and all-cause mortality in African Americans: The
NHEFS Cohort. Alcohol Clin Exp Res 2003;27:88–92.
14. Rimm EB, Klatsky A, Grobbee D, et al. Review of moderate
alcohol consumption and reduced risk of coronary heart disease: is the effect due to beer, wine, or spirits? BMJ 1996;312:
731–6.
15. Rimm EB, Williams P, Fosher K, et al. Moderate alcohol intake
and lower risk of coronary heart disease: meta-analysis of
effects on lipids and haemostatic factors. BMJ 1999;319:
1523–8.
16. Imhof A, Froehlich M, Brenner H, et al. Effect of alcohol consumption on systemic markers of inflammation. Lancet 2001;
357:763–7.
17. Teragawa H, Fukuda Y, Matsuda K, et al. Effect of alcohol consumption on endothelial function in men with coronary artery
disease. Atherosclerosis 2002;165:145–52.
18. Sacanella E, Badia E, Nicolas JM, et al. Differential effects of
moderate or heavy alcohol consumption on circulating adhesion molecule levels. Thromb Haemost 2002;88:52–5.
19. Zhou HZ, Karliner JS, Gray MO. Moderate alcohol consumption induces sustained cardiac protection by activating PKCepsilon and Akt. Am J Physiol Heart Circ Physiol 2002;283:
H165–74.
20. Kao WH, Puddey IB, Boland LL, et al. Alcohol consumption
and the risk of type 2 diabetes mellitus: Atherosclerosis Risk in
Communities Study. Am J Epidemiol 2001;154:748–57.
21. Wannamethee SG, Shaper AG, Perry IJ, et al. Alcohol consumption and the incidence of type II diabetes. J Epidemiol
Community Health 2002;56:542–8.
22. Zilkens RR, Puddey IB. Alcohol and type 2 diabetes—another
paradox? J Cardiovasc Risk 2003;10:25–30.
23. Dyer AR, Cutter GR, Liu KQ, et al. Alcohol intake and blood
pressure in young adults: The CARDIA Study. J Clin Epidemiol 1990;43:1–13.
24. Fuchs FD, Chambless LE, Whelton PK, et al. Alcohol consumption and the incidence of hypertension: The Atherosclerosis Risk in Communities Study. Hypertension 2001;37:1242–
50.
25. Blankenhorn D. Coronary arterial calcification: a review. Am J
Am J Epidemiol 2005;161:423–433
Med Sci 1961;242:1–9.
26. Pletcher MJ, Tice JA, Pignone M, et al. Using the coronary
artery calcium score to predict coronary heart disease events: a
systematic review and meta-analysis. Arch Intern Med 2004;
164:1285–92.
27. Friedman GD, Cutter GR, Donahue R, et al. CARDIA: study
design, recruitment, and some characteristics of the examined
subjects. J Clin Epidemiol 1988;41:1105–16.
28. Hughes GH, Cutter GR, Donahue R, et al. Recruitment in the
Coronary Artery Disease Risk Development in Young Adults
(CARDIA) Study. Control Clin Trials 1987;8(suppl):68S–73S.
29. Agatston AS, Janowitz WR, Hildner FJ, et al. Quantification of
coronary artery calcium using ultrafast computed tomography.
J Am Coll Cardiol 1990;15:827–32.
30. Carr JJ, Nelson JC, Wong ND, et al. Measuring calcified coronary plaque with cardiac CT in population-based studies: the
standardized protocol of the Multi-Ethnic Study of Atherosclerosis (MESA) and Coronary Artery Risk Development in
Young Adults (CARDIA). Radiology (in press).
31. Cutter GR, Burke GL, Dyer AR, et al. Cardiovascular risk factors in young adults: the CARDIA baseline monograph.
Control Clin Trials 1991;12(suppl):1S–77S.
32. Kauhanen J, Kaplan GA, Goldberg DE, et al. Pattern of alcohol
drinking and progression of atherosclerosis. Arterioscler
Thromb Vasc Biol 1999;19:3001–6.
33. Wiese JG, Shlipak MG, Browner WS. The alcohol hangover.
Ann Intern Med 2000;132:897–902.
34. Demirovic J, Nabulsi A, Folsom AR, et al. Alcohol consumption and ultrasonographically assessed carotid artery wall thickness and distensibility. The Atherosclerosis Risk in
Communities (ARIC) Study Investigators. Circulation 1993;
88:2787–93.
35. Temelkova-Kurktschiev T, Henkel E, Koehler C, et al. Lightto-moderate alcohol consumption, risk factor profile and early
atherosclerosis: The RIAD Study. Atherosclerosis 2001;156:
239–40.
36. Djousse L, Myers RH, Province MA, et al. Influence of apolipoprotein E, smoking, and alcohol intake on carotid atherosclerosis: National Heart, Lung, and Blood Institute Family Heart
Study. Stroke 2002;33:1357–61.
37. Kiechl S, Willeit J, Rungger G, et al. Alcohol consumption and
atherosclerosis: what is the relation? Prospective results from
the Bruneck Study. Stroke 1998;29:900–7.
38. Vliegenthart R, Geleijnse JM, Hofman A, et al. Alcohol consumption and risk of peripheral arterial disease: The Rotterdam
Study. Am J Epidemiol 2002;155:332–8.
39. Yang T, Doherty TM, Wong ND, et al. Alcohol consumption,
coronary calcium, and coronary heart disease events. Am J Cardiol 1999;84:802–6.
40. Mukamal KJ, Jensen MK, Gronbaek M, et al. Drinking frequency, potential intermediates, and risk of myocardial infarction in men. (Abstract 16). Circulation 2004;109(suppl):8.
41. Rumberger JA, Schwartz RS, Simons DB, et al. Relation of coronary calcium determined by electron beam computed tomography and lumen narrowing determined by autopsy. Am J
Cardiol 1994;73:1169–73.
42. Dawson DA. Beyond black, white and Hispanic: race, ethnic
origin and drinking patterns in the United States. J Subst Abuse
1998;10:321–39.