Alcohol and Coronary Heart Disease: Drinking Patterns and Mediators of Effect

The study of the effects of alcohol on health in human populations has spanned a wide spectrum, ranging from controlled feeding studies to cross-cultural comparisons to large scale prospective studies (1). Many detailed review papers have covered these areas of research — several are also included in this issue of the Annals of Epidemiology. In this paper we focus on the inverse association between moderate alcohol consumption and CHD and highlight recent evidence that contributes to the hypothesis that alcohol is causally related to lower risk through known biological mechanisms.

The first study to suggest an inverse association between alcohol consumption and CHD was published more than 30 years ago (2). Since then, additional data confirming this association have been collected in more than 25 countries from at least 100 different studies. Evidence from these studies suggests that beneficial changes in HDL cholesterol levels, clotting factors, insulin sensitivity, and markers of inflammation provide biological plausibility to the observed association. Recently, epidemiological studies have more carefully addressed the importance of drinking patterns, frequency, quantity, and beverage choice while also accounting for differences in physical activity, diet, and changes in alcohol consumption over the adult lifespan.

The preponderance of evidence suggests that “moderate” drinkers of alcohol have the greatest cardiovascular benefit. The definition of moderate drinking has a very broad range (from 5 to 60 g of alcohol per day) depending on the study population and the tool used for assessment. For purposes of this summary, we will use the definition defined by the 2005 USDA Dietary Guidelines (3). Moderate drinking is considered to be no more than one drink per day for women and no more than two drinks per day for men. A drink is more specifically defined as 12 ounces of regular beer, 5 ounces of wine, and 1.5 ounces of 80-proof distilled spirits. Thus, at the extreme, two drinks a day would be up to approximately 30 g of alcohol.

Substantial epidemiological evidence from geographical comparisons, large cohort studies, and several meta-analyses links moderate alcohol consumption with a lower risk of cardiovascular disease. There are too many epidemiological studies that have independently analyzed the association between alcohol and CHD risk to summarize in this paper, though they have been summarized elsewhere 4, 5, 6, 7. The population, size, and length of follow-up all vary greatly from study to study, which provides further support to the robustness and generalizability of the findings. For example, in the first study in 1974, Klatsky et al (2) analyzed recent health and lifestyle histories of 464 patients in the San Francisco Bay Area who had a multiphase health checkup and subsequently a first myocardial infarction. The alcohol histories of the cases suggested significantly lower alcohol intake compared with that of 50,000 healthy examinees. The Framingham Heart Study, a prospective epidemiologic cohort study established in Massachusetts in 1948 to evaluate potential risk factors for CHD, is also important because of the extremely long follow-up period (26 years in an early report of alcohol and CHD). With a single measure of alcohol at baseline, the authors reported an inverse association with subsequent CHD risk (8). Similar inverse associations have been documented in such diverse populations as those in France, Japan, Denmark, Germany, Finland, Korea, Great Britain, the United States, and many others 9, 10, 11, 12, 13, 14, 15. Despite the strength and consistency across studies, some still argue that generalization of the results may not be possible because of the selected nature of several of these study populations. However, general population surveys including the National Health and Nutrition Examination Survey (NHANES) in the United States have also found benefit from moderate alcohol consumption (16).

More recent studies of alcohol and CHD have focused on subgroups defined by age or health status. Although alcohol in moderation will likely provide greater benefit for older populations where rates of CHD are highest (10), the etiology of CHD is such that moderate consumption in middle age also is beneficial. Several important risk factors for CHD, such as obesity and the prevalence of diabetes, both of which have been increasing in younger adults, are consistently reported to be inversely associated with alcohol consumption 17, 18, 19, 20, 21, 22. In addition to subgroups across age strata, benefit from moderate alcohol consumption has been documented in many other subpopulations, such as those persons with diabetes and hypertension or patients followed up after myocardial infarction 23, 24, 25, 26. Thus, in summary, the epidemiological evidence for an inverse association between moderate alcohol consumption and CHD is extensive and generalizable to populations defined by age, ethnicity, geography, and prevalent health conditions.

