Ferrari P, Licaj I, Muller DC, et al (36 other authors). Lifetime alcohol use and overall and cause-specific mortality in the European Prospective Investigation into Cancer and nutrition (EPIC) study. Pre-publication; BMJ Open 2014;4:e005245. doi:10.1136/bmjopen-2014-005245.
Objectives: To investigate the role of factors that modulate the association between alcohol and mortality, and to provide estimates of absolute risk of death.
Design: The European Prospective Investigation into Cancer and nutrition (EPIC).
Setting: 23 centres in 10 countries.
Participants: 380 395 men and women, free of cancer, diabetes, heart attack or stroke at enrolment, followed up for 12.6 years on average.
Main outcome measures: 20 453 fatal events, of which 2053 alcohol-related cancers (ARC, including cancers of upper aerodigestive tract, liver, colorectal and female breast), 4187 cardiovascular diseases/coronary heart disease (CVD/CHD), 856 violent deaths and injuries. Lifetime alcohol use was assessed at recruitment.
Results: HRs comparing extreme drinkers (≥30 g/day in women and ≥60 g/day in men) to moderate drinkers (0.1–4.9 g/day) were 1.27 (95% CI 1.13 to 1.43) in women and 1.53 (1.39 to 1.68) in men. Strong associations were observed for ARC mortality, in men particularly, and for violent deaths and injuries, in men only. No associations were observed for CVD/CHD mortality among drinkers, whereby HRs were higher in never compared to moderate drinkers. Overall mortality seemed to be more strongly related to beer than wine use, particularly in men. The 10-year risks of overall death for women aged 60 years, drinking more than 30 g/day was 5% and 7%, for never and current smokers, respectively. Corresponding figures in men consuming more than 60 g/day were 11% and 18%, in never and current smokers, respectively. In competing risks analyses, mortality due to CVD/CHD was more pronounced than ARC in men, while CVD/CHD and ARC mortality were of similar magnitude in women.
Conclusions: In this large European cohort, alcohol use was positively associated with overall mortality, ARC and violent death and injuries, but marginally to CVD/CHD. Absolute risks of death observed in EPIC suggest that alcohol is an important determinant of total mortality.
There is no question that heavy alcohol consumption, especially when associated with smoking, increases the risk of a number of upper aero-digestive cancers that are commonly referred to as “alcohol-related cancers.” In the present study, the authors have also included in this group a number of other cancers that may be related less directly to the effects of alcohol; these include colorectal cancer and female breast cancer which, because they are so much more common, make up the large majority of cancers “related to alcohol.” All Forum members consider the EPIC study to be an important source of data on cancer. However, there were some concerns about certain aspects of the analysis, as described below.
Specific comments on the study: It was noted by Forum members that all estimates of alcohol were only from reported alcohol consumption at baseline, and no alcohol data were collected during a follow-up period that averaged 12.6 years. There were no data presented on the pattern of drinking (regular moderate versus binge drinking). Further, Forum members noted that there was no discussion of the effects of under-reporting of alcohol, which has been shown to markedly affect health effects of alcohol intake in epidemiologic studies.
In their Abstract, the authors focus on the “extreme” drinkers, those women who consume (≥30 g/day) or men who consume ≥60 g/day. They do not point out that only 2.4% of the women in this study consumed at this level, and little attention is given to the fact that almost all of the women were non-drinkers or light-moderate drinkers. Specifically, 10.1% of the women were lifetime never drinkers, 45.3% consumed 0.1-4.9 g/d, and 31.1% reported 5-14.9 g/d. A total of 11.0% reported 15-29.9 g/d and 2.4% reported ≥30 g/d. From the data presented, there is a clear U-shaped curve: among women; the highest risks for total mortality were in the abstainers (a 26% increase over the referent group of light drinkers) and the very small number of women in the highest drinking category (a 27% increase). Similar patterns were seen for all cancers (even for alcohol-related cancers) and for violent deaths; for CVD deaths, in comparison with non-drinkers, the estimated risks were lower for all alcohol categories.
