Association between alcohol consumption and mortality, myocardial infarction, and stroke in 25 year follow up of 49 618 young Swedish menBMJ 1999; 319 doi: http://dx.doi.org/10.1136/bmj.319.7213.821 (Published 25 September 1999) Cite this as: BMJ 1999;319:821
- Anders Romelsjö, associate professor (, )
- Anders Leifman, statistician
- Karolinska Institutet, Department of Public Health Sciences and Centre for Alcohol and Drug Prevention, Novum, S-141 57 Huddinge, Sweden
- Correspondence to: A Romelsjö
- Accepted 17 June 1999
Several epidemiological studies have shown that moderate alcohol consumption is associated with reduced mortality from cardiovascular diseases in middle aged and elderly subjects,1 but its effect in younger people is unknown. High alcohol consumption is associated with increased total mortality, but the findings for stroke have varied.1 2 We analysed the association between alcohol consumption and the incidence of myocardial infarction, stroke, and mortality in a 25 year follow up of military conscripts.
Subjects, methods, and results
This study is based on 49 618 Swedish men conscripted between 1 July 1969 and 30 June 1970 and born between 1949 and 1952.2 At conscription, all men were given two questionnaires with questions covering social background, behaviour, and use of alcohol and tobacco, and all met with a psychologist for assessment. The percentage of non-responders was between 1% and 2%.
From questions about quantity and frequency of consumption of different alcoholic beverages, we calculated usual consumption in terms of grams of 100% ethanol a day and categorised subjects into different consumption groups.1 Social, psychosocial, and behavioural variables were included as confounders, based on earlier studies and on bivariate analyses of the data in this study (see table).2 Using the Swedish personal identification number, we linked the questionnaire data to the Swedish register of causes of deaths and to the new national inpatient care register for 1970-95. We estimated unadjusted and adjusted relative risks in logistic regression analyses and calculated attributable proportions.3 Outcomes were total mortality (n=1473) and incidences of myocardial infarction (n=279, 38 fatal) and stroke (n=233, 30 fatal) as underlying causes of death or as main diagnoses at hospitalisation.
Compared with abstainers, alcohol consumers had higher unadjusted relative risks for all three outcomes, and the risks increased with increasing alcohol consumption, being significantly higher for consumers of ≥15 g ethanol/day. Adjusted analyses showed an increasing risk of death (significant) and stroke (non-significant) with increasing alcohol consumption but a decreasing risk of myocardial infarction (non-significant) (table) To a considerable extent, the increased mortality with high alcohol consumption was due to the strong association between drinking and smoking and the high risk associated with smoking: compared with not smoking, the relative risk of death was 3.02 for smoking 1-10 cigarettes a day and 5.20 for smoking ≥11 cigarettes a day. Consuming ≥15 g ethanol/day was associated with a relative risk of death of 1.37 (95% confidence interval 1.01 to 1.85). The attributable proportion in multivariate analysis for alcohol consumption, relative to abstention, was 14% for mortality (alcohol use caused 205 deaths) and 37% for stroke (causing 86 cases), while alcohol prevented 44 (16%) myocardial infarctions.
The validity of self reported alcohol consumption at conscription and as a measure of consumption during the follow up period can be questioned.3 Both non-differential underreporting at conscription and change in alcohol use over time lead to underestimation of an elevated relative risk.4 However, we found a clear association between level of alcohol use at conscription and risk of subsequent hospitalisation or death with a diagnosis of alcoholism, alcohol psychosis, or alcohol intoxication, with a significant relative risk of 5.71 for consumers of ≥15 g ethanol/day, indicating a stability over time. The results also indicate a cardioprotective effect of alcohol use in relatively young men, in whom myocardial infarction is rare. Possible biological mechanisms include an increase in high density lipoprotein cholesterol, a decrease in platelet coagulability, and a decrease in plasma fibrinogen associated with alcohol intake.1
Calculation of the attributable proportions clearly indicated that alcohol consumption had a negative net effect on the subjects' health up to the age of 45. The results support a restrictive alcohol policy and recommendations for little or no alcohol consumption by young men.5
Contributors: AR formulated the study hypotheses, was responsible for all phases of the study except the statistical analyses, and wrote the paper. AL was responsible for the statistical analyses and participated in planning and designing the study, interpreting the results, and editing the paper. AR is guarantor for the paper.
Funding This study was supported by grant 92-0300:3C from the Swedish Council for Social Research.
Competing interests None declared.