Alcohol Consumption and Stroke Mortality
20-Year Follow-up of 15 077 Men and Women
Background and Purpose Since stroke is a principal cause of death in elderly people, we analyzed the association between alcohol and stroke mortality in a cohort of 15 077 middle-aged and older men and women.
Methods Data on alcohol habits were obtained from a questionnaire in 1967. The subsequent 20 years yielded 769 deaths from stroke, of which 574 were ischemic. Relative mortality risks (RR) were estimated from logistic regression analyses with lifelong alcohol abstainers as a reference group. Adjustments were made for age and smoking.
Results No association was found between alcohol intake and hemorrhagic stroke. An elevated risk of ischemic stroke was found for men who drank infrequently, that is, a few times a year or less often (RR, 2.0; 95% confidence interval [CI], 1.3 to 3.2), for those who were intoxicated now and then (RR, 1.8; 95% CI, 1.1 to 2.8), and for those who reported “binge” drinking a few times in the year or less often (RR, 1.6; 95% CI, 1.1 to 2.5). Among women only ex-drinkers had an elevated risk of dying of ischemic stroke (RR, 3.3; 95% CI, 1.5 to 7.2). The risk was reduced for women who had an estimated average consumption of 0 to 5 g pure alcohol per day (RR, 0.6; 95% CI, 0.5 to 0.8); for those who did not drink every day (RR, 0.7; 95% CI, 0.5 to 0.9); and for those who never “went on a binge” (RR, 0.6; 95% CI, 0.5 to 0.8) or became intoxicated (RR, 0.7; 95% CI, 0.5 to 0.9).
Conclusions Drinking habits were associated only with deaths from ischemic stroke, and the risk patterns were different for men and women. In analyses, ex-drinkers should not be included with lifelong abstainers, since the former tend to run high health risks.
Studies on the relationship between alcohol consumption and cerebrovascular diseases, including stroke, have shown partly inconsistent results. There is evidence of an association between heavy drinking and stroke, and a positive dose-response relationship has been found between alcohol intake and hemorrhagic stroke.1 2 3 4 Concerning ischemic stroke, which is the major subtype of stroke, dose-response, inverse, or U-shaped associations have been reported.5 6 7 Alcohol consumption is believed to increase the occurrence of a hemorrhagic stroke mainly as a result of alcohol-induced hypertension.8 The reduced risk of ischemic stroke with light or moderate consumption has been attributed to reduced clotting of thrombocytes and reduced fibrinogen concentration, which reduce the risk of formation of thrombi in blood vessels.2 9
The association has generally been similar in studies with mortality as an end point and in studies of incidence based on hospital discharge records.10 11 Stroke is the third leading cause of mortality in most industrialized countries and a major cause of hospitalization, especially in older age groups.8 The evidence relating alcohol to stroke is more extensive in regard to the risk with heavy alcohol consumption than with moderate consumption, but relatively little has been done to investigate the role of light or moderate drinking.
Since stroke accounts for more female deaths (approximately 15%) than male deaths (approximately 10%),9 it is important to analyze the role of alcohol consumption in different types of stroke in women. Several studies have focused on men only, yet there is a tendency to extrapolate the results to women.1 3 4 5
The aim of the present study was to investigate the association, if any, between alcohol habits and stroke mortality (hemorrhagic and ischemic) in men and women. Alcohol habits were defined in terms of (1) average quantity of alcohol intake, ie, grams of pure alcohol per day as estimated from the quantities of alcoholic beverages consumed (beer, wine, liquor); (2) frequency of alcohol intake, ie, on a daily versus not-daily basis; (3) frequency of high-alcohol-consumption episodes; and (4) frequency of “feeling intoxicated.”
For this purpose, self-reported alcohol consumption data in a cohort of 15 077 men and women were linked with the national cause-of-death register during a period of 20 years.
Subjects and Methods
A cohort of subjects born between 1886 and 1925 was identified from the Swedish population-based twin register. The register, established in 1961 for the purpose of epidemiological investigations of diseases, has been used in several studies.12 13 14 In 1967 to 1968, baseline data for this follow-up were obtained from 15 077 twin individuals by a self-administered questionnaire. (Since genetic aspects were not the issue here, the fact that the cohort contained twins was not used in the present study.) In addition to sociodemographic characteristics, health status, and physical activity, the questionnaire included items on smoking and alcohol habits. The following questions concerned drinking habits: “Did you drink beer, wine, or liquor at any time during the past year? Yes/No” and if no: “Did you drink alcohol before that time? Yes/No” and if yes: “When did you stop drinking alcohol? In the year … ” and “Why did you stop? Because of trouble with alcohol/For another reason.”
