Beer binging and mortality: results from the kuopio ischaemic heart disease risk factor study, a prospective population based studyBMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7112.846 (Published 04 October 1997) Cite this as: BMJ 1997;315:846
- Jussi Kauhanen (), junior research fellowa,
- George A Kaplan, professorb,
- Debbie E Goldberg, research associatec,
- Jukka T Salonen, professora
- a Research Institute of Public Health and Department of Public Health and General Practice, University of Kuopio, Kuopio, Finland
- b Department of Epidemiology, School of Public Health, University of Michigan, Ann Arbor, MI 48109-2029, USA
- c Human Population Laboratory, Berkeley, CA 94704, USA
- Correspondence to: Dr Jussi Kauhanen Department of Public Health and General Practice, University of Kuopio, Box 1627, FIN-70211 Kuopio, Finland,
- Accepted 3 June 1997
Objective: To examine the association between beer binging (regular sessions of heavy beer drinking) and mortality.
Design: Prospective population based study with the baseline assessment of level of alcohol intake (dose), by type of drink and drinking pattern, previous and existing diseases, socioeconomic background, occupational status, involvement in organisations during leisure time, physical activity in leisure time, body mass index, blood pressure, serum lipids and plasma fibrinogen concentration, during an average of 7.7 years' follow up of mortality.
Subjects: A population sample of 1641 men who consumed beer who were aged 42, 48, 54, or 60 years at baseline.
Main outcome measures: All cause mortality, cardiovascular mortality, death due to external causes, fatal myocardial infarctions.
Results: The risk of death was substantially increased in men whose usual dose of beer was 6 or more bottles per session compared with men who usually consumed less than 3 bottles, after adjustment for age and total alcohol consumption (relative risk 3.01 (95% confidence interval 1.54 to 5.90) for all deaths; 7.10 (2.01 to 25.12) for external deaths; and 6.50 (2.05 to 20.61) for fatal myocardial infarction). The association changed only slightly when smoking, occupational status, previous diseases, systolic blood pressure, low density lipoprotein and high density lipoprotein cholesterol concentration, plasma fibrinogen concentration, body mass index, marital status, leisure time physical activity, and involvement in organisations were controlled for.
Conclusion: The pattern of beer binging is associated with increased risk of death, independently of the total average consumption of alcoholic drinks. The relation is not explained by known behavioural, psychosocial, or biological risk factors. Death due to injuries and other external causes is overrepresented among beer bingers, but a strong association with fatal myocardial infarction suggests that the pathway may also involve other acute triggers of severe health events.
The effects of drinking pattern on mortality and morbidity are less well known than the effects of total alcohol consumption
The binging style of drinking beer was associated with the risk of death from any cause and from cardiovascular and external causes and with fatal myocardial infarctions in middle aged men in Finland
The association was not explained by the total amount of alcohol consumption, and it remained after adjustments for several potential confounders
Strong association of beer binging with deaths from external causes and fatal myocardial infarction suggests that this type of drinking pattern may involve triggers of severe acute events
Heavy consumption of alcohol is known to increase mortality from all causes and especially from cardiovascular causes.1 2 3 4 Little evidence exists, however, about the health effects of drinking pattern beyond the effects of overall consumption. We examined prospectively the relation between binge drinking (regular sessions and heavy beer drinking) of beer and all cause mortality, cardiovascular mortality, deaths from external causes (injuries, poisonings, suicides, violence), and incidence of fatal and non-fatal myocardial infarction, with adjustment for total consumption of beer and other alcoholic drinks. We compared middle aged Finnish men who usually consume six or more bottles of beer per session with men who usually have less than three bottles, and we made adjustments for several potentially confounding variables.
Subjects and methods
The study population comprised a random, age stratified sample of middle aged men living in the city of Kuopio, Finland, or surrounding rural communities. The men were aged 42, 48, 54, or 60 years at the baseline examination that was carried out between March 1984 and December 1989. Of 3235 eligible men, 2682 participated,5 6 of whom 1849 reported drinking beer. Information on one or more covariables was missing in 208 men, so the analyses were based on a sample of 1641 men who drank beer.
