Brief Reports

Cardiovascular Risk Factors and Confounders Among

Nondrinking and Moderate-Drinking U.S. Adults
Timothy S. Naimi, MD, MPH, David W. Brown, MSPH, MS, Robert D. Brewer, MD, MSPH,
Wayne H. Giles, MD, MS, George Mensah, MD, Mary K. Serdula, MD, MPH, Ali H. Mokdad, PhD,
Daniel W. Hungerford, DrPH, James Lando, MD, MPH, Shapur Naimi, MD, Donna F. Stroup, PhD, MSc

Background: Studies suggest that moderate drinkers have lower cardiovascular disease (CVD) mortality
than nondrinkers and heavy drinkers, but there have been no randomized trials on this
topic. Although most observational studies control for major cardiac risk factors, CVD is
independently associated with other factors that could explain the CVD benefits ascribed
to moderate drinking.
Methods: Data from the 2003 Behavioral Risk Factor Surveillance System, a population-based
telephone survey of U.S. adults, was used to assess the prevalence of CVD risk factors and
potential confounders among moderate drinkers and nondrinkers. Moderate drinkers
were defined as men who drank an average of two drinks per day or fewer, or women who
drank one drink or fewer per day.
Results: After adjusting for age and gender, nondrinkers were more likely to have characteristics
associated with increased CVD mortality in terms of demographic factors, social factors,
behavioral factors, access to health care, and health-related conditions. Of the 30
CVD-associated factors or groups of factors that we assessed, 27 (90%) were significantly
more prevalent among nondrinkers. Among factors with multiple categories (e.g., body
weight), those in higher-risk groups were progressively more likely to be nondrinkers.
Removing those with poor health status or a history of CVD did not affect the results.
Conclusions: These findings suggest that some or all of the apparent protective effect of moderate
alcohol consumption on CVD may be due to residual or unmeasured confounding. Given
their limitations, nonrandomized studies about the health effects of moderate drinking
should be interpreted with caution, particularly since excessive alcohol consumption is a
leading health hazard in the United States.
(Am J Prev Med 2005;28(4):369 –373) © 2005 American Journal of Preventive Medicine

Introduction including breast cancer. And finally, more than 20% of

“moderate” drinkers in the U.S. general population re-

W
hen viewed prospectively, the initiation of alco-
hol consumption entails risk. In the U.S., about
30% of those who drink alcohol do so exces-
sively, and excessive drinking is the third leading actual
cause of death in the United States, killing 75,000 persons
annually.1,3,4 Furthermore, even those drinking at “mod-
erate” levels (as defined by their average daily consump-
tion) are at increased risk of death from certain causes,
From the Emerging Investigations and Analytic Methods Branch
(Brewer, Brown, Giles, T Naimi), the Behavioral Surveillance Branch
(Mokdad), the Cardiovascular Disease Branch (Mensah), Division of
Adult and Community Health, the Nutrition Branch, Division of
Physical Activity and Nutrition (Serdula), and the Office of the
Director (Lando, Stroup), National Center for Chronic Disease
Prevention and Health Promotion, the Center for Injury Prevention
and Control (Hungerford), Centers for Disease Control and Preven-
tion, Atlanta, GA; and the Division of Cardiology, Tufts–New England
Medical Center, Boston, MA (S Naimi)
Address correspondence and reprint requests to: Timothy Naimi,
Centers for Disease Control and Prevention, 4770 Buford Hwy NE,
MS K-67, Atlanta GA 30341. E-mail: tbn7@cdc.gov.
port binge drinking (drinking 5⫹ drinks at one time),
putting them at risk for injuries, violence, and other
adverse health and social outcomes.4
While moderate drinkers in most, but not all, study
populations appear to have a reduced risk of death
from cardiovascular disease (CVD), unmeasured or
residual confounding5–12 could complicate the inter-
pretation of these studies, especially since the strength
of the association between moderate drinking and CVD
outcomes is modest relative to other risk factors.13 It is
important to clarify whether CVD benefits are being
misattributed to moderate drinking, since initiating or
increasing alcohol consumption also carries risk. The
purpose of this study was to compare the prevalence of
CVD risk factors and potential confounders among
nondrinkers and moderate drinkers using data from
the Behavioral Risk Factor Surveillance System survey
(BRFSS). To date, there have been no randomized

