Internal and Emergency Medicine

, Volume 5, Issue 4, pp 291–297

Prevention of cardiovascular risk by moderate alcohol consumption: epidemiologic evidence and plausible mechanisms


    • “RE ARTU” Research Laboratories, “John Paul II” Centre For High Technology Research and Education in Biomedical SciencesCatholic University
  • Simona Costanzo
    • “RE ARTU” Research Laboratories, “John Paul II” Centre For High Technology Research and Education in Biomedical SciencesCatholic University
  • Maria Benedetta Donati
    • “RE ARTU” Research Laboratories, “John Paul II” Centre For High Technology Research and Education in Biomedical SciencesCatholic University
  • Licia Iacoviello
    • “RE ARTU” Research Laboratories, “John Paul II” Centre For High Technology Research and Education in Biomedical SciencesCatholic University
  • Giovanni de Gaetano
    • “RE ARTU” Research Laboratories, “John Paul II” Centre For High Technology Research and Education in Biomedical SciencesCatholic University
IM - Review

DOI: 10.1007/s11739-010-0346-0

Cite this article as:
Di Castelnuovo, A., Costanzo, S., Donati, M.B. et al. Intern Emerg Med (2010) 5: 291. doi:10.1007/s11739-010-0346-0


An inverse association between moderate alcohol intake and cardiovascular risk, in particular coronary disease and ischemic stroke, has been shown in many epidemiologic studies. In addition, several other diseases are also known to occur less frequently in moderate drinkers than in non-drinkers, whereas excess of drinking is invariably harmful. However, some concern has been recently raised about the possibility that at all dosages the harm of alcohol could overcome its beneficial effects. We present here the epidemiologic and mechanistic evidence to support the protective effect of moderate alcohol intake against cardiovascular disease and all-cause mortality.


AlcoholWineVascular riskTotal mortality


There is strong and consistent epidemiologic evidence of a significant inverse association between moderate alcohol consumption and cardiovascular disease (CVD), in particular coronary disease, ischemic stroke [13] and peripheral arterial disease [4, 5].

The 2006 Diet and Lifestyle Recommendations Scientific Statement from the American Heart Association Nutrition Committee [6] recommends that: “If you consume alcohol, do so in moderation (equivalent to no more than one drink for women or two drinks for men per day)”. In addition, moderate ethanol consumption has now emerged as a dominant component of a Mediterranean diet score, used as a predictor of lower mortality [7]. Several not primarily vascular diseases are also known to occur less frequently in moderate drinkers than in non-drinkers [8], whereas excessive alcohol consumption is unquestionably harmful [810]. Although the protective effect of moderate alcohol consumption on atherothrombotic vascular disease is widely accepted by the scientific community, some concern has been recently raised, particularly at regulatory agencies level, on the possibility that at all dosages the harmful effects of alcohol might overcome its benefit [10]. The relationship between alcohol and vascular risk or total mortality has been depicted as a J-shaped curve: the risk is indeed lower at light to moderate alcohol consumption, but increases at higher doses [9]. However, if low alcohol intake is inversely related to CVD, the other side of the coin shows an increased risk of certain cancers, cirrhosis and death from accidents mainly associated with increasing alcohol consumption [810].

The aim of this survey is to review the epidemiologic evidence supporting the relationship of alcohol dosing with vascular risk and all-cause mortality, but we shall start by a brief discussion of the biological plausibility of some major mechanisms that might contribute to the protective effect of moderate alcohol intake against CVD.

Mechanisms of action of alcohol as a protective factor against CVD

Alcohol consumption correlates with modification of several vascular and biochemical factors that have potential cardioprotective benefits. Increase in high-density lipoprotein (HDL) cholesterol levels, decrease in platelet aggregation via inhibition of prostaglandin synthesis and changes in fibrinogen, tissue-plasminogen activator (t-PA), and PA inhibitor (PAI-1) levels, are thought to represent major mechanisms to reduce the risk of cardiovascular events [11, 12].

Increase HDL levels

The increase in HDL reportedly associated with alcohol drinking has been for a long time considered one of the main mechanisms of alcohol action. However, recent data suggest that increased HDL cholesterol through medicaments has no significant effect on vascular risk [13]. By inference, the association of alcohol with elevation of HDL per se, without considering the effects on its function, appear nowadays of reduced clinical relevance.

