Intake of Beer, Wine, and Spirits and Risk of Stroke
The Copenhagen City Heart Study
Background and Purpose—Alcohol consumption has been associated with a protective effect on risk of ischemic stroke. There may, however, be differences in the effect of beer, wine, and spirits due to properties other than ethanol, a topic that has gained only little attention in stroke research.
Methods—Our analysis was a prospective cohort study of 13 329 eligible men and women, aged 45 to 84 years, participating in the Copenhagen City Heart Study. Information on alcohol habits and a number of socioeconomic and health-related factors was obtained at baseline. During 16 years of follow-up, 833 first-ever strokes occurred. Data were analyzed by means of multiple Poisson regression.
Results—We found indications of a U-shaped relation between intake of alcohol and risk of stroke. In analyses adjusted for age, sex, and smoking, intake of wine on a monthly, weekly, or daily basis was associated with a lower risk of stroke compared with no wine intake (monthly: relative risk [RR], 0.83; 95% CI, 0.69 to 0.98; weekly: RR, 0.59; 95% CI, 0.45 to 0.77; daily: RR, 0.70; 95% CI, 0.46 to 1.00). This effect of wine intake remained after complete adjustment for confounding variables (monthly: RR, 0.84; 95% CI, 0.70 to 1.02; weekly: RR, 0.66; 95% CI, 0.50 to 0.88; daily: RR, 0.68; 95% CI, 0.45 to 1.02). There was no association between intake of beer or spirits on risk of stroke.
Conclusions—The differences in the effects of beer, wine, and spirits on the risk of stroke suggest that compounds in the wine in addition to ethanol are responsible for the protective effect on risk of stroke.
Several epidemiological studies have found that moderate alcohol consumption is associated with a decreased risk of coronary heart disease.1 The association between alcohol and stroke is less well described, although many of the biological mechanisms suggested to influence the risk of cardiac disease are likely to influence the risk of ischemic stroke as well. Established risk factors for stroke such as systolic blood pressure and HDL are modified by the intake of alcohol.2 3 4 5 Furthermore, alcohol consumption leads to an increased tissue insulin sensitivity and has an antithrombotic effect on the coagulation system.6 7 In contrast to studies of ischemic heart disease, stroke studies are often hampered by the fact that the general definition of stroke includes both ischemic and hemorrhagic insults. Alcohol may be associated with a lower risk of cerebral ischemia, while it could cause an increased risk of intracranial hemorrhages.8
Beer, wine, and spirits may have different effects on the risk of cardiovascular disease, indicating that compounds other than ethanol may be of importance.9 10 Particularly, an additional beneficial effect of wine consumption has gained attention and has been suggested as one of many possible explanation of the “French paradox”: a low incidence of cardiovascular disease in the French population despite an unfavorable exposure to known cardiovascular risk factors.11
In the present study we examined the influence of alcohol intake and the different types of alcohol on risk of first-ever stroke in a large Danish prospective cohort study.
Subjects and Methods
The Copenhagen City Heart Study was initiated in 1976 and included 19 698 persons living in Østerbro and Nørrebro in Copenhagen, Denmark. The participants were randomly chosen within age and sex strata and invited by letter to 3 health examinations held in 1976–1978, 1981–1983, and 1991–1994. Less than 2% of the population was of nonwhite origin. There were 371 persons who died before the examination, and of the remaining 19 327 people, there were 5104 (26.4%) nonresponders and 14 223 (73.6%) responders, in all age groups. During the 16-year follow-up period, data were available for 13 329 eligible subjects aged 45 to 84 years. A total of 55 persons (0.04%) were lost to follow-up; of these, 53 were due to immigration, and there was no information regarding the other 2 persons. A detailed description of the examinations has previously been published.12
The participants were asked in multiple choice form whether they drank beer (bottles), wine (glasses), or spirits (units). The choices were “never/hardly ever,” “monthly,” “weekly,” or “daily.” Information for each person regarding intake of beer, wine, and spirits was entered into the model simultaneously.
The subjects were then classified according to the total weekly intake of alcohol: <1 beverage; 1 to 7 beverages; 8 to 14 beverages; 15 to 21 beverages; 22 to 41 beverages; and ≥42 beverages. One bottle of beer contains 12 g of alcohol, and this was considered the average for the other types of drinks. Abstinence due to treatment with drugs (disulfiram) was noted, and those subjects were excluded from the study (n=16).
