Wine Drinking and Risks of Cardiovascular Complications After Recent Acute Myocardial Infarction
Background— Scientific data on the clinical impact of moderate alcohol consumption after a recent acute myocardial infarction (AMI) are limited, and the specific effect of wine ethanol has not been studied.
Methods and Results— In survivors of a recent AMI, we analyzed the association between ethanol intake and the risk of recurrence. The patients were classified according to the amount of ethanol that they consumed regularly during follow-up. Major prognostic factors, including the severity of the prior AMI and drug treatment, were recorded and included in the analyses. Only patients with at least 2 reliable assessments of drinking (and dietary) habits were included (n=437). The average ethanol intake was 7.6% of the total energy intake, wherein wine ethanol represented 92% of the total. Among these patients, 104 cardiovascular complications occurred during a mean follow-up period of 4 years. In comparison with abstainers, the adjusted risk of complications was reduced by 59% (95% confidence interval: 17 to 80) in patients whose average ethanol intake was 7.7% of the total energy intake (about 2 drinks/day), and by 52% (95% confidence interval: 4 to 76) in those whose average ethanol intake was of 16% of energy (about 4 drinks/day).
Conclusion— Whereas moderate wine drinking was associated with a significant reduction in the risk of complications in this homogenous population of coronary heart disease patients, further studies are required to confirm the data, define the clinical and biological profile of the patients who would most benefit from wine drinking after recent AMI, and examine whether the relations found are due to ethanol or other wine ingredients.
Received April 8, 2002; revision received June 26, 2002; accepted June 26, 2002.
A number of case-control and cohort studies in different countries have reported decreased cardiovascular disease (CVD) rates among healthy moderate drinkers as compared with abstainers.1–9⇓⇓⇓⇓⇓⇓⇓⇓ In most studies, the benefit of drinking varied according to age and the underlying risk factors of CVD, and it was largely confirmed for subjects over 60 years of age at increased risks. In contrast, data on the impact of moderate ethanol drinking in patients with established CVD are limited and controversial.10–13⇓⇓⇓ Recent studies have shown either a small effect11 or no effect12 of moderate drinking on the risk of CVD. In fact, none of these studies were done in patients who were recovering from a recent acute myocardial infarction (AMI), no data were reported regarding the severity of the prior AMI, which is the main prognostic factor in these patients, and sometimes only the consumption of alcohol in the year before the AMI was used in the analyses.13 Finally, the specific effect of wine, which is often associated with a long-term and moderate drinking pattern (as opposed to the binge and irregular drinking patterns), has not yet been studied.
The aim of this study, which was conducted on survivors of a recent AMI, was to examine the association between habitual ethanol intake and the risk of CVD complications. The main prognostic factors and potential confounders, such as smoking, severity of the prior AMI, drug treatment, and dietary habits were carefully evaluated and included in the analyses.
The Lyon Diet Heart Study is a randomized secondary prevention trial initiated in 1988 to test whether a Mediterranean type of diet may prevent coronary recurrence and CVD complications after a first and recent (<2 months) AMI.14 The patients in the experimental group were instructed to follow a Mediterranean diet, but no specific recommendations were given regarding the consumption of alcoholic beverages. The patients of the control group were supposed to follow the dietary advice given by their attending physicians, along the lines of a prudent Western-type diet.14 The final report after a mean follow-up of 4 years was published in 1999 and confirmed that the Mediterranean diet is actually cardioprotective.14 Additionally, no major bias was detected in the trial.14 Complete data about clinical follow-up, drug treatment, dietary (including drinking) habits, and major prognostic factors were available in 437 patients in whom we analyzed the relationships between wine drinking and the risk of CVD complications.