Oxidized low-density lipoproteins (LDL) are known to play a major role in atherosclerosis and CVD, and are a commonly used marker for oxidative damage [14]. Moderate consumption of alcohol is associated with a reduction in the LDL susceptibility both to oxidation and aggregation. In addition, alcohol-induced elevations in HDL may facilitate the transport of LDL cholesterol back to the liver for reprocessing, thus dropping the cholesterol available to endothelial activation and atheroma formation [14]. As a consequence, we can infer that moderate alcohol intake might promote reduction in the toxic effects of elevated LDL. Nonetheless, although depending on some genetic characteristics (e.g., Apo E), alcohol increases LDL plasma levels [15], an apparently paradoxical effect since LDL are considered to be a risk factor for myocardial infarction. Possibly, regulation of lipids rather than LDL elevation alone should be taken into consideration with respect to alcohol benefits. More basic studies are needed to clarify this controversial point.

Antiatherogenic and antithrombotic properties

Moderate alcohol consumption is independently associated with less extensive coronary atherosclerosis in humans. In a study on more than 2,000 patients undergoing coronary angiography, alcohol consumption was associated with a lower percent of lumen narrowing in the main coronary vessels [16].

Several studies have shown that polyphenols, mainly contained in red wine, are able to inhibit platelet arachidonic acid metabolism and biosynthesis of thromboxane A2, a potent platelet aggregation inducer and a vasoconstrictor [17]. Moderate alcohol consumption may also affect fibrinogen concentration, platelet aggregation, t-PA and PAI-1 [11, 12, 1820].

Alcohol consumption and omega-3 polyunsaturated fatty acids

Omega-3 fatty acids (ω3 FA) consumption reduces the risk of sudden cardiac death [21]. The Lyon Diet Heart Study showed that moderate wine consumption was associated with higher plasma levels of “marine” ω3 FA independently from the dietary intake of specific plant and marine ω3 [22]. Thus, the protection resulting from moderate alcohol drinking might be mediated through increased ω3 FA. The association of alcohol consumption with ω3 FA was lately investigated in the framework of the IMMIDIET study that enrolled an adult population of about 2,000 people from three different European regions [23]. Eicosopentaenoic acid (EPA), docosahexaenoic acid (DHA) and EPA + DHA in plasma, on one hand and EPA and EPA+DHA in red blood cells on the other, were all positively associated with alcohol intake, independently from fish intake and other possible confounders related to life styles [23].

Ethanol preconditioning

Epidemiological data and several experimental animal studies suggest that low dose ethanol induces in the heart a chronic protective state that is independent from traditional (lipid and coagulation) risk factors [24, 25],

As reported in the FLORA study [26], the hearts of rats fed a diet rich in anthocyanins (that are part of the non alcoholic component of wine) were more resistant to regional ischemia and reperfusion insult. Because anthocyanins do not circulate a long time, it is likely that they are protective irrespective of the timing of their assumption in relation to the moment of ischemia.

However, it was also seen, at least in the rabbit [25] that if alcohol is present within the circulation during the ischemic insult, it could block its own protection. In all probability, ethanol is only a “preconditioning” agent, which means it is not protective if it is present in blood at the moment of ischemia [27].

Mechanisms of action of wine

Wine possibly acts through mechanisms that might provide additional cardiovascular benefits than that provided by ethanol. Red wine contains a wide variety of polyphenols, including phenolic acids, tannins, trihydroxy stilbenes (resveratrol), and flavonoids (catechin, epicatechin and quercetin) that influence low-density lipoprotein oxidation, platelet aggregation, endothelial function and smooth muscle cell proliferation [2832]. Some of these molecules inhibit lipoprotein oxidation, promote nitric oxide formation by vascular endothelium, reduce serum inflammatory biomarkers, inhibit thromboxane A2 biosynthesis in platelets and leukotriene biosynthesis in neutrophils, and regulate lipoprotein production and secretion. These actions occur through the inhibition of various enzymes, such as phospholipase A2, cyclooxygenases, phosphodiesterase, and several protein kinases involved in cell signaling [2838].

Mechanisms by which alcohol may increase cardiovascular risk

Heavy alcohol consumption and irregular (binge) drinking are associated with several adverse effects: fetal alcohol syndrome, liver cirrhosis, certain cancers, hypertriglyceridemia, hypertension, hemorrhagic stroke, obesity, alcohol intoxication and dependence [10, 39].

Blood pressure levels and prevalence of hypertension linearly increase with alcohol consumption [8].

Excessive alcohol intake is the second main cause of hyperlipidemia in the population [40]. Overweight subjects should limit or avoid alcohol to reduce unnecessary energy intake. Increased clotting and a reduced threshold for ventricular fibrillation are the main mechanisms explaining the negative effects of heavy drinking on CVD [39].

