Critique 214: Some methodologic problems in a recent paper urging changes in drinking guidelines – 23 April 2018

Wood AM, Kaptoge S, Butterworth AS . . . John Danesh, et al, for the Emerging Risk Factors Collaboration/EPIC-CVD/UK Biobank Alcohol Study Group (120 authors).   Risk thresholds for alcohol consumption: combined analysis of individual-participant data for 599 912 current drinkers in 83 prospective studies. Lancet 2018;391:1513-1523.

Authors’ Abstract

Background Low-risk limits recommended for alcohol consumption vary substantially across different national guidelines. To define thresholds associated with lowest risk for all-cause mortality and cardiovascular disease, we studied individual-participant data from 599 912 current drinkers without previous cardiovascular disease.  

Methods  We did a combined analysis of individual-participant data from three large-scale data sources in 19 high income countries (the Emerging Risk Factors Collaboration, EPIC-CVD, and the UK Biobank). We characterized dose–response associations and calculated hazard ratios (HRs) per 100 g per week of alcohol (12·5 units per week) across 83 prospective studies, adjusting at least for study or centre, age, sex, smoking, and diabetes. To be eligible for the analysis, participants had to have information recorded about their alcohol consumption amount and status (ie, non-drinker vs current drinker), plus age, sex, history of diabetes and smoking status, at least 1 year of follow-up after baseline, and no baseline history of cardiovascular disease. The main analyses focused on current drinkers, whose baseline alcohol consumption was categorised into eight predefined groups according to the amount in grams consumed per week. We assessed alcohol consumption in relation to all-cause mortality, total cardiovascular disease, and several cardiovascular disease subtypes. We corrected HRs for estimated long-term variability in alcohol consumption using 152 640 serial alcohol assessments obtained some years apart (median interval 5·6 years [5th–95th percentile 1·04–13·5]) from 71 011 participants from 37 studies.

Findings  In the 599 912 current drinkers included in the analysis, we recorded 40 310 deaths and 39 018 incident cardiovascular disease events during 5·4 million person-years of follow-up. For all-cause mortality, we recorded a positive and curvilinear association with the level of alcohol consumption, with the minimum mortality risk around or below 100 g per week. Alcohol consumption was roughly linearly associated with a higher risk of stroke (HR per 100 g per week higher consumption 1·14, 95% CI, 1·10–1·17), coronary disease excluding myocardial infarction (1·06, 1·00–1·11), heart failure (1·09, 1·03–1·15), fatal hypertensive disease (1·24, 1·15–1·33); and fatal aortic aneurysm (1·15, 1·03–1·28). By contrast, increased alcohol consumption was loglinearly associated with a lower risk of myocardial infarction (HR 0·94, 0·91–0·97). In comparison to those who reported drinking >0–≤100 g per week, those who reported drinking >100–≤200 g per week, >200–≤350 g per week, or >350 g per week had lower life expectancy at age 40 years of approximately 6 months, 1–2 years, or 4–5 years, respectively.

Interpretation  In current drinkers of alcohol in high-income countries, the threshold for lowest risk of all-cause mortality was about 100 g/week. For cardiovascular disease subtypes other than myocardial infarction, there were no clear risk thresholds below which lower alcohol consumption stopped being associated with lower disease risk. These data support limits for alcohol consumption that are lower than those recommended in most current guidelines.

Forum Comments

While this paper was based on a very large sample size, and leading epidemiologists are listed among the numerous authors, Forum members had some concerns about certain aspects of the study and the conclusions of the authors.

Exclusion of non-drinkers from the main analyses: All Forum reviewers commented on the fact that the authors had data on ex-drinkers and never-drinkers, yet chose to exclude all current non-drinkers in the main analyses in this paper:  The authors state “. . . never-drinkers might differ systematically from drinkers in ways that are difficult to measure, but which might be relevant to disease causation.  When including never-drinkers and ex-drinkers, we reproduced previously reported U-shaped associations of alcohol consumption with total cardiovascular disease and all-cause mortality {comment from Forum: shown only in the Appendix}.  However, we observed notable differences in baseline characteristics between never drinkers and current drinkers (e.g., in relation to sex, ethnicity, smoking, and diabetes status), supporting the validity of focusing on current drinkers in our main analysis.”