Abstainers could thus be categorized into lifelong abstainers and former drinkers. Those who had answered “Yes” to the question of whether they had drunk beer/wine/liquor during the preceding year were asked: “How often do you usually drink beer? Almost every day/Occasionally in the week/A couple of times a month/A few times in the year/More rarely/Never” and “How much do you usually drink when you drink beer? A couple of glasses/1 bottle/2 bottles/3 bottles or more.” The same questions were asked concerning wine and liquor. The options for the quantities of wine were: A couple of glasses/ bottle/ to 1 bottle/1 bottle or more. For liquor they were: Less than 5 cL/5 to 15 cL/15 to 35 cL/More than 35 cL (5 cL being equivalent to one drink). The average consumption of grams of pure alcohol per day was estimated from the answers to these questions. Drinking behavior was further investigated by two questions: “It happens relatively often that on special occasions one drinks more than usual. How often do you drink alcohol corresponding to at least half a bottle of spirits or two bottles of wine on the same occasion? Occasionally in the week/A couple of times a month/Occasionally in the month/A few times in the year/More rarely/Never” and “How often do you feel intoxicated? Often/Now and then/Rarely/Never.” These two variables, “binge drinking” and “feel intoxicated,” were analyzed separately and in combination.
Smoking habits were also recorded in detail. In the analysis, smoking status was divided into smoker (1 to 9 or ≥10 cigarettes per day) and nonsmoker. Nonsmokers were defined as persons smoking less than 5 to 10 packs of cigarettes, or the equivalent in cigars or pipe tobacco, during their entire life (1 cigarette=1 g tobacco).
The cohort was followed in the national cause-of-death register from 1968 through 1987. All death certificates of persons registered in Swedish parishes at the time of death were reviewed and recorded at Statistics Sweden. The cause-of-death register is more than 99% complete15 and has a validity of 97% in cases of cerebrovascular diseases.14 During the follow-up, 6886 deaths occurred, of which 769 were attributed to stroke as the underlying or contributory cause of death (cerebrovascular disease, codes 430 through 438 of the International Classification of Diseases, 8th Revision). Hemorrhagic stroke caused 195 deaths, and ischemic stroke caused 574 deaths. Permission to link the two data sets was obtained from the Data Inspection Board and the faculty’s Research Ethics Committee for the study.
The relative risk (RR) of mortality (cumulative incidence) was estimated from a logistic regression analysis with 95% confidence intervals (CI) with the use of the SAS catmod procedure.16 We included age and smoking in the multivariate model after testing each variable separately in relation to alcohol habits in univariate analyses. Age was used as a continuous variable. Lifelong alcohol abstainers were used throughout as the reference group. Ex-drinkers who reported that they had stopped drinking because of the problems it caused were initially analyzed separately. Since they were few, particularly among women, and since an analysis showed no difference between them and former drinkers who had stopped for other reasons, these two groups were treated together in the main analyses.
Table 1⇓ shows the cohort distributed by age, sex, drinking and smoking habits at baseline, the total number of deaths, and deaths from stroke during the follow-up. As can be seen, the level of alcohol intake was generally low; only 9% of men (640/7089) and 0.8% of women (61/7988) in the cohort had an estimated average alcohol intake of 15 g/d of alcohol or more. Less than 1% of men (52/7089) and 0.08% of women (6/7988) consumed more than 40 g/d of alcohol, that is, more than 3 (US standard) drinks per day. The frequency of intake of alcoholic beverages was considerably higher in men than in women: 25% (1746/7089) of the men versus 7% (578/7988) of the women stated that they drank beer, wine, and/or liquor practically every day.
Drinkers Versus Nondrinkers
The risk of dying of stroke was not increased for male consumers of alcohol as a group compared with lifelong abstainers: overall RR, 1.2; 95% CI, 0.8 to 1.6 (Table 2⇓). Nor was it high for female consumers; the latter ran a 30% lower risk of stroke (RR, 0.7; 95% CI, 0.5 to 0.8) than did the female lifelong abstainers (Table 3⇓). Female ex-drinkers, on the other hand, had more than triple the risk of ischemic stroke (RR, 3.3; 95% CI, 1.5 to 7.2). Male ex-drinkers had an insignificantly increased risk.
Quantity of Alcohol
For men, the quantity of alcohol was not associated with the risk of stroke (Table 2⇑). Women had the lowest risk at 0 to 5 g/d; the risk increased somewhat with a higher intake. However, this dose-response relation did not reach significance (Table 3⇑).
Frequency of Alcohol Intake
Table 2⇑ shows that men who drank infrequently (beer, wine, or liquor a few times a year or more rarely) ran twice the risk of dying of ischemic stroke than lifelong abstainers (RR, 2.0; 95% CI, 1.3 to 3.2). On the other hand, women who reported infrequent drinking had a 40% risk reduction in ischemic stroke (RR, 0.6; 95% CI, 0.4 to 0.9); those who drank once a week or once a month had a 30% risk reduction (RR, 0.7; 95% CI, 0.5 to 0.9); and those who drank almost daily had an RR of 0.8 (95% CI, 0.5 to 1.4; Table 3⇑).
