| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
(Circulation. 2000;102:489.)
© 2000 American Heart Association, Inc.
Editorial |
Alcohol Consumption, Diabetes, and Coronary Disease
An Epidemiological Perspective
From the Division of Cardiology, Emory University School of Medicine, Atlanta, Ga.
Correspondence to William S. Weintraub, MD, Division of Cardiology, Emory University School of Medicine, 1639 Pierce Dr, WMB 319, Atlanta, GA 30322. E-mail bill{at}ecor.eushc.org
Key Words: Editorials • alcohol • diabetes mellitus • coronary disease
Coronary artery disease and diabetes mellitus are common, serious medical conditions, with diabetes being a serious risk factor for coronary events as well. Although some data suggest that mild to moderate alcohol consumption may be beneficial in preventing coronary events, there is natural concern about the safety of even moderate alcohol consumption for diabetics who suffer from increased risk of coronary disease and may seek to use alcohol therapeutically to decrease their risk.
This important concern is addressed by Ajani et al in the current
issue of Circulation.1 Alcohol consumption
was defined by survey collected at baseline. Diabetes was considered to
be present if the participant gave a history of diabetes or used
insulin or other antidiabetic medication. Follow-up was 
5.5 years,
with mortality assessed from the National Death Index and cause of
death confirmed from death certificates. The authors examined the
outcomes of 87 938 US physicians, among whom there were 2790
diabetics. There were 850 deaths due to coronary heart disease,
717 in nondiabetics and 133 in diabetics. The effect of alcohol
consumption on mortality was expressed as the relative risk, adjusted
for other risk factors. Risk is the probability of an event. Relative
risk is the risk of an event in a group with a risk factor divided by
risk in the group without the risk factor. Thus, relative risks <1
mean the risk factor is actually protective. In the nondiabetics, the
relative risks for coronary deaths, compared with those who
drank no alcohol, were 1.02 for monthly, 0.82 for weekly, and 0.61 for
daily consumers of alcohol. In the diabetics, the relative risks for
coronary deaths, compared with those who drank no alcohol, were
1.11 for monthly, 0.67 for weekly, and 0.42 for daily consumers of
alcohol. In a subgroup of 21 852 men (510 with diabetes), the risk of
incident coronary events, including myocardial infarction and
revascularization, was assessed by annual survey,
with an average follow-up of 12 years. Similar reductions in risks of
incident coronary events were noted compared with the mortality
data. Thus, the study shows a lower risk of coronary deaths and
events with increasing alcohol consumption up to 1 drink per day. The
reduction in risk was similar in diabetics and nondiabetics.
This study confirms many epidemiological studies that have shown a decrease in cardiovascular events in patients who consume moderate amounts of alcohol.2 3 4 5 It should be noted that there is a J-shaped relationship between alcohol consumption and mortality, with minimal mortality being in the range of 1 to 2 drinks a day.6 With higher levels of alcohol consumption, mortality rises from noncardiac causes. A major portion of the protective effect of alcohol has been suspected in epidemiological studies to be due to an increase in HDL cholesterol with moderate alcohol consumption.7 In addition, alcohol may increase fibrinolytic activity8 and decrease platelet aggregation,9 and some alcohol-containing beverages, such as red wine, may contain antioxidants.10
There are far fewer data concerning the potential protective effects of
alcohol in diabetics. In a case-control study, alcohol was actually
reported to be a risk factor for coronary disease in
diabetics.11 In a smaller epidemiological study, there was
no association of alcohol consumption and coronary disease in
diabetics.12 A recent preliminary report from another
small study suggests that alcohol consumption in diabetics decreased
events.13 The only other major epidemiological study to
examine this specific question was conducted in 983 older diabetics
(mean age 68.6, 45.2% male) with 12.3 years of
follow-up.14 Compared with people who never drank and
controlled for multiple other risk factors, former drinkers had a
relative risk of coronary death of 0.69; those who drank <2
g/d (<1 drink/wk) had a relative risk of 0.54; those who drank 2 to 13
g/d had a relative risk of 0.44; and those who drank 
14 g/d (1
drink/d) had a relative risk of 0.21.
Thus, the epidemiological data are clear and consistent. There is a decreased relative risk of cardiac death in the population, and probably in the diabetics as well, who are mild to moderate consumers of alcohol. The first question to ask is whether these data are internally valid. The most serious challenge to the validity would be misclassification of alcohol consumption, diabetes, or cause of death. In general, such misclassification is likely to be random and, thus, cause a bias toward the null. That is, if the data are imperfect, it is much more likely that the effect of alcohol on mortality is larger, not smaller. Are the results generalizable? The present study is in men in their 50s. The results may not be entirely generalizable to either younger or older men or to women. The authors quote a companion study in women from the Nurses’ Health Study with similar findings. The study by Valmadrid et al14 was conducted in older men and women. The generalizability of these studies to younger patients who would be at low risk of coronary disease but at higher risk of potential alcohol abuse is most uncertain. These data cannot be readily applied to patients with established disease. The potential protective effects may be either greater or smaller.