Am J Prev Med 2005;28(4) 0749-3797/05/$–see front matter 369

© 2005 American Journal of Preventive Medicine • Published by Elsevier Inc. doi:10.1016/j.amepre.2005.01.011
clinical trials examining the relationship between mod-
erate alcohol consumption and any mortality endpoint.
Methods
The BRFSS is a population-based cross-sectional telephone
health survey of U.S. adults aged ⱖ18 years conducted by state
and territorial health departments with funding and technical
assistance provided by the Centers for Disease Control and
Prevention.14,15 In 2003, there were 250,496 respondents, and
the response rate was 54%. Average daily alcohol consumption
was calculated by multiplying the number of drinks typically
consumed by the fraction of days that alcohol was consumed
(i.e., the quantity–frequency method). Nondrinkers were de-
fined as those who did not drink alcohol during the past 30 days.
Moderate drinkers were defined as male drinkers who con-
sumed an average of two drinks per day or less, or female
drinkers who consumed an average of one drink per day or
less.16 Therefore, moderate drinkers included what some inves-
tigators refer to as light or occasional drinkers.
Two analyses compared nondrinkers to moderate drinkers;
those who drank in excess of moderate levels were excluded. In
the first analysis (reported in Table 1), the entire national
sample was used to compare nondrinkers (n ⫽116,841) to
moderate drinkers (n ⫽118,889). In the second analysis, data
from all 25 states that included questions about CVD history
were analyzed after excluding those with a history of CVD
(coronary disease, previous heart attack or stroke) or those with
poor health. This analysis included 45,771 nondrinkers and
45,170 moderate drinkers. Those with poor health were ex-
cluded because they may stop drinking (i.e., become de facto
nondrinkers) and “contaminate” the nondrinking group. Those
with a history of CVD were excluded to be sure that the
CVD-associated factors preceded the development of CVD.
The adjusted odds ratio (AOR) that a particular factor was
more likely to be associated with nondrinkers than moderate
drinkers was calculated. Odds ratios were adjusted for age
(continuous) and gender. For factors in which multiple risk
categories were assessed, the referent group was the group with
the lowest CVD risk based on the medical literature, and the
adjusted odds described the likelihood of being a nondrinker
for a particular risk category compared to the likelihood of
being a nondrinker in the referent category.
Self-reported CVD risk factors and/or confounders were
divided into five domains: demographic factors (e.g., age);
social factors (e.g., marital status, income); behavioral factors
(e.g., physical activity); health access (e.g., lack of insurance);
and health conditions (e.g., weight status, mental well-being).
The CVD risk score was calculated by summing the following
risk factors (each factor was worth 1 point): age (ⱖ45 years
for men or ⱖ55 years for women), current smoking, obesity
(body mass index [BMI]ⱖ30 kg/m2), diabetes, physical inac-
tivity, hypertension, and high cholesterol. Dental extractions