The following variables were assumed to confound the analyses: smoking (never smokers; ex-smokers; 1 to 10, 11 to 20, and >20 cigarettes per day); sex (men and women); monthly income in 1976 to 1978 (monthly income in 1976–1978 (<4000, 4000 to 10 000, and >10 000 Danish crowns, which is approximately equivalent to <666, 666 to 1667, and >1667 US dollars for exchange rates in 1977); physical activity in leisure time (almost completely physically passive or light physical activity <2 h/wk, light physical activity 2 to 4 h/wk, light physical activity >4 h/wk or more exhausting physical activity 2 to 4 h/wk, and exhausting physical activity >4 h/wk or regular hard training); educational level (education <8, 8 to 11, or >11 years); presence of diabetes mellitus (yes or no); cholesterol (<8 and >8 mmol/L)13 ; triglycerides in millimoles per liter; and body mass index (<20, 20 to 24, ≥25 and <30, and ≥30 kg/m2). Systolic blood pressure (millimeters of mercury) was regarded as a possible effect mediator of alcohol.
Information about prior stroke (hospitalized as well as nonhospitalized events) was obtained at the health examinations. Furthermore, the Danish National Patient Register and the Central Death Register were used to provide information about all persons with regard to stroke hospitalization and death. The International Classification of Diseases, codes 430 to 438 (8th revision), was used to identify persons who had had a stroke. The World Health Organization definition of stroke was used for validation: an acute disturbance of focal or global cerebral function with symptoms lasting >24 hours or leading to death with presumably no reasons other than of vascular origin.14 Subjects who suffered a stroke before the beginning of the study period were excluded from the analyses. This investigation is based on data of risk of first-ever stroke, thus excluding 178 persons who had had a stroke before the start of the study.
To distinguish between ischemic infarction, intracerebral hemorrhages, and subarachnoid hemorrhages, either CT or MR scan, autopsy, spinal fluid examination, or operation description was necessary. If a scan did not visualize an infarction but the person had symptoms that met the criteria of the stroke definition, a diagnosis of ischemic infarction was made.15 The diagnosis of stroke was not applied in cases in which a scan had revealed signs of prior cerebrovascular disease but without history of symptoms. Unspecified stroke relates to a patient for whom no information of stroke subtype was available.
The follow-up period was 16 years (from January 1, 1977, to December 31, 1992). We used data from participants aged 45 to 84 years and who attended the initial study examination; the present analyses are therefore based on 6067 men and 7262 women. The age reduction was applied because most strokes occurred in this age group, because very young people may differ in respect to susceptibility to alcohol,16 and to diminish uncertainty about the reliability of stroke diagnoses among very old persons.17
The data were analyzed by means of multiple Poisson regression analysis,18 19 and the statistical software package STATA was used.20 Tests for linearity of variables and interactions were based on log likelihood ratio tests at the 5% level. Although income was considered a potential confounder, the variable was omitted in the final analyses because it had no statistical effect in the present set of analyses. Adjustment for age was done by 4-year lexis groups,19a ie, a subject contributed to the age effect only with the years within the relevant age strata and was included in the following age group when that was reached. This implies that a person reaching the age limits within the time frame of this study could be included or excluded. The incidence was assumed to be constant within each 4-year age interval.
During the 16 years of follow-up of 13 329 eligible persons, 833 strokes occurred: 442 (53%) unspecified, 310 (37%) ischemic infarctions, 47 (6%) intracerebral hemorrhages, and 34 (4%) subarachnoid hemorrhages. There were 346 events (42%) among women and 487 (58%) among men. One hundred fifty-nine (19%) of the stroke patients died within the initial 28 days.
There was a high proportion of men in the groups consuming high amounts of alcohol, and the proportion of smokers increased with higher alcohol consumption (Table 1⇓). The proportion of subjects with cholesterol levels >8 mmol/L fell with a high alcohol intake, as did the proportion of subjects receiving antihypertensive treatment. Both mean concentration of triglycerides and mean systolic blood pressure were highest in the groups with high intake of alcohol (Table 1⇓).
In analyses adjusted for age, sex, and smoking, subjects drinking <1 unit of alcohol per week had a significantly higher risk of stroke compared with the reference group of subjects drinking 1 to 7 units per week (relative risk [RR], 1.33; 95% CI, 1.10 to 1.62) (Table 2⇓). Subjects who drank ≥42 units per week also had an increased risk of stroke (RR, 1.56; 95% CI, 1.15 to 2.23). There were no statistically significant differences between the intermediate consumption groups and light drinkers. The increased risk in the low- and high-consumption groups indicates a U-shaped relation.