Despite this evidence, some have raised doubt as to whether the observed inverse association between alcohol and CHD risk is causal. In 1988, Shaper et al (27) suggested an alternative explanation—the “sick quitter” hypothesis—suggesting that the reference category of abstainers were at higher risk of CHD not because they did not drink alcohol, but rather because they stopped drinking because of existing illness. Fillmore et al (28) recently published a subsequent review and meta-analysis supporting the hypothesis by Shaper and colleagues and adding further theories of systematic misclassification error due to changes in alcohol consumption with aging and poor health. Unfortunately, in their review (28), the inclusion criteria were excessively selective and thus the results were difficult to interpret.

Although the hypothesis put forward by Shaper et al may have a theoretical basis, a simple misclassification among a small percentage of nondrinkers cannot fully explain why men and women who consume one to three drinks per day have a 10% to 40% lower risk of CHD when compared with the 10% to 60% of the population (depending on the study) who abstain (29). There is substantial evidence to refute the “sick quitter” hypothesis. First, subsequent analyses from several large cohorts with data on lifelong or long-term abstainers have successfully refuted the hypothesis 30, 31. Second, moderate alcohol consumption not only reduces the risk of incident CHD in healthy individuals, but also reduces CHD and mortality in individuals with hypertension, diabetes, and existing CHD 23, 25, 26, 32. Thus, if the inverse association persists in a “sick” population, this suggests that changes in alcohol consumption after diagnosis do not appreciably change the shape of the relationship between alcohol and subsequent disease. Third, and most importantly (as we discuss in more detail below), there are more than 70 metabolic studies which document important changes in the cardiovascular risk profile caused by daily drinking (29).

Some have also suggested that the inverse association between alcohol and CHD may not be causal because moderate drinkers may also be more likely to eat better, exercise more, and live a healthier life. Although most prospective studies of alcohol and CHD control for potential confounders, uncontrolled or residual confounding could still explain the beneficial effect attributed to alcohol. Ideally, a long-term randomized trial of alcohol would help to rule out confounding. Such trials have been conducted to study changes in markers of CHD such as HDL cholesterol, triglycerides, glycemic control, and clotting factors, but long-term trials which follow hard clinical cardiovascular events may be difficult to conduct because of cost, ethical considerations regarding the randomization of alcohol, and finally the generalizability of results among what would need to be a very highly selective population who would be willing to be randomized to a specific drinking pattern for a long period.

We recently conducted a re-analysis of men in the Health Professionals Follow-up Study (HPFS) to examine the association between alcohol and CHD in the strata of men already at low risk based on other health lifestyle choices. This effectively eliminates the potential for residual confounding by healthy behavior because only men with a healthy lifestyle were included in the analysis.

Briefly, the HPFS is a prospective investigation of the dietary etiologies of heart disease and cancer among 51,529 men, aged 40 to 75 years at baseline in 1986 9, 31. The population consists of dentists, veterinarians, pharmacists, optometrists, osteopathic physicians, and podiatrists. In our initial analysis from this population with 4 years of follow-up (9), we found a strong inverse association for moderate alcohol consumption that persisted even after exclusion of men with baseline health conditions which may have led to a change in drinking pattern. In a subsequent analysis of these men with 12 years of follow-up and 1,418 incident CHD cases, we investigated the role of drinking pattern. Men with the lowest rates of CHD consumed alcohol 3 to 4 days per week, regardless of the type of beverage (beer, wine, spirits) or whether or not alcohol was consumed with meals (31).

Finally, in our most recent analysis with 16 years of follow-up and 2,183 incident CHD cases, we identified 5 factors that define a low-risk individual: (1) nonsmoker, (2) body mass index <25 kg/m2, (3) moderate-to-vigorous activity 30 minutes or more per day, (4) the top 40% of the distribution for a healthy diet score, and (5) moderate alcohol consumption. Individually, these modifiable lifestyle factors were associated with lower risk of CHD, but a combination of lifestyle factors had an even greater impact (33). Men who adhered to all 5 low-risk factors had the lowest risk of CHD (relative risk = 0.13; 95% confidence interval [CI] = 0.09–0.19) compared with men with no low-risk factors during the 16-year follow-up period. To address the issue of residual confounding by healthy lifestyle among drinkers, we restricted the analysis to the 8,867 healthy men who adhered to the first 4 low-risk behaviors (not including alcohol). We then examined the association for alcohol consumption among this group of the healthiest men. In this group we still had 106 incident CHD cases and found that men who drank moderately (15.0–29.9 g/d) had a relative risk of 0.38 (95% CI, 0.16–0.89) compared with abstainers. This strong inverse association between moderate alcohol consumption and CHD in predominantly healthy individuals adds further evidence to support the hypothesis that the inverse association is causal, and not confounded by healthy lifestyle behaviors (34).