due to infection/periodontal disease were analyzed because
oral hygiene and dental infections are independently associ-
ated with CVD.17
Results
Of the individual factors, 27 of 30 (90%) were signifi-
cantly more prevalent among nondrinkers than mod-
370 American Journal of Preventive Medicine, Volume 28, Number 4
erate drinkers; only male gender and smoking were
significantly more associated with moderate drinking.
This was true before and after adjusting for age and
gender. Examples of factors more commonly associated
with nondrinking status included being older and
nonwhite, being widowed or never married, having less
education and income, lacking access to health care or
preventive health services, having comorbid health
conditions such as diabetes and hypertension, having
lower levels of mental well-being, being more likely to
require medical equipment, having worse general
health, and having a higher CVD risk score.
For factors in which there were multiple risk catego-
ries, there was a graded relationship between increas-
ing levels of risk and an increased likelihood of being a
nondrinker (Table 1). For example, those with progres-
sively lower levels of income, education, physical activ-
ity, and overall health status, or those with higher BMIs
or higher CVD risk scores, were progressively more
likely to be nondrinkers than moderate drinkers.
Excluding those with either poor health or a history
of CVD from the analysis did not substantially affect the
results. Of the 30 factors assessed, 26 (87%) were still
significantly more common among nondrinkers than
moderate drinkers; only male gender and current
smoking were significantly more common among mod-
erate drinkers (data not shown). Examples of factors
that were significantly associated with nondrinking
status included being older and nonwhite; being unem-
ployed (AOR⫽1.28); having an income of ⬍$25,000
(AOR⫽2.85); lacking health insurance (AOR⫽1.50);
having diabetes (AOR⫽2.20); reporting more un-
healthy mental days (AOR⫽1.20); requiring medical
equipment (AOR⫽1.97); and having a CVD score of
ⱖ5 (AOR⫽2.91).
Discussion
The purpose of this population-based study was to assess
the prevalence of known CVD risk factors and potential
confounders among nondrinkers and moderate drinkers
in order to determine if these factors could account for at
least some of the apparent protective effect of moderate
drinking on CVD. In sum, it appears that moderate
drinkers have many social and lifestyle characteristics that
favor their survival over non-drinkers, and few (if any) of
these differences are likely due to alcohol consumption
itself. Overall, 90% of risk factors were significantly more
common among nondrinkers; only 2 risk factors were
more common among moderate drinkers. Furthermore,
among factors with multiple risk categories, higher levels
of risk (e.g., the highest BMI category) were progressively
more strongly associated with nondrinking status. The
findings were similar before and after adjusting for age
and gender, and after excluding those with poor health or
a history of CVD from the analysis. These results suggest