Forum member Skovenborg commented: “The differences between never drinkers and current drinkers were that the group of never drinkers (n = 53,851) included more females (70.4% vs 44.3%), more non-whites (47.8%. vs 7.2%), less smokers (14.6% vs 21.3%), less heavy smokers (6.74 vs 17.0 pack years), less academic education (34.9% vs 51.4%), less physical inactive persons (5.2% vs 18.6%), higher diabetes prevalence (7.1% vs 3.7%). According to The EPIC-InterAct Study, the higher diabetes prevalence among never drinkers might be explained by the lower risk of type 2 diabetes in current drinkers (Beulens et al).

“It is hard to understand that a group of participants with a favorable lifestyle regarding non-smoking and physical activity should be excluded as a reference group. And without a substantial protective effect of alcohol consumption on important diseases like myocardial infarction it is difficult to explain that the all-cause mortality risk level is the same in never drinkers as in participants with the highest drinking level >300 grams/week. Further, in the Nurses’ Health Study (participants with same sex and same education level and same socio-economic status) 18,967 of 84,630 participants were never drinkers. The only difference between never drinkers and moderate drinkers were a higher level of non-smoking among never drinking nurses: 62% vs. c. 30% (Chen et al).  Also, the Emerging Risk Factors Collaboration excluded 109,315 of 356,819 eligible participants (30.6%) as ex-drinkers and never drinker. The same figures in EPIC-CVD were 5,099 of 31,135 (16.3%) and in UK Biobank 32,416 of 358,833 (9.0%) (Høyer et al).  The large differences in exclusion rates has not been explained by the authors.”

The authors of this paper state: “For all-cause mortality, there was a positive and curvilinear association with alcohol consumption, with the lowest risk for those consuming below 100 g per week. Associations were similar for men and women.”  Skovenborg commented: “A meta-analysis of 34 prospective studies (including many of the studies used by Wood et al) found a J-shaped relationship between alcohol and total mortality: consumption of alcohol up to 4 drinks per day in men and 2 drinks per day in women was inversely associated with total mortality. (Di Castelnuovo et al). These results have been found in many other studies and they are expected considering the recognized gender differences in body composition and biology. The results found in the present paper are highly unusual – a fact, however, that is neither mentioned nor explained by the authors.”  Reviewer Svilaas agreed about the failure of the authors to discuss the net health effects of any alcohol in their text: “A strange study protocol, and hidden information within the article.”

Reviewer Stockley wrote: “I completely agree with the importance of showing the effects of moderate drinking versus never-drinking, which is brushed over in the paper’s discussion based on drinker population differences. From the figure shown only in the supplementary material, it is noted that when lifetime abstainers are compared to light/moderate consumers, the risk of all-cause mortality is lower for light/moderate consumers.  Specifically, for all-cause mortality, never-drinkers have a greater risk of all-cause mortality than consumers who drink up to 250 grams per week of alcohol, i.e. up to 25 standard drinks/week, at 10 grams of alcohol per standard drink.  Further, the risk of all cause-mortality for abstainers is approximately the same as those consuming >250-≤350 grams per week of alcohol, i.e. 25-35 standard drinks/week.

“Concerning cardiovascular disease, the risk of all cardiovascular events is lower for light/moderate consumers, and in particular, there is a j-shaped relationship between the risk of alcohol intake and all cardiovascular events; the greatest reduction in the risk is at 100-200 grams per week of alcohol, i.e. 10-20 standard drinks/week. Also, never-drinkers even have a greater risk of all cardiovascular events than consumers who drink >600 grams per week of alcohol, i.e. more than 60 standard drinks/week.”