After complete adjustment for confounding factors, the risk of stroke was statistically insignificant between the reference group and the other groups. However, the estimates did not change substantially; only the CIs became wider, and thus the U-shaped relation remained. Exclusion of systolic blood pressure increased the risk slightly in the high-consumption group. Inclusion of subarachnoid hemorrhage and intracerebral hemorrhage had a modest effect on the results, with a tendency of altering the RR estimates toward the value of 1. First-order interaction between alcohol consumption and sex was statistically insignificant (χ2=3.14, 5 df; P=0.68).
Beer, Wine, or Spirits
Men drank beer and spirits more often than women, whereas there was no difference between the sexes in regard to intake of wine (Table 3⇓). There was a low proportion of persons reporting a frequent consumption of wine who had had <8 years of education. There were no differences between the groups with regard to body mass index, mean concentration of triglycerides, proportion with cholesterol concentration >8 mmol/L, and physical activity in leisure time (not shown).
The effect of drinking beer, wine, and spirits on risk of stroke varied among the 3 types of alcohol (Table 4⇓). In simple adjusted analyses, subjects who drank wine had a statistically significant decreased risk of stroke in all 3 frequency groups compared with subjects who never/hardly ever drank wine. In contrast, no statistically significant effect of drinking either beer or spirits was found in either of the frequency groups.
After complete adjustment for confounding factors, subjects who drank wine weekly had a significantly lower risk of stroke, while the difference was marginally significant for subjects reporting a daily consumption of wine (P=0.06). Intake of neither beer nor spirits was associated with risk of stroke. Exclusion of systolic blood pressure in the model had only a modest effect and did not change the overall results. Inclusion of intracerebral hemorrhage and subarachnoid hemorrhage did not alter the estimates substantially. The differences between the various frequency groups could be expressed as linear, with the RR (95% CI) for beer, wine, and spirits being 1.03 (0.96 to 1.11), 0.85 (0.77 to 0.94), and 0.99 (0.91 to 1.08), respectively. There was no interaction between sex and wine consumption (χ2=0.004, 1 df; P=0.95), wine and beer intake (χ2=0.232, 3 df; P=0.97), or intake of wine and spirits (χ2=2.766, 3 df; P=0.43).
Our analyses suggest, in accordance with other studies, that there is a beneficial effect of intake of a small amount of alcohol on risk of stroke. In addition, we have found that this effect is strongest among wine drinkers, since subjects who drank wine had a lower risk of stroke than subjects who never drank wine. The risk of stroke among wine drinkers was not influenced by beer or spirits consumption.
The sampling procedures of the study population ensured that a representative proportion of the Copenhagen population was selected for the investigation. All data on exposure were collected before the end points occurred, precluding recall bias. Validation of stroke events was possible for all participants regardless of socioeconomic status, thereby avoiding selection bias of the case ascertainment. Nonfatal nonhospitalized stroke events were detected at the health examinations. Therefore, it is likely that only few events were missed in the analyses. Any such nuisance would lead to a lower estimated RR, and the results are therefore to be considered conservative estimates of the true risk differences.
Alcohol consumption was self-reported, which may raise questions regarding the validity of the data. It is possible that the true alcohol consumption was underreported, especially among heavy drinkers.21 In case of underreporting, the true higher risk in the group with high alcohol consumption would relate to an even higher alcohol consumption. Similarly, for the wine results, underreporting would consequently reflect that daily wine consumption is beneficial and not only weekly wine intake.
The validity of self-reported alcohol intake in a simple questionnaire, like the one used in this study, has been compared with a more detailed dietary interview, and there was only little or no systematic variation for the 3 types of alcoholic beverages.22 Therefore, it is unlikely that bias in alcohol reporting is responsible for the findings of differences in risk of stroke among drinkers of beer, wine, and spirits.
Information regarding diet, a potential confounder,23 has not been obtained in the Copenhagen City Heart Study. If differences in diet could explain the results between the 4 wine groups, it would require that drinking wine was directly associated with a beneficial intake of food ingredients, independently from potential confounders already controlled for. Furthermore, diet should be quite strongly inversely related to risk of stroke. We consider that to be an unlikely explanation for the results.