The methods used for the validation and classification of the complications have been reported.14 Only those clinical events requiring admission to the coronary care unit and invasive investigations such as coronary angiography were retained. An independent Endpoint Committee blinded to patients’ group assignment validated and classified the endpoints out of the raw data obtained from hospital files. The main risk factors of coronary heart disease (CHD), including blood pressure and blood lipids, were assessed at randomization. Other prognostic factors or potential confounders (eg, smoking and drug treatment) were also recorded at randomization and during follow-up. The diet evaluation comprised a 24-hour recall and a frequency questionnaire. Alcohol consumption was estimated in the same way to obtain detailed data about the frequency, amount, and type of alcohol drinks usually consumed. For each patient, the ethanol intake per 24 hours was calculated in energy and converted into a percent contribution to the total energy intake. This way of normalization was preferred to control for variations in body mass index, lifestyle (physical exercise and energy expenditure), and dietary factors (energy intake) that potentially confound the associations between ethanol intake and the risk of CHD.15,16⇓ To provide clinically meaningful information, an optimal classification (an optimal indicator of exposure) would be on the basis of serial measurements of ethanol intake over the entire period of follow-up. Thus, the patients were classified according to their habitual consumption of ethanol during follow-up. Only patients with at least 2 reliable assessments of their drinking habits were retained. In case of a major discrepancy between the different visits (in case of irregular or binge drinking, for instance), patients were excluded from analysis (n=35). Because the few women (n=49) with available data showed a wide variation in ethanol intake throughout follow-up, thus entailing a risk of misclassification, and also because they were not equally distributed into the 4 quartiles, they were all excluded from this analysis. Using the calculated mean consumption of ethanol, patients were categorized into quartiles of ethanol consumption. Zero percent of energy intake per day derived from ethanol (non-drinkers) was quartile 1, <5.41% of total energy intake per day was quartile 2, >5.41 but <9.84% of energy was quartile 3, and >9.84% of energy was quartile 4. Thus, patients categorized as non-drinkers (quartile 1) were total and permanent abstainers during the entire follow-up.
The Cox proportional hazards model was used to quantify the associations between habitual ethanol intake and the rate of recurrences. A composite outcome as previously described14 was used, and it included the primary endpoints of the Lyon trial (cardiac death and nonfatal AMI), the major secondary endpoints (episodes of unstable angina or overt heart failure, stroke, and pulmonary embolism), and the other secondary endpoints (recurrent angina, surgical or medical revascularization procedures, post-angioplasty restenosis, and thrombophlebitis). The censoring date for each patient was the date of the earliest primary (if any) or secondary endpoint event. Risk ratios were calculated by comparing the frequency of the endpoints in the patients of quartiles 2 through 4 with that in the reference quartile (quartile 1, non-drinkers) and were adjusted for the diet group assignment. They were also computed with multivariate proportional hazards models controlling simultaneously for diet group assignment, age, and current smoking. The reported risk ratios were not adjusted for several predictors of CVD complications affected by ethanol consumption (namely high-density lipoprotein cholesterol, body mass index, triglycerides, blood glucose) because those are effects of exposure and should not be controlled.4,15⇓ Because of the importance of blood pressure and total cholesterol as major risk factors in secondary prevention,14 however, these 2 variables were included in the multivariate model.
The data about 353 male patients (163 on the Western diet and 190 Mediterranean) were available. Ethanol intake accounted for 8% of total energy intake in the Western (control) group (range: 0% to 25.9%), wherein wine ethanol represented 88% of the total. The average ethanol intake in the Mediterranean group was 7.3% (range: 0% to 31.4%), wherein wine ethanol represented 95%. Thus, most of the ethanol consumed (92%) by these patients came from wine, in agreement with published data about the French adult population.5 Among the patients in quartile 4, the number of chronic heavy drinkers (ethanol intake >20% of energy) was quite small (n=17), and only 3 patients exceeded 25% (31.4%, 27.6%, and 25.1%). There were no irregular or binge drinkers in that group of patients (see Methods). The baseline characteristics of patients are shown in Table 1. The main aspects of their habitual diet were computed separately for each diet group (Table 2), because a combined presentation would have confused the data. Clearly, in terms of dietary habits, which is a major risk factor of recurrence,14 there was no major difference across the quartile categories within each diet group. Drinkers did not have a healthier diet than abstainers. Abstainers were also not older and did not smoke more than drinkers (Table 1). Table 3 shows the infarction history, which reflects the severity of the prior AMI, and the medications at randomization. The use of the various medications was also remarkably stable during the follow-up, 14 and there was no significant difference across the quartile categories in this regard.