Nevertheless, both positive and negative reported mechanisms have often been deduced from epidemiologic data, and are not real mechanisms. The actual relevance of inhibition of some enzymes and lipoprotein oxidation is indirect and often questioned. Further basic studies, able to define plausible mechanisms of action are warranted.

Epidemiologic evidence that moderate alcohol consumption decreases both cardiovascular risk and total mortality

Despite the fact that the healthy effect of moderate intake of alcohol is by now well accepted [12, 41]; important issues remain to be discussed about the relationship between alcohol and CVD, including the possible effect of confounding, the differences in types of alcoholic beverages, the optimal amount of alcohol intake, associated with health benefit, the individual or environmental modulation of the alcohol-related effects and the pattern of drinking.

The role of life style

It has been argued that uncontrolled confounding by associated lifestyle factors play a major role in the association between moderate alcohol drinking and coronary protection [42] based on the results showing that drinkers have many healthier characteristics than non-drinkers and thus have lower ischemic heart disease risk [43]. However, Mukamal et al. [44] show that the differences between abstainers and drinkers were attenuated, eliminated, or even reversed by additional adjustment for sociodemographic factors. Moreover, a similar pattern was apparent for physical activity, a widely recommended behavior for the prevention of CVD. Despite this fact, concerns about the possibility of confounding in studies on physical activity have been largely overshadowed.

The differences among alcoholic beverages

Many epidemiologic studies have explored the hypothesis that consuming alcohol in the form of wine might confer a protection against CHD above that expected from its alcohol content [2]. As mentioned above, wine might indeed conceivably exert additional ethanol unrelated beneficial effects. Despite a large number of experimental studies that confirm such a hypothesis, epidemiologic evidence of a greater effect of wine has not been definitely established. To test such a hypothesis, our group performed two meta-analyses based on the 26 studies reporting comparisons between different alcoholic beverages (wine and beer) and the relative risk (RR) of CVD [2]. A first meta-analysis was conducted on studies that investigated the relation between vascular events and specific alcoholic beverages, irrespective of the amount consumed. Fifteen studies were included, involving 208,096 persons. The overall RR of wine drinkers with respect to non-drinkers was 0.68 [95% confidence interval (95% CI) 0.59–0.77], whereas the protection associated with beer drinkers showed an overall RR 0.78, 95% CI 0.70–0.86. However, following the exclusion of the studies that did not simultaneously adjust for different types of alcoholic beverages, there was no difference in the RRs of CVD between wine and beer drinkers (0.75 vs. 0.77) when compared with abstainers. In contrast, an important difference between wine and beer consumption was observed in the second meta-analysis conducted on studies reporting trend (dose–response) analysis. Ten studies were included involving 176,042 persons. In wine drinkers a clear inverse dose–effect curve (J curve) was found, while all the fitted models used failed to show any significant relationship between different amounts of beer intake and vascular risk. Therefore, it was only possible to identify a daily dose of wine, namely 150 ml, under which the beneficial protection against CVD was no more significant.

If wine is better than beer or spirits remains to be elucidated. Further studies addressing this issue should be of large sample size and carefully designed, because differences between beverages, if any, are expected to be limited and might reflect differences in the risk factor patterns, traditions, or geographic distribution among categories of drinkers rather than a true difference in the protection from CVD risk.

The most healthy amount of alcohol intake

As already mentioned, the dose–response relationship between wine or alcohol intake and rate of CVD has been depicted as a J-shaped curve [13, 9]. The J-shaped relationship between alcohol consumption and clinical events reflects non-drinkers having higher incidence and mortality rates than light or moderate drinkers, but similar or lower rates than heavy drinkers. In our meta-analysis [2], an “average dose–response” curve was obtained fitting a non-linear model. The observed relationship confirmed a “J-shaped” curve since, after an initial decrease in the vascular risk by increasing amounts of wine, the curve reached a plateau at higher intake, and tended to be reverted at the highest amounts explored. A maximum reduction was predicted at 750 ml/day, but statistical significance was only reached up to the amount of 150 ml/day, equivalent to 12–18 g of ethanol per day. In conclusion, epidemiologic evidence indicates that the amount of wine/alcohol for which the balance between risk and benefit is the best is in the range of 1–2 drinks a day, in agreement with American Heart Association guidelines [45].

Is the effect of alcohol different in men and women?