Men consumed more alcohol, as only 1.6% were never drinkers, 14.0% reported 0.1-4.9 g/d, 28.3% reported 5-14.9 g/d, 27.5% reported 15-29.9 g/d, 20.4% reported 30-59.9 g/d, and 8.2% reported ≥ 60 g/d. Hence, more than one-quarter of the men reported that they consumed 30 or more grams/d of alcohol. For men, there was generally a U-shaped curve, with lower death rates for light to moderate drinkers. However, the heaviest drinkers (≥ 60 g/d) had the highest risk of death for overall mortality and for deaths from cancers and other causes of death.
For analyses separating smokers and non-smokers, among never-smokers, essentially the highest risk was among abstainers, and consumption up to 30 g/d had no adverse effects for never-smokers. At every level of drinking, smokers had increased risk of death in comparison with non-smokers.
Reviewer Lanzmann-Petithory was also concerned that “The focus of this paper is on the extremes of alcohol consumption and there are incomplete comparisons between wine drinkers and beer drinkers. While the authors admit in their discussion that there are nutritional and cultural confounding factors, saying ‘Although we believe that this finding is relevant, we call for cautious interpretations of these results, as the lifestyle profile of wine and beer drinkers is profoundly different,’ they translate that in the results of the abstract to ‘Overall mortality seemed to be more strongly related to beer than wine use, particularly in men.’”
Reviewer Thelle noted: “It is peculiar that never-drinking women (and men – but there are few events) have increased risk of violent deaths, but that only underlines the need of using the low consumers as a reference group. The heterogeneity across geographical areas with Northern Europe showing stronger associations with total and overall mortality suggests drinking pattern more than average amount as a driving force. It is a pity that large cohorts from the north were excluded due to lack of information in some comparisons. Still, it is my view that this report is as good as observational studies can be regarding alcohol and mortality.”
Forum member Skovenborg praised the work of the EPIC study, but thought that the analytical strategy for the present study raise some questions: “Categories of alcohol use: Never drinkers are discarded as the reference group by the authors, which might be reasonable concerning men (only 1.5% of men reported having never consumed alcohol) but hardly in women, where never drinkers made up 10% of the total cohort. Information on lifetime alcohol use was available on 76% of the cohort, allowing separate consideration of former drinkers and lifetime abstainers, but this is not discussed.
“The authors quote risk of misclassification of alcohol quantity and lack of accuracy in reporting prevalent morbid conditions at baseline in the group of never drinkers as argument for their decision to discard never drinkers as the reference category. The authors do not present evidence to support their assumptions about EPIC never drinkers, neither do they look for possible explanations for the association of never drinking and increased risk of death due to violence and injury – a finding that might also be purely coincidental.
“Moderate drinkers (0.1 – 4.9 g/day): this is a strange designation for an alcohol intake that in most other studies would be named as light or very modest. Categorizing an alcohol intake of 0.1 – 4.9 g/day as moderate is a suitable means to create confusion. Apart from that a steady, light intake of 4.9 g alcohol per day might well have a biological effect obscuring the difference between moderate vs no consumption of alcohol and risk of disease and death. Most studies would define 5 – 14.9 g/day as light to moderate (this group is not given a name in this study). Further, most studies would consider the consumption of 15 – 29.9 g/day as moderate for men and perhaps heavy for women; the authors give no name for this category.”
Reviewer Skovenborg was also concerned about misclassification due to underreporting: “The authors do not discuss the risk of misclassification due to underreporting. Underreporting of alcohol intake is a source of bias with respect to adverse effects of light–moderate drinking. In a recent study by Klatsky et al (Cancer Causes Control 2014;25:693-699) increased risk of cancer in association with alcohol consumption was concentrated in the stratum of study participants who had other recorded data indicating alcohol misuse, and who were thus suspected of underreporting their intake. Klatsky et al concluded that the apparent increased risk of cancer among light-moderate drinkers might be substantially due to underreporting of intake. Underreporting would also explain the finding of this paper that the rate advancement period on the risk of death was larger when the reference category was set to 0.1-15 g/day than when using a threshold of 5 g/day.” Reviewer De Gaetano agreed that “The underreporting issue is of utmost importance.”