Forum member Teissedre noted: “A problem with the analyses in this paper is that their group of ‘never-drinkers’ is not very good and does not ‘match’ with their ‘drinkers’: imbalance between men and women, not the same ethnicity, not the same level of education, etc. Like good epidemiologists, the authors tell us that comparisons with this group are not possible because there are known confounding factors . . . So for them, the discussion related to moderate versus no alcohol consumption is not scientifically sound.  It’s a shame that they did not extract a subgroup of never-drinkers paired with similar drinkers and describe the associations.”

Forum member Keil wrote: “By eliminating the non-drinkers, this approach does not allow to make comparisons between non-drinkers and light to moderate to heavy drinkers. But this is what the scientific and the public health community would like to know and a lot of good data to this point have been published over the last thirty years.  When reading the actual publication I do not understand why important information in the appendix does not appear in the publication and is not discussed and interpreted there.  I know of Lancet publications dealing with complex data sets which are three times the length of this paper.  Space should not have been a problem for the Lancet.

“Further, I have my problems understanding the assessment of alcohol intake in quite a number of studies. For example, in about 5 of the 83 included studies (including PROCAM, PROSPER and SHIP), the ascertainment method of alcohol consumption was unknown and there are more studies whose ascertainment methods are unclear.  There are a number of occupational cohort studies included, for which it is known that they are prone to bias with regard to the assessment of alcohol consumption.”

Reviewers Keil and Ellison also noted that in the data presented in the appendices, the association between baseline alcohol consumption and HDL-cholesterol showed no linear trend. Ellison stated: “HDL levels for ‘Never-drinkers’ is given as 1.38, which is identical to that of ‘All-current drinkers.’  One of the most consistent relations shown in essentially all epidemiologic studies is a strong positive relation between alcohol intake and HDL.  Also, in this study fibrinogen levels are higher for never-drinkers than for light-drinkers (0-50 and 50-100 g/week), whereas the opposite is usually found.  As increased HDL and lower fibrinogen are two important mechanisms of alcohol’s protective effect against cardiovascular disease, the findings in this paper make suspect the underlying data for these analyses.”

Richard Harding of the UK, a co-author of previous British guidelines for alcohol consumption and an invited participant on a BBC2 programme discussing this paper last week, was asked to comment on this paper. He stated: “‘The absence of an abstainer category in the presentation of the results gives the grossly misleading impression that 100g of alcohol/week is an acceptable threshold of risk, from which mortality risk increases with rising intake.  But in fact you have to drink this much to get the benefit, which is then eroded with higher levels of intake until the mortality rate for abstainers is reached.  The overall finding is actually little different from many other studies of this type.”  As for excluding non-drinkers from their main analyses, he added: “This seems ridiculous, like excluding non-smokers from a smoking study.  But the risk for ex-drinkers and non-drinkers is tucked away on page 31 of the Supplementary Appendix, and there is the real reason for their exclusion: much higher risk for non-drinkers than for subjects in all but the highest categories of alcohol consumption.”

Reviewer Waterhouse summarized the views of many Forum members: “A growing number of studies seem to be doing their level best to obscure their own data showing that drinking alcohol might have health benefits.  This new paper largely omits any comparison between drinkers and non-drinkers in the abstract or even the text, hiding these results in supplementary tables not normally reviewed by the press or anyone other than the most diligent of readers.  The surprising (or not so surprising) result from their data is that the moderate consumption of alcohol (200 grams/week or less) appears to protect drinkers (compared to never-drinkers) by reducing death from cardiovascular disease events by more than 30% and total mortality by more than 15%.  The authors claim that this result is compromised by differences between drinkers and non-drinkers in their primary data sources, but the job of epidemiologists is to attempt to resolve those differences in the course of their data analysis. These author abdicated this responsibility.”

Differences in effect according to the type of alcoholic beverage: Wood et al also state in this paper: “Associations of baseline alcohol consumption with all-cause mortality were stronger in drinkers of beer or spirits than of wine, and in those drinking less frequently (when consuming the same weekly amount), including binge drinkers.” Skovenborgs’s comment on this: “The well-known effects of drinking pattern is confirmed in the study results, however, these results are not mentioned in the abstract and not taken into consideration in the conclusion regarding drinking guidelines.”