In our meta-analysis of studies that enrolled only men [2], the protection offered by wine was found to be surprisingly small (13%) and not significant, in contrast with studies enrolling both sexes (47%). No sex-related difference was observed for beer consumption. In a meta-analysis of studies on alcohol and stroke [3], greater protection in women than in men was apparent. Confronting dose–response curves separately in men and women in a meta-analysis on alcohol and total mortality [9], we observed that the protection was apparent up to three drinks per day in men, but only up to two drinks per day in women, whereas the maximal risk reduction was similar (17%; 99% CI 15–19% and 18%; 99% CI 13–22%, respectively). The pooled curves for men and women were different at the range at which alcohol was protective—in fact, the inverse association in women apparently disappeared at doses lower than in men—but comparable regarding the maximal protection.

Thus, a possible sex difference in the protective effect of alcohol might exist and explain apparently controversial results in different epidemiologic studies. Large studies are needed to test whether women are more susceptible indeed to the benefit of alcohol or if they more likely drink lower amounts of alcohol, thus receiving its maximal advantage [46, 47].

The role of the pattern of drinking

Experimental studies [48] suggest that drinking wine at meals provides maximal health effect, by prevention of the development of atheromatous lesions, while a binge pattern of drinking has been associated with higher risk of CHD [49, 50]. In the large health professional follow-up study [51, 52], the authors studied the association of alcohol consumption with the risk of myocardial infarction [51] and ischemic stroke [52], focusing their investigations on the pattern of drinking (frequency and quantity of alcohol consumption). Although they failed to observe any correlation between health effect of alcohol and the proportion consumed with meals, they found a role of drinking frequency: the highest protective effect on risk of myocardial infarction was among men who drank alcohol three or more days a week, whatever the amount consumed. On the other hand, alcohol consumption of 10–30 g/day on 3–4 days/week appeared to be associated with the lowest risk of ischemic stroke when compared with abstention or other amount of alcohol intake.

Moderate regular drinking, possibly during meals appears as the ideal behavior, while binge drinking is to be absolutely avoided.

Protection against total mortality

If low alcohol intake is inversely related to CVD, the other side of the coin shows an increased risk of certain cancers, cirrhosis and death from accidents associated with increasing alcohol consumption [1, 10]. As a consequence, one could suspect that even at lower dosages the benefit of alcohol could be overcome by its harmful effects [10, 42, 47]. To test such a hypothesis, we performed a meta-analysis including 34 prospective studies on alcohol and all-cause mortality [9]. We pooled findings from more than one million subjects and about 95,000 deaths from any cause as measured outcomes. The J-shaped relationship observed between total mortality and increasing amounts of alcohol consumed indicate that low to moderate consumption of alcohol (≤1 drink/day in women and ≤2 drinks/day in men) significantly reduces total mortality, while higher doses increase it.

Is the beneficial effect of alcohol a real one?

As mentioned above, we paid special attention to the possible effect of confounding. Twenty-nine studies showed adjusted RRs at least for age; among them, 15 were adjusted for social status too, and 6 for social status and dietary markers. Pooled curves for different levels of adjustment were significantly different (P < 0.0001) showing that part of heterogeneity is attributable indeed to adjustment. However, the average maximal protection observed in not adjusted studies was 36%, it decreased in adjusted studies to an average of 17% (anyway a substantial and statistically significant effect). The observed difference between the 5 not adjusted and the 29 adjusted studies could also be due to factors different from the level of adjustment, as different groups of studies were compared. Therefore, we performed an additional analysis of adjusted or not unadjusted data obtained from the same studies. In the latter case, the effect due to known confounders (age, smoking, social status, dietary factors) led to the reduction in the maximal average protection from 19 to 16%; for analogy, even in the pessimistic hypothesis that residual (possibly unknown) confounding would have a similar strength as the known one in lowering the protection, one can assume that the “real” (maximal) protection against total mortality associated with low consumption of alcohol would be higher than 10–12%.

In conclusion, although the apparent protection by alcohol or wine decreases when data are adjusted, thus confirming the importance of confounding in assessing drinking effects, it largely remains in a range of undoubted public health value [9].

The problem of former drinkers

We also investigated the degree to which the inclusion of ex-drinkers in control group could influence the results of our meta-analysis on alcohol and total mortality [9]. The inclusion in the control groups of people who had stopped drinking owing to illness might induce an overestimation of the protection offered by drinking in moderation. We tested this hypothesis by comparing studies that used as reference group, the category of no alcohol intake and/or excluded former drinkers with studies which, in contrast, included in their reference group occasional or former drinkers or people reporting low occasional alcohol intake: the protection was indeed somewhat lower in the first studies, but still remained statistically significant.