The strongest evidence to support a causal association between alcohol and CHD comes from metabolic studies which demonstrate changes in lipids, hemostatic factors, inflammatory markers, and glycemic control parameters. In a previous meta-analysis, we summarized the effects of alcohol on a variety of these biomarkers from experimental studies (29) and found strong and consistent evidence linking moderate alcohol intake with higher concentrations of HDL cholesterol and apolipoprotein A1 and lower concentrations of fibrinogen. Since the publication of this article, several cross-sectional and experimental studies have also reported strong positive associations between alcohol and increased insulin sensitivity 35, 36, 37. The combined beneficial effect of alcohol on the underlying biological mechanisms, specifically HDL cholesterol, fibrinogen, and hemoglobin A1c (as a marker of glycemic control and insulin resistance), can almost entirely explain the reduced CHD risk among moderate drinkers.

Historically, early studies by Langer et al. (38) estimated that approximately 50% of the benefit of moderate alcohol consumption could be explained by the direct effects of alcohol on HDL cholesterol. We extended this analysis and included other biomarkers of CHD. In nested case-control studies of CHD from the HPFS and the Nurses' Health Study, we modeled the inverse association between alcohol and CHD among men and women with and without controlling for the potential biological mediators, HDL cholesterol, fibrinogen, and hemoglobin A1c (39) (Fig. 1). In the figure, each subsequent model after the basic model controls for an additional biomarker thought to be in the causal pathway between moderate alcohol consumption and reduced risk of CHD. For both the men and the women, approximately half of the benefit (the magnitude of the regression coefficient is halved) is explained away after including HDL cholesterol in the model. Subsequent models illustrate the change in the coefficient after controlling for fibrinogen, hemoglobin A1c and all 3 biomarkers simultaneously. The final model which controls for all 3 potential mediators suggests that all of the benefit of alcohol in men and 80% of the benefit of alcohol in women can be explained by these biomarkers in the causal pathway between alcohol and CHD.

View Large Image Download to PowerPoint

Figure 1 The inverse association between alcohol consumption and risk of CHD in the Health Professionals Follow-up Study (A) and the Nurses' Health Study (B). The β-coefficient (also called the regression coefficient) is the natural log of the odds ratio in frequency of consumption per week. The basic model controls for age, smoking, and date of blood sample return (39).

The importance of the role of drinking pattern and specifically beverage choice is still disputed. Similar to the argument that moderate drinkers may have a healthier lifestyle, there are documented differences in healthy lifestyle characteristics by predominant beverage choice and drinking pattern, which vary by country and cultural setting (40). Thus it is more likely that in any given culture, the alcohol-containing beverage which is consumed in moderation, most frequently, and with meals will be of greatest benefit. For example, in a large cohort from France (where wine is the most frequent alcohol beverage of consumption), Renaud et al. (11) evaluated the health risk of both wine and beer and found that both beverage choices reduced the risk of cardiovascular death, albeit wine was more beneficial. In Germany, where much of the alcohol consumed is beer, the risk of CHD events for drinkers was one half that of nondrinkers (12). In Japan, where spirits are the common beverage of choice, Kitamura et al. (13) found that alcohol intake reduced premature incidence of CHD among urban Japanese middle-aged men; within that study, less than 1% of alcohol consumed was from grape wine. When we summarized the available evidence, the inverse association between moderate alcohol intake and CHD was consistent, regardless of beverage choice (41). This suggests that the direct metabolic effects of ethanol are largely responsible for the cardiovascular benefits and that other constituents of the beverages contribute little if anything to further benefit.

The evidence discussed above provides substantial support for the hypothesis that moderate drinking reduces the risk of CHD. Beer, wine, and spirits all have demonstrated significant benefits. These benefits are likely mediated through strong and lasting effects of alcohol on HDL cholesterol, fibrinogen, and glycemic control. The “sick-quitter” hypothesis and the concern that moderate drinkers lead a healthier lifestyle may explain a small proportion of the benefit attributed to alcohol in some studies, but recent studies which have removed sick quitters, updated alcohol and covariate information on diet and lifestyle factors, and separately documented benefits of alcohol among healthy and unhealthy populations further add to the evidence that moderate alcohol consumption is causally related to a lower risk of CHD.