Forum member Teissedre wrote: “You will have noticed that wine drinkers are less likely than drinkers of beer or hard liquor to have adverse outcome events. The authors reported the differences according to type of beverage, but then added: ‘However, people showing these behaviours had higher baseline levels of smoking and other indicators of lower socioeconomic status, suggesting the potential for confounding effects.’  It is shown in the Supplementary Appendix that, for cardiovascular disease subtypes, hazard ratios tended to be higher in beer and spirit drinkers than in wine drinkers, but not significantly so in direct comparisons involving a common set of participants.”  Ellison thought that an attempt might have been made to estimate effects of any alcohol intake (versus none) when these known confounders were adjusted for in the analysis, perhaps by stratification.  Others pointed out that in previous studies, wine consumers generally show greater beneficial effects on health than those of drinkers of other beverages, even when all known confounders are included in the analysis.

Effects of under-reporting of alcohol intake: The authors did not discuss the estimated effects of under-reporting of alcohol intake on their results.  Skovenborg noted: “Self-reported alcohol consumption data are prone to bias and are challenging to harmonise across studies conducted over different time periods that used varying instruments and methods to record such data.  The most important bias is underreporting.  In The Svalbard Study, a unique setting for validation of self-reported alcohol consumption, the self-reported volume accounted for approximately 40% of the sales volume (Høyer et al).  The coverage of sales estimates by surveys varies between 39% (Germany) and 56% (France).

“The Kaiser Permanente Study found underreporting of alcohol intake partially explained (1) the increased prevalence of hypertension among persons reporting one to two drinks per day (Klatsky et al, 2006); (2) the apparent increased risk of cancer among light-moderate drinkers (Klatsky et al, 2014); and (3) the apparent magnitude of benefit of lighter drinking (Klatsky & Udaltsova). Further, in an observational study from Italy of men aged 45-64 who were followed for total mortality from 1965 to 1995, men reporting drinking about 5 drinks per days had a longer life expectancy than occasional drinkers and heavy drinkers.  Underreporting is not an issue in this population with almost no non-drinkers and total acceptance of drinking wine with your food (Farchi et al).”

Need for different advice according to age and sex: The primary beneficial effects that have been demonstrated from moderate alcohol intake are primarily for the “diseases of ageing.”  These include cardiovascular disease, in particular, but also there is increasing evidence that moderate consumption may reduce the risk of developing type-2 diabetes, dementia, osteoporosis, and several diseases related to inflammation.  While heavy consumption also relates to upper digestive and certain other cancers, light-to moderate intake has little effect on most such conditions; the primary exception is that many studies show increased risk of incident breast cancer (but not death from such cancer) among women reporting as low as one drink/day.

These diseases are conditions of middle-aged and older adults, and are exceedingly rare in young people. Thus, it is important that when giving advice to the public regarding alcohol, it should point out the age groups for which each set of recommendations apply.  In general, there are few detectable beneficial health effects among the young, where excessive and/or binge drinking is a common problem that is associated with many adverse effects on personal and public health.  In terms of health effects, advice to the young should focus on avoiding alcohol or markedly limiting intake; for older adults, advice should take into consideration the beneficial health effects when moderate drinking is incorporated into a “healthy lifestyle”.

Further, guidelines regarding alcohol intake should be different for men and women. Physiologically, women process alcohol differently from men, and numerous studies have shown that similar amounts of alcohol may have greater biological effects among women.  The present study advises similar recommendations for both men and women.

Inadequate discussion of the importance of the pattern of alcohol consumption: Many Forum members commented on the inadequacies of the authors in discussing the pattern of drinking. Stated reviewer Stockley: “What is really missing for me is any discussion on the importance of pattern of consumption, as the data is expressed as a weekly rather than daily amount, although there is a brief acknowledgement that binge drinking is associated with greater risks of all-cause mortality.”  It has been shown consistently that the lower risk of developing coronary disease for moderate drinkers results from the regular use of alcohol, and this protection is diminished or lost when consumption occurs only on week-ends, and especially with binge drinking.  Although differences between consumption on 2 or fewer versus 2 or more days per week are mentioned in the text, this should have been discussed more broadly and included in any suggestions for changing guidelines.  Further, it would have been useful to see the effects of different cut-points for frequency of drinking (e.g., 3-4 days/week, 5-7 days/week).