However, there is a more fundamental limitation. All the data to date are from epidemiological studies in which alcohol is treated as a risk factor, or in this case, a protective factor with a relative risk of <1. Association does not prove causation or protection. A series of criteria can be applied, albeit in a flexible manner, to a risk factor to suggest causation, as originally outlined by Hill.15 These are (1) strength of association, (2) dose-response effect, (3) lack of temporal ambiguity, (4) consistency of the findings, (5) biological plausibility, (6) coherence of the evidence, and (7) specificity of association. Aside from criterion 7, alcohol consumption would appear to meet the Hill criteria. However, the fundamental problem with an epidemiological association such as this is the potential for confounding.16 A confounder is a risk factor for the disease, in this case coronary death, the correction of which will reduce or eliminate a biased estimate of the true disease-exposure relationship, in this case the relationship of coronary deaths and alcohol consumption. The confounder must vary in prevalence in the exposure (ie, alcohol consumption) groups. Confounding may be assessed by examining the subgroups of patients with and without the risk factor or, more commonly, by multivariate analysis, which is a method to account for multiple confounders at once. However, multivariate analysis cannot correct for unmeasured confounders. The potential confounder in this case, as pointed out by the authors, is that healthier people may be the ones who consume moderate amounts of alcohol. There is precedent for the occurrence of problems of this type. Hormone replacement therapy in women has been shown in epidemiological studies to be associated with decreased incidence of cardiovascular events.17 18 However, in the randomized Heart and Estrogen-Progestin Replacement Study (HERS), hormone replacement therapy could not be shown to decrease events.19
To examine risk factors for disease, such as cigarette smoking, epidemiological studies are the only population-based method available to study association and ultimately infer causation. In the case of a form of therapy, randomized trials can overcome the bias caused by confounders by making sure that the confounders are evenly distributed between groups. The problem in consideration of moderate alcohol consumption is that it is extremely unlikely that a randomized trial would or even could be mounted. Where does this leave us? In men and women without prior coronary disease from their 50s to perhaps their early 70s, both diabetics and nondiabetics, there is a decreased incidence of coronary deaths in people who consume mild to moderate amounts of alcohol. Extrapolation of these data to older patients, younger patients, or patients with established coronary disease is probably not warranted. Furthermore, this association can only be considered suggestive and not proof of causation. Thus, from a public policy point of view, it would seem that moderate alcohol consumption is safe and may be beneficial from the point of view of cardiovascular risk in selected populations, both diabetic and nondiabetic. However, there is probably not sufficient evidence to recommend alcohol consumption to decrease risk in any population.
Acknowledgments
The author thanks Lesley Wood for her expert editorial review.
Footnotes
The opinions expressed in this editorial are not necessarily those of the editors or of the American Heart Association.
(Circulation. 2000;102:489-490.)
References
1.
Ajani UA, Gaziano JM, Lotufo PA, et
al. Alcohol consumption and risk of coronary heart
disease by diabetes status. Circulation.. 2000;102:500–505.
2.
Pearson TA. Alcohol and heart disease.
Circulation. 1996;94:3023–3025.
3.
Friedman GD, Klatsky AL. Is alcohol good for your
health? N Engl J Med.. 1993;329:1882–1883.
4. Rimm EB, Giovannucci EL, Willett WC, et al. Prospective study of alcohol consumption and risk of coronary disease in men. Lancet. 1991;338:464–468.[Medline] [Order article via Infotrieve]
5.
Fuchs CS, Stampfer MJ, Colditz GA, et al. Alcohol
consumption and mortality among women. N Engl J
Med. 1995;332:1245–1250.
6.
Bradley KA, Donovan DM, Larson EB. How much is too
much? Advising patients about safe levels of alcohol consumption.
Arch Intern Med.. 1993;153:2734–2740.
7.
Gaziano JM, Buring JE, Breslow JL, et al. Moderate
alcohol intake, increased levels of high-density lipoprotein and its
subfractions, and decreased risk of myocardial infarction. N
Engl J Med. 1993;329:1829–1834.
8.
Ridker PM, Vaughan DE, Stampfer MJ, et al. Association
of moderate alcohol consumption and plasma concentration of
endogenous tissue-type plasminogen
activator. JAMA. 1994;272:929–933.
9. Rubin R, Rand ML. Alcohol and platelet function. Alcohol Clin Exp Res. 1994;18:105–110.[Medline] [Order article via Infotrieve]
10. Renaud S, de Lorgeril M. Wine, alcohol, platelets, and the French paradox for coronary heart disease. Lancet. 1992;339:1523–1526.[Medline] [Order article via Infotrieve]
11. Thulaseedharan N, Augusti KT. Risk factors for coronary heart disease in noninsulin dependent diabetes mellitus (NIDDM). Indian Heart J. 1995;47:471–476.[Medline] [Order article via Infotrieve]
12. Rosengren A, Welin L, Tsipogianni A, et al. Impact of cardiovascular risk factors on coronary heart disease and mortality among middle aged diabetic men: a general population study. BMJ. 1989;299:1127–1131.
13. Diem P, Fajfr R, Mueller B, et al. Effects of alcohol consumption on mortality in type-2 diabetic subjects. Diabetes. 1999;48:A86. Abstract.
14.
Valmadrid CT, Klein R, Moss SE, et al. Alcohol intake
and the risk of coronary heart disease mortality in persons
with older-onset diabetes mellitus. JAMA. 1999;282:239–246.
15. Hill AB. Principles of Medical Statistics. 9th ed. New York, NY: Oxford University Press; 1971:309–323.
16. Kleinbaum DG, Kupper LL, Morgenstern H. Confounding. In: Epidemiologic Research: Principles and Quantitative Methods. New York, NY: Lifetime Learning Publications; 1982:242–265.
17.
Folsom AR, Mink PJ, Sellers TA, et al. Hormonal
replacement therapy and morbidity and mortality in a prospective study
of postmenopausal women. Am J Public Health. 1995;85:1128–1132.
18.
Grodstein F, Stampfer MJ, Manson JE, et al.
Postmenopausal estrogen and progestin use and the risk of
cardiovascular disease. N Engl J
Med. 1996;335:453–461.
19.
Hulley S, Grady D, Bush T, et al. Randomized trial of
estrogen plus progestin for secondary prevention of coronary
heart disease in postmenopausal women. JAMA. 1998;280:605–613.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||