Why was there not a greater effect of light-to-moderate drinking on cardiovascular diseases in this study: The effects of alcohol on cardiovascular disease in these analyses also raised questions. Several reviewers thought that this may relate to the lack of data on the pattern of drinking. Further, Dr. Roger Corder, Professor of Experimental Therapeutics at the William Harvey Research Institute in London, was asked to comment on the rather modest reduction in risk of coronary heart disease reported in this report. “I’m not really surprised that alcohol, and wine in particular, appears in this study to confer less protection against cardiovascular disease than earlier studies in France from the 1970s and 1980s. Many ‘modern’ wines are higher in alcohol than in the past, when it was unusual to see a wine with more than 12.5% alcohol by volume. With a wine 11% alcohol, one-half of a bottle would contain about 30 g alcohol; for modern wines it is frequently over 40 g alcohol in the same volume. Our studies of polyphenols in wine show lower levels than seen in the past. The present study doesn’t differentiate red or white drinkers, and the percent of white wine drinkers in Germany and Denmark is likely high. So when you take into account that modern red wines may be low in polyphenols, this study shouldn’t be expected to show the benefit from wine drinking that earlier French studies did. Ultimately it is just a study of the risks associated with alcohol consumption.”
Forum member Finkel suggested that “The higher alcohol level in many modern wines may relate partly to global warming, partly to viticultural practices enhancing ripening, perhaps partly to a misguided impression of vintners that higher alcohols lead to higher scores and more sales. We really need detailed systematic data, which most published studies lack.”
Many Forum members noted that the emphasis of this paper was almost exclusively on the dangers of any alcohol, without indicating that the data indicate that the dangers were primarily with “heavy drinking.” Lanzmann-Petithory considered the last sentence of the authors’ abstract to be misleading. “With their data, they could as well reasonably have concluded that ‘Heavy drinking is an important determinant of total mortality,’ and even more accurately, that ‘Never drinking is an important determinant of total mortality.’ The emphasis of the authors appears to be more political than scientific.”
Reviewer Waterhouse also had some major concerns about the overall message of the paper. “This paper goes to great lengths to raise concern about dangers from alcohol consumption. One of their conclusions is that alcohol consumption is “positively associated with overall mortality.” Their discussion focuses on the increased mortality risk of the < 3% of women who consume large amounts of alcohol. From their data presented, it appears that abstinence is more dangerous than heavy drinking for women, with an increased mortality risk of 34% compared to 27% for heavy alcohol consumption among non-smokers. But, since there are four times as many teetotalers as heavy drinkers, the cost to public health from abstinence is 4 times as great as heavy drinking. Also, since the lowest risk of mortality from any perspective is among light to moderate alcohol consumers, their overall conclusion that alcohol consumption increases mortality is incorrect. It is also clear that their classification of alcohol-related cancers is highly questionable because in women never-drinkers have a greatly increased risk! It appears to me that the authors have deliberately overlooked the hazards of abstinence in order to advance an anti-alcohol agenda.”
The large European Prospective Investigation into Cancer and Nutrition Study (EPIC) has released a new analysis of the relation of alcohol consumption to mortality. The study concluded that alcohol use was positively associated with overall mortality, alcohol-related cancers, and violent death and injuries, but marginally to cardiovascular disease, and that absolute risks of death observed in EPIC suggest that alcohol is an important determinant of total mortality.
There is no question that heavy alcohol consumption, especially when associated with smoking, increases the risk of a number of upper aero-digestive cancers that are commonly referred to as “alcohol-related cancers.” In the present study, the authors have also included in this group a number of other cancers that may be related less directly to the effects of alcohol; these include colorectal cancer and female breast cancer which, because they are so much more common, make up the large majority of cancers “related to alcohol.” Forum members consider the EPIC study to be an important source of data on cancer, but had a number of questions about the analysis and especially about the conclusions of the authors.