Other studies have addressed the question of a relation between alcohol consumption and stroke in different settings.24 There are discrepancies in the statistical strength of the results; some studies found a protective effect of moderate alcohol consumption,8 25 26 27 28 29 whereas other authors concluded that there was no effect.30 High alcohol consumption may be associated with an increased risk of stroke, and, as in our study, there is evidence that at least some of this effect is mediated through systolic blood pressure.4 5 8 Furthermore, high alcohol consumption has been associated with an increased risk of hemorrhagic stroke.31 Our study confirms these findings, but the effect of alcohol consumption was significant only in analyses controlling for age, sex, and smoking. When lifestyle factors were entered into the model, the association between alcohol intake and stroke was attenuated, and the nonsignificant results are likely merely to reflect the powerful confounding effect of these covariates rather than a lack of power. It is possible that the inconsistency in the literature on the effect of alcohol on cerebrovascular disease is partly due to differences in adjustment for some of the lifestyle factors.
Alcohol is purported to influence risk of stroke through an action on the coagulation system and/or atherogenesis similar for both sexes. There were few women in the high-consumption groups and consequently few stroke events (one in the highest-consumption group and 3 in the second highest group). Therefore, the results in these alcohol categories are largely based on stroke events that occurred in men. In the other groups and in the analyses for beer, wine, and spirits, a sufficient number of stroke events in both men and women indicated that there were no differences between analyses stratified by sex or grouped.
The follow-up period covers the period in which neuroimaging was introduced as a routine diagnostic tool in Danish hospitals, which is reflected by a relatively high proportion of unspecified strokes in this study. Our stroke diagnosis was based on clinical signs, and neuroimaging had no or little effect.32 A large proportion of unspecified strokes may explain the modest effect of exclusion of hemorrhagic stroke in this investigation. Although a vast majority of these strokes can be expected to be of ischemic origin, ≈10% may have been hemorrhagic. The possible misclassification will lead to a negation of the difference. Therefore, although the results in the analyses excluding hemorrhagic strokes support a differential effect of intake of alcoholic beverages on risk of ischemic and hemorrhagic strokes, they are presumably too low.
A few authors have discussed whether the effect of alcohol on risk of cardiovascular disease depends on the type of alcohol ingested.9 33 34 We found that weekly and daily consumption of wine was associated with lower risk of stroke. Furthermore, the relation could be explained as a linear relationship that was statistically significant, indicating a dose-response effect. There was no interaction between wine consumption and intake of beer and spirits, and therefore the protective effect on risk of stroke is solely connected with intake of wine and not with a certain amount of alcohol. In a recent correlation study from Spain, it was found that low wine consumption, together with sedentary lifestyle, was associated with increased cerebrovascular death.35 In the Nurse’s Health Study,8 which included women aged 34 to 59 years, consumption of wine was also associated with a reduced risk of stroke, but the investigation was based on relatively few events (66 ischemic strokes, 12 intracerebral hemorrhages, 28 subarachnoid hemorrhages, and 14 unclassified). Wine contains flavonoids and tannins, which are components presumed to prevent cardiovascular disease; this may be a possible explanation of the apparent protective effect in this study.36 37 38 On the other hand, it has been suggested that the beneficial effect of wine consumption is merely related to drinking pattern. Wine may be consumed with meals to a greater extent than beer and spirits; the latter 2 may be consumed more irregularly throughout the day. These differences in “timing” may be important. In a recent study, intake of alcohol was shown to reduce potential atherogenic postprandial alterations of the blood.39
Our data do not allow discrimination between ex-drinkers and life-long abstainers in the reference groups. The lower risk of stroke found among subjects reporting a weekly consumption of wine could be due to a higher risk among ex-drinkers. However, our reference groups are composed of people who reported a low alcohol consumption rather than people who are teetotalers, which reduces this problem considerably. Furthermore, in a large study it was found that ex-drinkers had higher risk of mortality from cardiovascular and coronary artery diseases in unadjusted analyses but not in analyses adjusted for age, sex, race, body mass index, marital status, and education.40 In this study we have adjusted for additional confounding factors, and if ex-drinkers should bias the results, it was to be expected that in addition to wine consumption, intake of beer and spirits would also have been associated with a reduced risk of stroke, which was not found.
Our data indicate that intake of wine is associated with lower risk of stroke. We cannot determine whether these results are due to components in wine in addition to ethanol or whether this is an effect due to intake of ethanol with meals. Whatever the biological mechanism may be, the consistency of the results, the lack of obvious biases, and the biological plausibility suggest that there may be a beneficial effect of intake of wine on risk of stroke.
This study was supported by funds from Direktør Werner Richter og Hustrus Legat, the Danish National Board of Health, and Copenhagen University. The Copenhagen City Heart Study was funded by the Danish Heart Foundation. We are grateful to G. Jensen, J. Nyboe, and A.T. Hansen, who, together with P. Schnohr, initiated and conducted the Copenhagen City Heart Study. We acknowledge M. Appleyard for her support and assistance.