The extensive examination protocol and measurements have been described in detail.5 6 7 8 9 10 11 The subjects completed a self administered questionnaire and participated in a clinical examination. They were instructed that before the baseline examinations they should abstain from drinking alcohol for 3 days, smoking for 12 hours, and eating for 12 hours. The collection of blood specimens and the measurement of serum lipids and plasma fibrinogen concentrations were carried out as described previously.7 9 10 11
Alcohol consumption—We assessed alcohol consumption with a structured quantity and frequency method using the Nordic alcohol consumption inventory.12 13 Usual frequency of intake and usual dose (in glasses or bottles) were recorded for each type of drink (beer, wine, strong wine, spirits) with a structured response form. We calculated the measure for average weekly intake of various drinks on the basis of known alcohol content. A third of a litre bottle of ordinary beer (class III in Finland) contains 12 g of ethanol, and strong beer (class IV) 14 g. The usual dose of beer was determined according to the following scale: <1 bottle, 1 bottle, <2 bottles, 2 bottles, 3 bottles, 4 or 5 bottles, 6–9 bottles, >9 bottles. We used the total mean consumption all alcoholic drinks (grams of ethanol/week) as the covariate in adjusted models.
Health—We evaluated baseline health on the basis of the questionnaire, which was checked in an interview, and on the basis of an examination by a physician. Current coronary heart disease was defined as (a) a history of acute myocardial infarction, (b) angina pectoris, (c) positive angina pectoris on effort in the Rose interview14 (part of the questionnaire), or (d) the use of nitroglycerin tablets at least once a week. The diagnosis of other cardiovascular diseases (stroke, cardiac insufficiency, claudication) was based on the questionnaire and the physician's examination. The criteria for diabetes were diagnosed non-insulin dependent or insulin dependent diabetes, or fasting blood glucose concentration >11 mmol/l (WHO criteria), or both.
Psychosocial factors—We classified the occupational status of the participants as white collar, blue collar, or farmer on the basis of self reported main lifetime occupation.9 Information on marital status and unemployment was obtained from the questionnaire. Involvement in organisations during leisure time, which we have earlier reported to be associated with mortality,15 was used as the indicator of social contacts. For the assessment of depression, we included the Minnesota multiphasic personality inventory depression subscale with T score conversion.16
Behavioural and biological risk factors—We estimated the lifelong exposure to smoking (‘'cigarette years'’) as the product of years smoked and the number of cigarettes (or other tobacco equivalents) smoked daily at the time of examination. We assessed physical activity in leisure time from a 12 month history, modified from the Minnesota leisure time physical activity questionnaire; the intensity of physical activity was expressed in metabolic equivalents of oxygen consumption.8 10 We calculated body mass index as the ratio of weight (kg) to height (m)2. The measurement protocol of resting blood pressure has been described in detail.7 9
The follow up of all cause mortality and deaths due to cardiovascular and external causes was achieved through linkage to the national death registry, which is maintained for all Finnish citizens. We classified cardiovascular deaths and deaths due to external causes according to codes 390–459 and 800 and above, respectively, of the ICD-9 (international classification of diseases, 9th revision). We ascertained the incidence of acute myocardial infarctions, both fatal and non-fatal, by linkage to the registry of acute myocardial infarctions, established for the MONICA project (an international study conducted under the auspices of the World Health Organisation to monitor trends in and determinants of mortality from cardiovascular disease).17 The coverage of this registry was virtually 100% of the coverage of the national death registry.7 The mean follow up time for all cause mortality was 7.7 (maximum 10.8) years and for acute myocardial infarction 5.6 years (maximum 8.8 years). The end date for the whole cohort was set as the end of the last year from which registry information was available.