Should abstention be recommended to CVD patients?

A meta-analysis on five cohorts of CVD patients [53] found a 20% reduction in all-cause mortality risk in moderate drinkers in comparison with abstainers. Janszky et al. [54] found that among survivors from a primary myocardial infarction, moderate drinkers (up to 20 g/day) have lower secondary cardiovascular and all-cause mortality than abstainers. Regular (non-binge) and moderate alcohol intake is significantly associated with a reduced incidence of secondary cardiovascular and all-cause mortality in patients with a history of cardiovascular events [55]. If not contraindicated, patients who drink alcohol should not exceed 1–2 drinks/day for women or up to 2–3 drinks/day for men, as a component of a balanced cardioprotective dietary pattern with appropriate energy intake levels.

A last question: should a randomized controlled clinical trial be performed on the protective effect of alcohol in CVD disease?

Randomized controlled trials usually offer a more solid answer than observational studies to many questions in medicine, but have mainly been restricted to the efficacy of drugs; controlled intervention trials on diet in general and on alcohol in particular are difficult to be performed, mainly because of selection, blinding or compliance and ethical problems [47]. The sponsor(s) of this kind of trial is also difficult to identify. Notwithstanding these difficulties, a large randomized controlled trial on alcohol, comparing a reference group of long-term abstainers, devoid of former or occasional drinkers and an “intervention” group of light to moderate no-binging drinkers, with appropriate follow-up, collecting not only vascular, but total mortality data too, would help in clarifying the effect of alcohol on health. Randomised controlled trials on intermediate endpoints could perhaps be more feasible and offer a valuable alternative.

At the present moment, however, one has to rely upon observational studies such as those analyzed here or prospective studies where participants spontaneously decrease or stop drinking. Interestingly enough, the first study of the latter type [56] supports the inverse relation of moderate alcohol intake with CVD.


Available epidemiologic data, that are limited at the moment to observational studies, confirm the hazards of excess drinking, but also indicates the existence of potential windows of alcohol intake that may confer a net beneficial effect of drinking, at least in terms of survival, both in men and in women. Methodological limitations of observational study design, the role of uncontrolled confounding and the optimal choice of the reference group are important issues to be considered. However, the protection by alcohol would anyway remain in a range of undoubted public health value, even considering all possible confounders.

In a general public health perspective, alcoholic beverages should be avoided whenever consumption would put an individual or others at risk, such as during pregnancy or before driving. Moreover, alcohol intake, even in moderation, must be avoided by young people, as they are at very low cardiovascular risk and tend to misuse alcohol (binging) [8, 57].

Besides insisting on the control of risk factors, abstainers should be informed that, in the absence of contraindications and in the context of healthy eating and lifestyle, low-moderate alcohol consumption may contribute to better health. Asymptomatic subjects at high cardiovascular risk or those who have had a cardiovascular event might also be informed of the cardiovascular risk reduction associated with low-moderate drinking. People who are already regular light-moderate alcohol consumers should be encouraged to continue. The hazards of excess drinking should be always highlighted, and heavy drinkers pushed to cut their consumption to a moderate level [6].

The rates of vascular and total mortality, as well as the development of hypertension or type 2 diabetes, are lower for people who drink low to moderate amounts of alcohol than for people who do not drink at all or drink heavily [1, 5861]. However, some concerns have been raised about whether it is advisable to encourage adult people to drink small amounts regularly rather than abstain completely, especially among poor populations and in low-income countries where the disease burden per unit of alcohol consumption is greater [10, 62].

The cardioprotective nature of alcohol has been attributed to both its antithrombotic properties and its ability to decrease LDL levels and/or their oxidation. Moreover, wine and especially red wine, due to its polyphenol content might offer additional advantages and greater cardiovascular benefits than alcohol alone.

Genetic regulation of the individual response to alcohol has been recently proposed [63, 64].

A large randomized controlled trial on alcohol, comparing a referent group of long-term abstainers, devoid of former or occasional drinkers and an “intervention” group of light to moderate no-binging drinkers, with appropriate follow-up, collecting not only vascular, but total mortality too will without doubt help in clarifying the effect of alcohol on health.

Further investigations in this emerging field will hopefully assist in interpreting heterogeneous findings in different settings and populations.

Conflict of interest


Copyright information

© SIMI 2010