During a mean follow-up of 4 years, 104 complications occurred. All but 9 (1 cerebral stroke, 1 peripheral embolism, and 7 episodes of overt heart failure) were CHD recurrences. There were 4 deaths, 14 recurrent AMIs, and 15 episodes of unstable angina. There were also 24 episodes of recurrent angina that required admission to the coronary care unit but did not fully meet the criteria for AMI or unstable angina, as previously defined.14 Finally, there were 17 cases of postangioplasty restenosis, and 21 patients needed myocardial revascularization (bypass surgery in 3 cases and coronary angioplasty in 18 cases). There were 36, 34, 18, and 16 complications in the quartiles 1, 2, 3, and 4, respectively, and we found a significant inverse trend across ethanol intake categories in the diet-adjusted models (P=0.01, Table 4). In comparison with the abstainer group, the risk of recurrence was lower among the patients whose ethanol intake was about 7.7% (quartile 3, about 2 drinks per day) or 16% (quartile 4, an average of 4 to 5 drinks per day) of the total energy intake (Figure). Control for potential confounders in multivariate analysis only slightly changed the risk ratios (Table 4). Inclusion of AMI severity measures (Table 3) in the model did not substantially change the results. Analyses were repeated including only the 42 hard endpoints (death, AMI, cerebral stroke, pulmonary embolism, and unstable angina). This analysis provided similar trends as those reported with the 104 complications. Despite the small number of cases (8 and 7 cases in quartiles 3 and 4 versus 11 and 16 in quartile 1 and 2), the risk ratios were 0.65 and 0.69 after adjusting only for diet (P=0.01) when comparing the quartiles 3 and 4 with the reference quartile 1, and 0.53 and 0.58 after multiple adjustments (P=0.07).
This study shows that in middle-aged male survivors of a recent AMI, moderate wine ethanol drinking was associated with a significant reduction in the risk of CVD complications. This is in line with most (but not all) previous studies about ethanol consumption and the risk of CHD.1–13⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓ Regarding the specific association between wine drinking and lower CHD risk, however, ecological studies are thought to be confounded by lifestyle, diet, and other cultural factors.17 In addition, because there is no study comparing exclusive wine drinkers with total ethanol abstainers and because the consumption of wine but not of other alcoholic beverages has been associated with a healthier lifestyle, which could partly explain the lower CHD risk,18–20⇓⇓ it is unclear whether (or to what extent) wine drinking protects against CHD independently of any confounding lifestyle factor. For instance, in a Californian study,18 subjects who preferred wine were shown to smoke less, have more education, and have more temperate drinking habits than those who preferred other alcoholic beverages. Binge drinking is indeed associated with loss of the protective effect seen with regular moderate drinking, and sometimes poses an increased risk of cardiac death.21 Another example is given by Danish researchers who reported that in their country, wine drinking is often associated with the intake of healthy foods, which the authors defined as fruit, vegetables, fish, salads, and olive oil,19 a type of diet that was unambiguously shown to protect against CHD.14 Finally, among British men aged 40 to 59 years, a large part of the greater benefit seen in wine drinkers relative to other drinkers was attributed to lifestyle characteristics, in particular low rates of smoking and obesity.20 This raises the question of whether the potential protection against CHD is actually related to the drink (wine) or to the lifestyle of the drinker.22
To help to solve this important question, and because wine trials are not feasible, it is important to design studies that measure and control for the main lifestyle and health confounders of the association between wine ethanol and CHD. One solution is to study this association in the context of secondary prevention, where the conventional risk factors of CHD and the main prognostic factors are, in principle, systematically measured. The data on secondary prevention of CHD, however, are limited and conflicting.8,10–13⇓⇓⇓⇓ Thun et al8 reported that the adjusted death rates from CHD were 30% to 40% lower among men with preexisting CHD who reported at least 1 drink daily than among non-drinkers, whereas Shaper and Wannamethee12 found that regular ethanol drinking in men with established CHD was not associated with any significant benefit for CHD as compared with occasional drinking. On the other hand, Muntwyler et al11 reported that in a large cohort of middle-aged men with a history of myocardial infarction, there was a reduction of total and CVD mortality with alcohol consumption, the maximum apparent benefit (a 24% reduction of CVD mortality) coming from 2 to 6 drinks per week. The surprising features of the results by Muntwyler et al were that risk reduction was more apparent for non-cardiovascular rather than CVD mortality.23 Unfortunately, most studies of patients with known CHD did not include information about the severity of CHD and other major prognostic cofactors and potential confounders, such as medicines and habitual diet, and an insufficient control of these factors may have affected the overall findings of the studies. It should also be noted that the main alcoholic beverage drunk by British or US patients was not wine, and that assessments of ethanol intake and other confounders were mostly made on the basis of a single measurement, which may have resulted in either misclassification over time or an underestimation of the effect of certain factors, including ethanol drinking.4,22⇓
For these reasons, the Lyon trial14 offered the unique opportunity to examine the association between wine drinking and the risk of CVD complications in a very homogeneous cohort of French middle-aged male survivors of a recent AMI. The study included repeated measures of a number of risk and prognosis factors in secondary prevention of CHD, with a careful assessment of lifestyle and dietary factors. The results show that moderate wine drinking was associated with a significant decrease in the risk of CVD complications. The inverse relation between wine drinking and the risks seemed to be independent of the major predictors of CVD, including the conventional risk factors (smoking, cholesterol, and blood pressure), the severity of the prior AMI, drug treatment, and dietary habits. The homogeneity of that population, with a narrow age range, no women, a small number of heavy drinkers, no binge or irregular drinkers, and the fact that the drinkers consumed almost exclusively wine (which accounted for more than 90% of the ethanol consumed), is an important factor supporting the results of this study.