that residual confounding or unmeasured effect modifi-
Table 1. Distribution of selected cardiovascular risk factors and potential confounding factors by drinking status, and the
adjusted relative odds of being a nondrinker among people with those factors, Behavioral Risk Factor Surveillance System
survey, 2003 (continued)
Nondrinkers Moderate drinkers Adjusteda relative
(%, SE) (%, SE) odds (95% CI) of
Risk factor/confounder (n ⴝ 116,841) (n ⴝ 118,889) being a nondrinker
DEMOGRAPHIC FACTORS
Age (years)
18–34 27.2(0.28) 32.6(0.27) —
35–54 36.1 (0.27) 42.1 (0.27) —
55–64 14.2 (0.18) 12.4 (0.16) —
ⱖ65 22.5 (0.22) 12.9 (0.17) —
Gender (% men) 39.1 (0.29) 55.0 (0.27) —
Race/ethnicity
White, NH 66.0 (0.29) 74.7 (0.28) 1.00 (referent)
Black, NH 11.6 (0.19) 7.2 (0.15) 1.95 (1.84–2.07)
Hispanic 14.7 (0.26) 11.7 (0.24) 1.66 (1.22–1.52)
Asian 3.1 (0.14) 2.3 (0.11) 1.87 (1.64–2.14)
American Indian/Alaska Native, NH 1.7 (0.07) 1.4 (0.06) 1.51 (1.35–1.70)
Other 2.8 (0.10) 2.7 (0.10) 1.36 (1.22–1.52)
SOCIAL FACTORS
Marital status
Married/unmarried couple 61.7 (0.28) 65.4 (0.26) 1.00 (referent)
Divorced/separated 11.9 (0.17) 11.1 (0.15) 1.05 (1.00–1.09)
Widowed 10.1 (0.14) 4.6 (0.10) 1.43 (1.34–1.52)
Never married 16.4 (0.24) 18.9 (0.24) 1.16 (1.11–1.22)
Unemployed 6.0 (0.16) 5.4 (0.14) 1.29 (1.20–1.40)
Education
College graduate 22.7 (0.23) 38.8 (0.26) 1.00 (referent)
Some college 25.7 (0.24) 28.0 (0.25) 1.56 (1.50–1.62)
High school graduate 34.6 (0.27) 26.1 (0.25) 2.23 (2.14–2.32)
⬍High school 17.1 (0.23) 7.1 (0.17) 4.06 (3.81–4.32)
Income
ⱖ$50,000 28.7 (0.28) 49.3 (0.29) 1.00 (referent)
$25,000–$49,999 32.5 (0.29) 29.8 (0.26) 1.83 (1.75–1.91)
⬍$25,000 38.8 (0.30) 20.9 (0.25) 2.96 (2.84–3.10)
BEHAVIORAL FACTORS
No “leisure-time” physical activity 31.9 (0.26) 17.7 (0.22) 2.02 (1.94–2.09)
Overall physical activity level
Recommended 40.6 (0.29) 49.7 (0.28) 1.00 (referent)
Insufficient 38.0 (0.28) 40.2 (0.27) 1.11 (1.07–1.15)
Inactive 21.4 (0.24) 10.1 (0.17) 2.31 (2.19–2.42)
Smoking status
Never smoker 59.0 (0.28) 50.9 (0.27) 1.00 (referent)
Former smoker 22.8 (0.23) 26.7 (0.24) 0.68 (0.66–0.71)
Current smoker 18.2 (0.22) 22.4 (0.23) 0.76 (0.73–0.79)
Diet—vegetables (<5 servings/day) 75.1 (0.25) 77.3 (0.23) 1.00 (0.96–1.04)
HEALTHCARE ACCESS
No health insurance 16.0 (0.23) 13.5 (0.21) 1.49 (1.42–1.57)
No personal doctor 18.1 (0.25) 20.9 (0.25) 1.07 (1.02–1.12)
Couldn’t see doctor due to cost 14.0 (0.20) 11.2 (0.19) 1.35 (1.28–1.42)
No influenza shot (age >50) 46.4 (0.38) 48.0 (0.42) 1.14 (1.08–1.19)
No cholesterol screening 25.0 (0.27) 25.4 (0.26) 1.26 (1.21–1.32)
No colorectal cancer screening 41.0 (0.81) 36.4 (0.86) 1.34 (1.21–1.49)
(age >50)
HEALTH CONDITIONS
Diabetes 11.4 (0.17) 4.6 (0.11) 2.37 (2.23–2.52)
Hypertension 31.2 (0.25) 21.7 (0.22) 1.44 (1.39–1.49)
Body mass index (kg/m2)
⬍18.0 1.5 (0.08) 1.0 (0.06) 1.69 (1.43–2.00)
18.0–24.9 36.4 (0.28) 41.1 (0.27) 1.00 (referent)
25.0–29.9 34.9 (0.28) 38.0 (0.27) 1.12 (1.08–1.16)
30.0–34.9 17.0 (0.22) 14.0 (0.20) 1.44 (1.37–1.51)
35.0–39.9 6.5 (0.14) 4.1 (0.11) 1.82 (1.69–1.96)
ⱖ40.0 3.8 (0.11) 1.9 (0.07) 2.28 (2.06–2.51)
High blood cholesterol 36.1 (0.29) 32.4 (0.28) 1.05 (1.01–1.09)