Frequency of High-Alcohol-Consumption Episodes
Men who reported binge drinking on rare occasions (a few times per year or more rarely) had an elevated risk of ischemic stroke (RR, 1.6; 95% CI, 1.1 to 2.5). Women who reported never binge drinking had a reduced risk of ischemic stroke (RR, 0.6; 95% CI, 0.5 to 0.8).
Frequency of Feeling Intoxicated
Men who reported that they often or sometimes felt intoxicated had a greater risk of ischemic stroke (RR, 1.8; 95% CI, 1.1 to 2.8). Women who never felt intoxicated had a reduced risk of dying of ischemic stroke (RR, 0.7; 95% CI, 0.5 to 0.9).
The patterns were similar when the two variables were combined: men who reported binge drinking and/or intoxication had an elevated risk of ischemic stroke, while women who had never experienced such episodes were less at risk (Table 4⇓).
This study suggests an elevated risk of death due to ischemic stroke for middle-aged and elderly men who were infrequent drinkers, for men who reported binge drinking on rare occasions, and for men who reported that now and then they felt intoxicated. No positive or inverse dose-response relationship was seen. Lifelong abstainers were used as the reference group. Stroke mortality in male ex-drinkers did not differ significantly from that of consumers of alcohol.
For middle-aged and elderly women a different pattern emerged. While ex-drinkers had the highest risk—threefold—of dying of ischemic stroke compared with lifelong abstainers, there was a considerably reduced risk of ischemic stroke in women with the lowest drinking levels. A low drinking level in this case meant, at most, 5 g/d of pure alcohol, which corresponds to two or three glasses of wine a week. Although they merged fatal and nonfatal stroke, Stampfer at al7 in their study of 87 000 female nurses found a similarly decreased risk of ischemic stroke among those drinking moderate amounts of alcohol (1.5 to 14.9 g/d). At the same time, their data suggested that moderate alcohol consumption may increase the risk of hemorrhagic stroke among middle-aged women. In our study no association was found regarding hemorrhagic stroke in men or women. However, since ischemic stroke accounted for 75% of all strokes, associations with ischemic stroke also had an effect on the findings for total stroke. This shows the need to analyze different subtypes of stroke separately; otherwise, the effects found may be “diluted” since risk factors such as alcohol may have different relationships with different subtypes of stroke.
Age and smoking were controlled for in the analyses as potential confounders. In contrast to some other studies,3 7 we did not control for hypertension or cardiovascular diseases since, as Beaglehole and Jackson8 point out, if alcohol consumption increases stroke occurrence through alcohol-induced hypertension, controlling for blood pressure in the analyses will obscure the relationship between alcohol and stroke.
Drinking habits were measured by a questionnaire regarding the quantity and frequency of alcohol intake, binge drinking, and subjective feelings of intoxication. The validity of self-reported drinking habits, ie, the extent to which accurate information is provided by the respondent, has often been debated. In reality, however, there is still no better method for investigating this domain of human behavior, which is a good reason for trying to improve questionnaire and interview methods.17 Respondents may overestimate or underestimate the amount drunk, and the amount may also vary over time. Nevertheless, as Doll et al18 point out, studies of alcohol use and subsequent mortality can yield useful results, particularly among people in middle or old age with long-established habits. In our study no validation of the self-reported data was attempted by hematologic or biomedical markers of alcohol intake because such tests were not in use in 1967 to 1968, when the data were compiled for the present follow-up. The autopsy rate was 23%. In Sweden an autopsy is performed on practically all persons who die outside of the hospital to document the cause of death. Analyses of the relationship between alcohol habits and risk of stroke were attempted separately on these cases and showed the same tendencies as for the combination. However, the figures in subgroups were too small for safe conclusions.
The level of alcohol intake in the cohort was generally low; less than 1% reported a consumption of 40 g/d or more, and 20% of the men and 44% of the women drank no alcohol at all at the baseline of the study. This may be due to the fact that the subjects were elderly or because that generation had a more restrictive attitude toward alcohol. More than 20 years ago, this was particularly true of women. Heavy drinkers may also be less willing to answer the questionnaire about lifestyle. However, even if the cohort may not be quite representative of a “normal” population, it can be used for our purpose of investigating the relationship between various amounts of alcohol intake and mortality from stroke.