- Received June 25, 1998.
- Revision received August 18, 1998.
- Accepted September 7, 1998.
- Copyright © 1998 by American Heart Association
Maclure M. Demonstration of deductive meta-analysis: ethanol intake and risk of myocardial infarction. Epidemiol Rev. 1993;15:328–351.
Lieber CS. Metabolism and metabolic effects of alcohol. Med Clin North Am. 1984;68:3–33.
Facchini F, Chen YD, Reaven GM. Light to moderate alcohol intake is associated with enhanced insulin sensitivity. Diabetes Care. 1994;17:115–119.
Grønbæk M, Deis A, Sørensen TIA, Becker U, Schnohr P, Jensen G. Mortality associated with moderate intakes of wine, beer, or spirits. BMJ. 1995;310:1165–1169.
Rimm EB, Klatsky AL, Brobbee D, Stampfer MJ. Moderate alcohol consumption and reduced risk of coronary heart disease: is the effect due to beer, wine, or spirits? BMJ. 1996;312:731–736.
Appleyard M, Hansen T, Schnohr P, Jensen G, Nyboe J. The Copenhagen City Heart Study: Østerbroundersøgelsen: a book of tables with data from the first examination (1976-78) and a five year follow-up (1981-83). Scand J Soc Med. 1989;170(suppl 41):1–160.
Lindenstrøm E, Boysen G, Nyboe J. Influence of total cholesterol, high density lipoprotein cholesterol and triglycerides on risk of cerebrovascular disease. BMJ. 1994;1994:11–15.
Lindgren A, Norrving B, Rudling O, Johansson B. Comparison of clinical and neuroradiological findings in first-ever stroke. Stroke. 1994;25:1371–1377.
Klatsky AL, Armstrong MA, Friedman GD. Alcohol and mortality. Ann Intern Med. 1992;117:646–654.
Bonita R, Broad JB, Anderson NE, Beaglehole R. Approaches to the problems of measuring the incidence of stroke. Int J Epidemiol. 1995;24:535–542.
Grønbæk M, Deis A, Sørensen TIA, Becker U, Borch-Johnsen K, Müller C, Schnohr P, Jensen G. Influence of sex, age, body mass index, and smoking on alcohol intake and mortality. BMJ. 1994;308:302–306.
McCullagh P, Nelder JA. Generalized Linear Models. New York, NY: Chapman and Hall; 1983.
Clayton D, Hills M. Statistical Models in Epidemiology. New York, NY: Oxford University Press; 1993.
Statistical Software, Release 5.0. College Station, Tex: Stata Corp; 1997.
Poikolainen K. Underestimation of recalled alcohol intake in relation to actual consumption. Br J Add. 1985;80:215–216.
Klatsky AL, Armstrong MA, Friedman GD. Alcohol use and subsequent cerebrovascular disease hospitalizations. Stroke. 1989;20:741–746.
KiyoharaY, Kato I, Iwamoto H, Nakayama K, Fujishima M. The impact of alcohol and hypertension on stroke incidence in a general Japanese population. Stroke. 1995;26:368–372.
Palomäki H, Kaste M. Regular light-to-moderate intake of alcohol and the risk of ischemic stroke. Stroke. 1993;24:1828–1832.
Rodgers H, Aitken PD, French J, Curless RH, Bates D, James OFW. Alcohol and stroke. Stroke. 1993;24:1473–1477.
Wannamethee SG, Shaper AG. Patterns of alcohol intake and risk of stroke in middle-aged British men. Stroke. 1996;27:1033–1039.
Donahue RP, Abbott RD, Reed DM, Yano K. Alcohol and hemorrhagic stroke. JAMA. 1986;297:519–522.
Truelsen T, Prescott E, Grønbæk M, Schnohr P, Boysen G. Trends in stroke incidence. Stroke. 1997;28:1903–1907.
Artalejo FR, Guallar-Castillón P, Gutiérrez F, Banegas JR, Calero JR. Socioeconomic level, sedentary lifestyle, and wine consumption as possible explanations for geographic distribution of cerebrovascular disease in Spain. Stroke. 1997;28:922–928.
Caarnacini A, Arfelli G. Selected nutritional components of wine. Alcologia. 1994;6:41–49.
Locher R, Suter PM, Vetter W. Ethanol suppresses smooth muscle cell proliferation in the postprandial state: a new antiatherosclerotic mechanism of ethanol? Am J Clin Nutr. 1998;67:338–341.