(continued on next page)

Am J Prev Med 2005;28(4) 371

Table 1. Distribution of selected cardiovascular risk factors and potential confounding factors by drinking status, and the
adjusted relative odds of being a nondrinker among people with those factors, Behavioral Risk Factor Surveillance System
survey, 2003
Nondrinkers Moderate drinkers Adjusteda relative
(%, SE) (%, SE) odds (95% CI) of
Risk factor/confounder (n ⴝ 116,841) (n ⴝ 118,889) being a nondrinker
HEALTH CONDITIONS (continued)
Current asthma 8.7 (0.16) 6.9 (0.14) 1.19 (1.13–1.27)
Any teeth removed (10 states)b 54.8 (0.63) 40.7 (0.54) 1.57 (1.46–1.68)
Arthritis 32.3 (0.25) 23.5 (0.22) 1.24 (1.19–1.28)
General health status
Excellent 17.3 (0.22) 24.7 (0.23) 1.00 (referent)
Very good 28.3 (0.26) 37.4 (0.26) 1.05 (1.01–1.10)
Good 31.2 (0.26) 28.0 (0.25) 1.53 (1.47–1.61)
Fair 15.9 (0.21) 8.0 (0.16) 2.57 (2.42–2.73)
Poor 7.3 (0.14) 2.0 (0.07) 4.58 (4.18–5.02)
>14 unhealthy physical days 14.1 (0.20) 6.2 (0.13) 2.23 (2.12–2.36)
>14 unhealthy mental days 10.6 (0.18) 8.0 (0.15) 1.32 (1.25–1.40)
>14 activity limitation days 16.3 (0.29) 7.5 (0.20) 2.15 (2.00–2.32)
Use medical equipment 9.4 (0.15) 3.9 (0.10) 2.13 (2.00–2.28)
CVD risk scorec
0 23.5 (0.26) 29.0 (0.25) 1.00 (referent)
1 25.1 (0.25) 31.1 (0.26) 1.00 (0.95–1.05)
2 21.4 (0.23) 21.0 (0.22) 1.24 (1.18–1.31)
3 15.8 (0.19) 12.0 (0.17) 1.58 (1.49–1.67)
4 9.2 (0.15) 5.0 (0.12) 2.12 (1.98–2.28)
ⱖ5 5.0 (0.12) 1.9 (0.07) 3.19 (2.89–3.52)
Note: For factors in which multiple categories were assessed, the referent group was the lowest risk group. Example interpretation for education:
compared to those with a college education (the referent group), those with less than a high school education had 4.06 times the odds of being
nondrinkers, after adjusting for age and gender. An identical analysis was performed after excluding those with either poor health or a history
of CVD (see results).
a
The odds ratios and 95% CIs were adjusted for age (continuous) and gender.
b
Refers to tooth removal due to gum disease or periodontal infection.
c
The CVD risk score was calculated by summing the following: age (ⱖ45 for men; ⱖ55 for women); smoking; obesity (body mass index ⱖ30);
diabetes; physical inactivity; hypertension; and high cholesterol.
CI, confidence interval; CVD, cardiovascular disease; NH, non-Hispanic; SE, standard error.

cation would bias observational studies in favor of moder-
ate drinkers.
In studies of moderate drinking and CVD, it can be
difficult to control for confounding because CVD is
multifactoral, and because moderate drinking has a rela-
tively small effect size relative to other risk factors. Fur-
thermore, while most studies control for some CVD risk
factors, there are emerging factors that are also indepen-
dently associated with CVD. Examples of such factors
include poverty, psychological characteristics, newly dis-
covered CVD risk factors (e.g., homocysteine, C-reactive
protein, multiple adverse childhood experiences), and
as-yet unidentified factors.13,18 –20 Finally, although com-
binations of CVD-related factors (e.g., diabetes coupled
with lack of health insurance) are synergistic in terms of
risk,21–24 it is not generally practical or possible to assess all
such combinations for effect modification and most stud-
ies do not address effect modification at all.
This study did not compare moderate drinkers to
those who never drank alcohol (such analyses are done
in an effort to remove “sick quitters,”25 but this infor-
mation was unavailable in BRFSS). However, excluding
sick quitters have not explained observed differences in
CVD outcomes in U.S. studies.26 Furthermore, exclud-
ing those in poor health, which is another way to
remove sick quitters,5 did not substantially affect the
results of the analysis. Although it is possible that
alcohol-related nonresponse could bias the study, non-
response is typically highest among those drinking in
excess of moderate levels, and these individuals were
excluded from our analyses since we were only studying
nondrinkers and moderate drinkers. And finally, al-
though information on risk factors and health condi-
tions was based on self-report, it seems likely that
nondrinkers (who had less education and healthcare
access) would actually be less likely to be aware of their
prevalent health conditions and risk factors, suggesting
that the associations we observed between nondrinking
status and CVD risk factors were conservative.
Widespread scientific and public perceptions about the
benefits of moderate drinking 27–32 may carry public
health risk since approximately 30% of current drinkers
drink excessively,4 and excessive drinking accounts for
75,000 deaths annually in the United States.3 If large
numbers of nondrinkers begin drinking for their health,
some initiates will undoubtedly drink excessively and/or
suffer adverse effects from alcohol. This creates an ethical
dilemma for the scientific and public health communi-
ties, particularly in the absence of randomized controlled
trials about the impact of moderate alcohol consumption