Major weaknesses of the study are that an assessment of alcohol intake was obtained only at a baseline visit, with no further assessments during a follow-up period averaging 12 years, and especially, no information was available on the pattern of drinking of subjects (e.g., regular moderate versus binge drinking). Further, Forum members noted that there was no discussion of the effects of under-reporting of alcohol, which has been shown to markedly affect health effects of alcohol intake in epidemiologic studies. In fact, recent large studies show that most of the cases of cancer that appear to relate to “light-to-moderate drinking” actually relate to underreporting of consumption by subjects who are found, from other collected medical data, to be heavy users or abusers of alcohol.
The authors focus on “extreme drinkers,” which consist of women who consume (≥30 g/day) or men who consume ≥60 g/day. They do not point out that only 2.4% of the women in this study consumed at this level, and little attention is given to the fact that almost all of the women were non-drinkers or light-moderate drinkers. From the data presented in the paper, there is a clear U-shaped curve among women: the highest risks for total mortality were in the abstainers (a 26% increase over the referent group of light drinkers) and the very small number of women in the highest drinking category (a 27% increase).
For men, 8.2% were in the highest drinking category, reporting an average consumption of ≥ 60 g/d (5 – 6 typical drinks). For men, there was generally a U-shaped curve, with lower death rates for light to moderate drinkers. However, the heaviest drinkers had the highest risk of death for overall mortality and for deaths from cancers and other causes of death. For both men and women, at every level of drinking, smokers had an increased risk of death in comparison with non-smokers.
Some reviewers were also concerned about the unusual categorization of alcohol intake used in the study. Referring to subjects reporting 0.1 – 4.9 g/day as “moderate drinkers” is a strange designation for an alcohol intake that in most other studies would be named as light or very modest, and may lead to confusion in interpreting results. (Most studies would define 5 – 14.9 g/day as light to moderate drinkers, and this group is not given a name in this study).
Overall, this study tends to show a U-shaped relation between alcohol consumption and mortality. The data presented focus primarily on the highest categories of drinking, levels that are well known to relate to many diseases and mortality. For truly light-to-moderate consumption, however, there is overwhelming epidemiologic data that such drinking relates to lower mortality risks, and the present study does not contradict such an association.
* * *
Comments in this critique by the International Scientific Forum on Alcohol Research were provided by the following members:
Erik Skovenborg, MD, Scandinavian Medical Alcohol Board, Practitioner, Aarhus, Denmark
Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA
Dag S. Thelle, MD, PhD, Senior Professor of Cardiovascular Epidemiology and Prevention, University of Gothenburg, Sweden; Senior Professor of Quantitative Medicine at the University of Oslo, Norway
Giovanni de Gaetano, MD, PhD, Department of Epidemiology and Prevention, IRCCS Istituto Neurologico Mediterraneo NEUROMED, Pozzilli, Italy
Dominique Lanzmann-Petithory,MD, PhD, Nutrition/Cardiology, Praticien Hospitalier Hôpital Emile Roux, Paris, France
David Van Velden, MD, Dept. of Pathology, Stellenbosch University, Stellenbosch, South Africa
Pierre-Louis Teissedre, PhD, Faculty of Oenology – ISVV, University Victor Segalen Bordeaux 2, Bordeaux, France
Fulvio Ursini, MD, Dept. of Biological Chemistry, University of Padova, Padova, Italy
Creina Stockley, PhD, MBA, Clinical Pharmacology, Health and Regulatory Information Manager, Australian Wine Research Institute, Glen Osmond, South Australia, Australia.
Fulvio Mattivi, PhD, Head of the Department Good Quality and Nutrition, Research and Innovation Centre, Foundazione Edmund Mach, in San Michele all’Adige, Italy
Andrew L. Waterhouse, PhD, Marvin Sands Professor, Department of Viticulture and Enology, University of California, Davis; Davis, CA, USA
Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway
Ulrich Keil, MD, PhD, Institute of Epidemiology and Social Medicine, University of Münster, Münster, Germany
R. Curtis Ellison, MD, Section of Preventive Medicine & Epidemiology, Boston University School of Medicine, Boston, MA, USA
In addition to Forum members, comments on this paper were sought from Dr. Roger Corder, Professor of Experimental Therapeutics at the William Harvey Research Institute in London.