We analysed the associations between dose of beer and all cause mortality, cardiovascular mortality, deaths from external causes, and fatal and non-fatal acute myocardial infarction using the Cox proportional hazards model.18 To control for the possible non-linearity of the relation, we grouped the beer variable into <3 bottles (reference group), 3–5 bottles, 6 bottles or more. The first analyses included only age as a covariate. To check the changes in the risk estimates when total amount of alcohol was adjusted for, we added the weekly intake of beer, spirits, and wine (total alcohol intake) to the model. We then adjusted for a number of potential confounders. We controlled for smoking, leisure time physical activity, occupational status, marital status, social contacts, and depression in one model, and for previous diseases (diagnosed coronary heart disease, stroke, other cardiovascular disease, diabetes) in the second. We then added, in turn, systolic blood pressure, serum low density and high density lipoprotein cholesterol concentrations, plasma fibrinogen concentration, and body mass index in a series of separate models to show possible sources of confounding. Finally, we fitted a full model including all covariates.
Between March 1984 and December 1994, 136 deaths were registered in the study population. Of these, 59 deaths were due to cardiovascular disease and 25 to external causes. Of the 137 new acute myocardial infarctions, 28 were fatal. Table 1 shows the distribution of beer drinkers by the usual dose per session and the background characteristics of these groups. The number of outcome events and the corresponding mortality and incidence in each category are presented in table 2.
All cause mortality
The age adjusted Cox survival model showed a positive dose-response relation between the beer dose per session and all cause mortality (table 3). Even after adjustment for the total alcohol intake the relative risk of death was three times higher in the high dose group than in the low dose (reference) group and was also raised in the moderate dose group (table 3, 1). The difference in relative risk between the high and the low dose groups remained significant in the Cox models that further adjusted for behavioural, social, and psychosocial factors, disease history, and current diseases, depression, systolic blood pressure, serum low density and high density lipoprotein cholesterol concentrations, plasma fibrinogen concentration, and body mass index (table 3). After simultaneous adjustment for all covariates, the relation between six or more beers per session and all cause mortality was still significant (2.05 (1.01 to 4.14)).
The relative risk of cardiovascular death was higher in the high dose group than in the low dose group, but the difference was not significant when adjusted for age and total alcohol intake (2.42 (0.87 to 6.72)) (1). In a model with additional adjustment for previous diseases the estimated relative risk was significantly increased (3.64 (1.28 to 10.35)), but in the full model the relative risk decreased to 2.08 (0.69 to 6.27). The estimated relative risks in the moderate dose group varied between 1.60 (0.88 to 2.89) in the first model to 1.27 (0.70 to 2.39) in the full model.
In the age adjusted model the risk of death due to external causes (violence, suicides, injuries, poisonings) was seven times as high as that of the low dose group, and further adjustment for total alcohol intake did not notably change the relative risk estimate (table 4, 1). The association remained significant in every model, and, with all covariates included, the relative risk was still 5.78 (1.40 to 23.91) (table 4). The risk was not significantly increased in the moderate dose group.
We found no significant relation between the usual beer dose and myocardial infarctions (table 5). In fatal myocardial infarction, however, there was a clear gradient. In the high dose group the risk of fatal myocardial infarction was more than six times that of the low dose group when age and total alcohol consumption were controlled for (1). Adjustments for other covariates had no appreciable effect on the relative risk, except in the fibrinogen model. The raised risk of fatal myocardial infarction remained significant when all covariates were included in the same model (table 5). In the moderate dose group the raised risk relative to the low dose group was marginally significant.
Many population based studies have shown a J shaped or U shaped association between use of alcohol and all cause mortality.1 2 3 4 It is unclear, however, whether drinking pattern may have specific health consequences beyond total alcohol consumption. In this study the pattern of heavy acute intake of beer was associated with all cause mortality, deaths from external causes, and fatal myocardial infarctions in middle aged men, regardless of their total average consumption of beer, wine, and spirits. The risks were significantly higher among men who usually drank six or more beers at a time than among men who usually had less than three beers at a time. Our findings suggest that drinking pattern may have independent effects on health that are not explained by total consumption.