372 American Journal of Preventive Medicine, Volume 28, Number 4


What This Study Adds . . .
Observational studies suggest that moderate
drinking may lower cardiovascular disease (CVD)
mortality.
Although most studies control for “major”
CVD-associated risks, CVD is independently asso-
ciated with many factors.
In this population-based study, 27/30 (90%) of
CVD-associated factors were more prevalent
among nondrinkers than moderate drinkers.
Furthermore, among factors with multiple risk
strata, increasing risk was progressively associated
with nondrinking status.
These findings suggest that the apparent pro-
tective effect of moderate drinking may be due to
confounding.
on any mortality endpoint, including all-cause mortality.
For this and other reasons, current clinical and public
health guidelines focus on reducing drinking among
those who drink excessively, and do not recommend that
people begin or increase drinking for health reasons.33,34
Continued caution is particularly justified since there are
safe and effective strategies to decrease CVD mortal-
ity,35–37 many of which are underutilized.38
No financial conflict of interest was reported by the authors of
this paper.
References
1. Mokdad AH, Stroup D, Marks JS, Gerberding J. Actual causes of death in
the United States, 2000. JAMA 2004;291:1238 – 45.
2. Centers for Disease Control and Prevention. Available at: www.cdc.gov/
alcohol/. Accessed December 10, 2004.
3. Stahre M, Brewer RD, Naimi TS, et al. Alcohol-attributable deaths and years
of potential life lost due to excessive alcohol use in the U.S. Morb Mortal
Wkly Rep 2004;53:866 –70.
4. Naimi TS, Brewer RD, Mokdad AH, Denny C, Serdula M, Marks JS. Binge
drinking among U.S. adults. JAMA 2003;289:70 –5.
5. Wannamathee SG, Shaper AG. Alcohol, coronary heart disease and stroke:
an examination of the j-shaped curve. Neuroepidemiology 1998;17:288 –95.
6. Shaper AG, Wannamathee SG. Epidemiological confounders in the rela-
tionship between alcohol and cardiovascular disease. In: Paoletti R, Klatsky
AL, Poli A, Zakhari S, eds. Moderate alcohol consumption and cardiovas-
cular disease. Dordrecht: Kluwer, 2000:105–12.
7. Criqui MH, Ringel BL. Dies diet or alcohol explain the French paradox?
Lancet 1994;344:1719 –23.
8. Shaper AG, Wannamathee SG. The J-shaped curve and changes in drinking
habit. In: Chadwick DJ, ed. Alcohol and cardiovascular disease. Chichester:
John Wiley and Sons, 1998:173–93.
9. Criqui M. Do known cardiovascular risk factors mediate the effect of
alochol on cardiovascular disease? In: Chadwick DJ, ed. Alcohol and
cardiovascular diseases. Chichester: John Wiley and Sons, 1998:159 –72.
10. Rehm J, Greenfield TK, Rogers JD. Average volume of alcohol consump-
tion, patterns of drinking, and all-cause mortality: results form the U.S.
National Alcohol Survey. Am J Epidemiol 2001;153:64 –71.
11. Thun MJ, Peto R, Lopez AD, et al. Alcohol consumption and mortality
among middle-aged men and elderly U.S. adults. N Engl J Med
1997;337:1705–14.
12. Murray RP, Connett JE, Tyas SL, et al. Alcohol volume, drinking pattern,
and cardiovascular disease morbidity adn mortality: is there a u-shaped
function? Am J Epidemiol 2002;155:242– 8.
13. Splaver A, Lamas GA, Hennekins CH. Homocysteine and cardiovascular
disease: biological mechanisms, observational epidemiology, and the need
for randomized trials. Am Heart J 2004;148:34 – 40.
14. Nelson DE, Holtzman D, Waller M, et al. Objectives and design of the
Behavioral Risk Factor Surveillance System. Paper presented at American
Statistical Association National Meeting, Dallas TX, 1998.
15. Nelson DE. Reliability and validity of measures from the Behavioral Risk
Factor Surveillance System (BRFSS). Social Prev Med 2001;46(suppl
1):S3– 42.