Our findings do have potential limitations. First, the sample size and total number of events were rather small (the counterpart of studying a very homogeneous population), and the data have to be confirmed in large-scale studies. It is noteworthy, however, that in a recent meta-analysis involving more than 200 000 persons, an inverse relationship was found between wine drinking and vascular diseases.24 Second, we do not have information regarding the drinking habits before AMI, whereas abstainers before AMI were shown to be at particularly high risk.13 It is clear, however, that it is the way of drinking after and not before AMI that is potentially protective during the follow-up. Third, a great deal of prudence is required when attempting to extent our results to other groups, including women, young people (below 45 years of age), or elderly people (over 75 years of age), to the context of primary prevention, or to other alcoholic beverages and drinking patterns. Finally, because of the use of strict exclusion criteria in the Lyon trial and because some patients were lost to follow-up very early, in particular because of early CVD complication, and were excluded from this analysis in the absence of 2 reliable assessments of drinking habits, the studied cohort was at rather low risk and probably not representative of the average post-AMI patients. Caution is therefore required before making recommendations for secondary prevention in general, and further studies are warranted to better define the clinical and biological profile of the patients who would most benefit from moderate wine drinking after AMI. Because essentially all alcohol consumed by these patients was in the form of wine, our data cannot cast light on the specificity of any apparent benefit from non-alcoholic wine ingredients.
In conclusion, despite a small sample size, this investigation suggests that in a very homogeneous population of patients with established CHD, after controlling for many potential confounders, wine drinking is associated with a reduced risk of CVD complications after a recent AMI. Further studies are required to confirm the data, however, and to examine whether the relations found are due to ethanol or to other wine ingredients.
- ↵Pearson TA. Alcohol and heart disease. Circulation. 1996; 94: 3023–3025.
- ↵Gronbaek M, Deis A, Sorensen T, et al. Mortality associated with moderate intakes of wine, beer, or spirits. BMJ. 1995; 310: 1165–1169.
- ↵Djoussé L, Levy D, Murabito JM, et al. Alcohol consumption and risk of intermittent claudication in the Framingham Heart Study. Circulation. 2000; 102: 3092–3097.
- ↵Doll R, Peto R, Hall E, et al. Mortality in relation to consumption of alcohol: 13 years’ observations on male British doctors. BMJ. 1994; 309: 911–918.
- ↵Shaper AG, Wannamethee SG. Alcohol intake and mortality in middle aged men with diagnosed coronary heart disease. Heart. 2000; 83: 394–399.
- ↵de Lorgeril M, Salen P, Martin JL, et al. Mediterranean diet, traditional risk factors, and the rate of cardiovascular complications after myocardial infarction: final report of the Lyon Diet Heart Study. Circulation. 1999; 99: 779–785.
- ↵Hennekens CH, Buring JE. Epidemiology in Medicine. Boston, Mass: Little Brown & Co; 1987.
- ↵Hu FB, Stampfer MJ, Rimm E, et al. Dietary fat and coronary heart disease: a comparison of approaches for adjusting for total energy intake and modelling repeated dietary measurements. Am J Epidemiol. 1999; 149: 531–540.
- ↵Goldberg IJ, Mosca L, Piano MR, et al. Wine and your heart: a science advisory for healthcare professionals from the Nutrition Committee, Council on Epidemiology and Prevention, and Council on Cardiovascular Nursing of the American Heart Association. Circulation. 2001; 103: 472–475.
- ↵Tjonneland A, Gronbaek M, Stripp C, et al. Wine intake and diet in a random sample of 48763 Danish men and women. Am J Clin Nutr. 1999; 69: 49–54.
- ↵Chenet L, McKee M, Leon D, et al. Alcohol and cardiovascular mortality in Moscow: new evidence of a causal association. J Epidemiol Community Health. 1998; 52: 772–774.
- ↵Klatsky AL. Is the drink or the drinker?: circumstantial evidence only raises a probability. Am J Clin Nutr. 1999; 69: 2–3.
- ↵Di Castelnuovo A, Rotondo S, Iacoviello L, et al. Meta-analysis of wine and beer consumption in relation to vascular risk. Circulation. 2002; 105: 2836–2844.