Pattern of alcohol intake is often considered to be more important than frequency of drinking as a determinant of actual alcohol misuse.19 Few studies, however, have tried to examine this in a general population, in which most individuals be classified not as misusers but as moderate drinkers. Even the most recent recommendations on alcohol use that are based on large population based studies20 21 emphasise the limits of average weekly consumption and not the way people reach, exceed, or keep themselves within the limits.
Congruence with other studies
Our finding that average consumption and drinking pattern may have separate effects is supported by a recent case-control study, in which drinking pattern seemed to modify the risk of stroke in light to moderate drinkers, and sporadic or occasional pattern was associated with higher risk than regular and evenly distributed consumption.22 Cross sectional analyses in the lung health study23 showed that alcohol consumption was itself inversely related to body mass index, whereas greater modal intake (number of alcoholic drinks on a day that alcohol is consumed) was positively associated with greater body mass index.
We were especially interested in beer drinking as over half of all alcohol in Finland is consumed as beer. Wine, for example, still holds a fairly narrow share of the market.24 Alcohol consumption per head of population in Finland is similar to many other Western countries and in 1993 was 6.8 litres pure ethanol, the same as in the United States and slightly less than in the United Kingdom.24 Population surveys in Finland indicate that the traditional way of drinking infrequently but heavily has remained quite common.25 Thus, beer binging was well represented in our study population.
Prospective findings from the Copenhagen city heart study showed that the frequency of drinking beer was not associated with all cause mortality, but three to five beers a day implied a reduction in cardiovascular mortality compared with those who never drank beer.26 Drinking three to five glasses of spirits a day was associated with an increased risk of both all cause and cardiovascular deaths, whereas daily intake of wine was related to significantly lower mortality. That study, however, focused on frequency of intake and type of drink, and not on drinking pattern. To enhance the validity of the measurement we used a detailed quantity and frequency scale that included separate questions on each type of drink. As in other population studies self reported use of alcohol is still likely to underestimate the real consumption level.3
The reason that the men who indulged in heavy drinking sessions had an increased risk of death in our follow up study might come from beer itself, from the pattern of heavy acute intake, or from other characteristics that are associated with men who prefer to drink six or more beers at a time. After adjustment for total beer consumption, the pattern of heavy acute intake remained a significant predictor of mortality. Thus, the total average amount of beer drinking, or intake of other types of alcohol, did not explain the association. Regarding potential confounding, the Kuopio ischaemic heart disease risk factor study15 allowed us to adjust for several known risk factors and background variables. In general the observed associations did not change after adjustment for smoking, physical activity, occupational status, marital status, social contacts, depression, or previous and existing diseases. Systolic blood pressure, obesity, and serum lipids and plasma fibrinogen concentrations did not, however, explain the effect. After simultaneous adjustment for all the covariates, the risk of any death, death from external causes, and fatal myocardial infarction remained significantly raised.
Association with sudden events
The magnitude of the relation was strongest with deaths from external causes and fatal myocardial infarctions. Risks of injuries, poisoning, violence, and suicide apparently increase with acute intoxication. Thus, six or more bottles of beer may well imply greater likelihood of these events. Our results on fatal myocardial events suggest that heavy acute intake of beer may involve acute triggers of severe pathophysiological events in the myocardium or the coronary arteries, or both. These events might include arrhythmia, ischaemia, and possibly thrombotic processes.
Although relative risks in the group of men drinking six or more beers seem to be high, the attributable risk in the population becomes important only if this drinking pattern becomes sufficiently common. In our study only 4% of men drank six or more beers.
The literature on possible differential effects of various types of alcoholic drink is growing.26 27 28 Our findings show that it is equally important to examine carefully the relation between drinking patterns—especially occasional heavy intake—and health outcomes.
We thank Kimmo Ronkainen, John Lynch, Sue Everson, and Richard Cohen for their help and comments.
Funding: This research was supported in part by the Alcoholic Beverage Medical Research Foundation, and by grants from the National Heart, Lung, and Blood Institute (HL44199), the Academy of Finland, and the Yrjö Jahnsson Foundation, Finland.
Conflict of interest: None.