16. U.S. Department of Agriculture and U.S. Department of Health and
Human Services. Nutrition and your health: dietary guidelines for Ameri-
cans. 5th ed. Home and Gardening Bulletin 232. Washington DC: U.S.
Government Printing Office, 2000.
17. Mattila KJ, Valtonen VV, Nieminen M, Huttunen JK. Dental infection and
the risk of new coronary events: prospective study of patients with docu-
mented coronary artery disease. Clin Infect Dis 1995;20:588 –92.
18. Cahalin D, Cisin IH, Crossley HM. Demographic and sociological corre-
lates of levels of drinking. In: American drinking practices. New Brunswick
NJ: Rutgers Center on Alcohol Studies, 1969:18 – 64.
19. Dong M, Giles WH, Felitti VJ, et al. Insights into causal pathways for
ischemic heart disease: and adverse childhood experiences study. Circula-
tion 2004;110:1761– 6.
20. Williams JE, Paton CC, Siegler IC, et al. Anger proneness predicts coronary
artery heart disease risk: prospective analysis from the atherosclerosis risk
in communities (ARIC) study. Circulation 2000;101:2034 –9.
21. Parodi PW. The French paradox unmasked: the role of folate. Med
Hypotheses 1997;49:313– 8.
22. Kopp P. Resveratrol, a phytoestrogen found in red wine. A possible
explanation for the conundrum of the French paradox. Eur J Endocrinol
1998;138:619 –20.
23. McNamee R. Confounding and confounders. Occup Environ Med
2003;60:227–34.
24. Szklo M, Nieto FJ. Epidemiology: beyond the basics. Gaithersburg MD:
Aspen Publications, 2000.
25. Tsubono Y, Yamada S, Nishino Y, et al. Choice of comparison group in
assessing the health effects of moderate alcohol consumption. JAMA
2001;286:1177– 8.
26. Mukmal 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.
27. Rosen M. Drink to this: the evidence is growing: for most people, having
one or two glasses of alcohol every day is healthier than not drinking at all.
Boston Globe, Health Science section, April 6, 2004, pp. C11, C13.
28. Zuger A. The case for drinking (all together now: in moderation!). New
York Times, December 31, 2002.
29. Hu FB, Manson JE, Stampfer MJ, et al. Diet, lifestyle, and the risk of type 2
diabetes mellitus in women. N Engl J Med 2001;345:790 –7.
30. Klatsky AL. Drink to your health? Scientific American 2003:75– 81.
31. Raloff J. When drinking helps: sorting out for whom a nip might prove
therapeutic. Sci News 2003;163:155– 6.
32. Doyle M. Wine health claims allowed. Sacramento Bee. Available at:
www.sacbee.com/content/lifestyle/taste/story/6213819p–7168466c.html.
Accessed April 2, 2003.
33. US Department of Health and Human Services and US Department of
Agriculture. Dietary Guidelines for Americans 2005. Alcoholic Beverages.
Available at: http://www.health.gov/dietaryguidelines/dga2005/document/
pdf/Chapter9.pdf. Accessed March 9, 2005.
34. Goldberg IJ, Moska L, Piano MR, Fisher EA. Wine and your heart: a science
advisory for healthcare professionals from the Nutrition Committee, Coun-
cil on Epidemiology and Prevention, and Council on Cardiovascular
Nursing of the American Heart Association. Circulation 2001;103:472–5.
35. National Cholesterol Education Project. The detection, evaluation and
treatment of high blood cholesterol in adults (Adult Treatment Panel III).
Available at: www.nhlbi.nih.gov/guidelines/cholesterol/atp3full.pdf. Ac-
cessed June 30, 2003.
36. Shepherd J, Cobbe SM, Ford I, et al. Prevention of coronary heart disease
with pravastatin in men with hypercholesterolemia. N Engl J Med
1995;333:1301–7.
37. Wilson K, Gibson N, Willan A, Cook D. Effect of smoking cessation on
mortality after myocardial infarction. Arch Intern Med 2000;160:339 – 44.
38. Coffield AB, Maciosek MV, McGinnis MJ, et al. Priorities among recom-
mended clinical preventive services. Am J Prev Med 2001;21:1–9.
Am J Prev Med 2005;28(4) 373