In the main analyses, we considered persons who consume alcohol occasionally during the year or more rarely as drinkers. We also performed analyses in which we excluded these persons from the consumers. This lowered the stroke risk, although only marginally and without influencing the original trends, for men in the category of 0 to 5 g/d. There is therefore a possibility that men in the cohort who reported rare drinking, binge drinking, and feeling intoxicated were periodic heavy drinkers. These men, whose average amount of alcohol consumed may not be very high when considered per day, had a clearly increased risk of ischemic stroke. According to Hillbom,10 even occasional heavy drinking may increase the risk of stroke. However, although several studies suggest an association between heavy alcohol intake and an increased risk of stroke,1 2 3 4 5 the results thus far are not univocal internationally. For example, in two recent studies, one from Britain and one from Italy, no association was found.19 20
Some studies suggest a possible health-protective effect of light and moderate alcohol use, particularly against ischemic stroke,7 21 but it has also been argued that this may be an effect of the control group, usually nondrinkers, in whom abstention may be associated with other risk factors for disease.22 23 In our analyses, ex-drinkers were separated from lifelong abstainers. Former drinkers, at least the women, in fact had the highest risk of ischemic stroke. Also, in a cohort study of male Japanese physicians in which ex-drinkers were analyzed separately, they appeared to have the highest risk of ischemic stroke.4
In conclusion, in this cohort of light and moderate alcohol drinkers, middle-aged and older men and women had different risk patterns concerning ischemic stroke. We found no associations with hemorrhagic stroke. The results of the analyses are greatly dependent on the composition of the reference group. Therefore, it is important to distinguish between total alcohol abstainers and ex-drinkers since, in this study, female ex-drinkers were found to have an increased mortality risk.
This study was supported in part by the John D. and Catherine T. MacArthur Foundation and by grants from the Swedish Medical Research Council (B91-27X-09530-01 and K93-27P-09761-03A) and the Swedish Council for Social Research (92-0300:3C).
- Received April 24, 1995.
- Revision received June 29, 1995.
- Accepted July 3, 1995.
- Copyright © 1995 by American Heart Association
Shaper AG, Phillips AN, Pocock SJ, Walker M, Macfarlane PW. Risk factors for stroke in middle aged British men. Br Med J. 1991;302:1111-1115.
van Gijn J, Stampfer MJ, Wolfe C, Algra A. The association between alcohol and stroke. In: Verschuren PM, ed. Health Issues Related to Alcohol Consumption. Washington, DC: ILSI Press; 1993:43-80.
Kono S, Ikeda M, Tokudome S, Nishizumi M, Kuratsune M. Alcohol and mortality: a cohort study of male Japanese physicians. Int J Epidemiol. 1986;15:527-532.
Gorelick PB, Rodin MB, Langenberg P, Hier DB, Costigan J. Weekly alcohol consumption, cigarette smoking and the risk of ischemic stroke: results of a case-control study at three urban medical centers in Chicago, Illinois. Neurology. 1989;39:339-343.
Beaglehole R, Jackson R. Alcohol, cardiovascular diseases and all causes of death: a review of the epidemiological evidence. Drug Alcohol Rev. 1992;11:275-290.
Cohen GR, Duffy JC. Alcohol consumption and cerebrovascular disease. In: Duffy JC, ed. Alcohol and Illness. Edinburgh, Scotland: Edinburgh University Press; 1992:52-63.
Hillbom ME. What supports the role of alcohol as a risk factor for stroke? Acta Med Scand. 1987;717(suppl):93-106.
Camargo CA. Moderate alcohol consumption and stroke: the epidemiologic evidence. Stroke. 1989;20:1611-1626.
Cederlöf R, Friberg L, Lundman T. The interaction of smoking, environment and heredity and their implications for disease etiology. Acta Med Scand. 1977;612(suppl):1-128.
Gerhardsson M, Floderus B, Norell SE. Physical activity and colon cancer risk. Int J Epidemiol. 1988;17:743-746.
Höglund D. Quality control of death certificates. In: Promemoria. Stockholm, Sweden: Statistics Sweden; 1983:5.
SAS Institute. SAS User’s Guide Statistics, Version 6. Cary, NC: SAS Institute; 1990: vol 1.
Doll R, Peto R, Hall E, Wheatley K, Gray R. Mortality in relation to consumption of alcohol: 13 years’ observation on male British doctors. Br Med J. 1994;309:911-918.
Ben-Shlomo Y, Markowe H, Shipley M, Marmot MG. Stroke risk from alcohol consumption using different control groups. Stroke. 1992;23:1093-1098.
Palomäki H, Kaste M. Regular light-to-moderate intake of alcohol and the risk of ischemic stroke: is there a beneficial effect? Stroke. 1993;24:1828-1832.
Shinton R, Sagar G, Beevers G. The relation of alcohol consumption to cardiovascular risk factors and stroke: the West Birmingham Stroke Project. J Neurol Neurosurg Psychiatry. 1993;56:458-462.