Reviewer Thelle noted the estimated decrease in life expectancy at age 40 for consumers of 200 to 350 g/week of alcohol, but stated: “. . . again, this calculation is only showing what is going on among the drinkers and not broken down by different beverages or drinking pattern.” Forum member Goldfinger wrote: “An extrapolated j-shaped curve published in a 2002 meta-analysis (DiCasterlnuovo et al) showed the lowest cardiovascular risk at 750 ml wine daily!”  [However, Forum member de Gaetano (the senior author of the paper) responded: “A maximum reduction was predicted at 750 mL/day, but statistical significance was only reached up to the amount of 150 mL/day.”]

Forum member Lanzmann-Petithory stated: “The main results of this study are confounded, not taking into account the drinking pattern and the alcoholic beverage (as described by Ruidavets et al), without speaking of the problem of excluding abstainers, covered by other Forum members. It has no sense to make weekly recommendations because to drink 200 g of alcohol/week in the form of 20 drinks of spirits during the weekend has nothing to do with 2-3 glasses of wine every day during meals.”  Reviewer Van Velden added: “I also feel that the pattern of drinking is important, as well as underreporting of consumption.  We know that wine, in particular, when consumed with meals regularly and moderately, can be an important part of a healthy lifestyle, and combats inflammation.  We are moving away from cholesterol as the main cause of CVD towards widespread inflammation.  Further, genetic differences in population groups will also have an influence on the results.”

Effects of alcohol on other health conditions: Reviewer Keil commented: “In the present paper, there is no mention of cognitive functioning, of osteoporosis, of diabetes risk, or other conditions favorably related to moderate alcohol consumption.  Instead more and more voices are to be heard that light to moderate alcohol consumption produces dementia.  With regard to big numbers, I think that an early paper in 1997 from the NEJM (Thun et al) makes a lot more sense to me; that paper showed that people who are drinkers and non-smokers have the lowest and those who are non-drinkers but smokers have the highest mortality. The other categories are in-between.”  Forum member Estruch noted: “The dietary habits were not adequately evaluated in the analysis.  In our experience the effects of alcoholic beverages differ according to the dietary pattern followed. Thus, the effects of moderate drinking on cancer incidence may vary from protection when the participants followed a Mediterranean diet (Schwingshackl et al) to increase the risk with other less healthy dietary patterns.”

What should be the basis of population guidelines?  Finally, after a discussion of some apparent weaknesses of the analyses and interpretation by the authors of the results in the present paper, it might be useful to discuss what should be the factors considered when setting drinking guidelines for the public.  As described by Harding and Stockley, “A comparison of worldwide recommendations on alcohol consumption reveals wide disparity among countries.  This could imply that many of the recommendations do not adequately accommodate the science, given that the science is equally valid worldwide.  Such a view, however, would be an oversimplification of the problem that those who formulate such guidelines face.

“The objective of guidelines is to influence and change behavior among target populations. It follows, therefore, that several factors then become relevant: behavior that is thought to be in need of change, the culture and mindset of the target populations, and the kind of message that is likely to be effective. There are some tensions between advice intended only to reduce the prevalence of misuse and that which also seeks to reflect the evidence on the beneficial health effects of moderate consumption.”  Harding and Stockley conclude: “It is important, therefore, when formulating recommendations on maximum levels of alcohol consumption to recognize these potential problems and to seek ways of resolving them.”

Forum member Ellison added: “In terms of providing data that can be used when setting guidelines, we note that this paper is based on prospective studies from high-income societies covering a number of countries, with a very large numbers of subjects. While a variety of appropriate analyses were done, results from many are shown only in the Supplement, and many important results are not included in the abstract and little discussed in the text.  Thus, the results described by the authors in their conclusions are insufficient, when considered alone, for setting drinking guidelines.”

Factors affecting the authors’ conclusions and implications of this paper: Professor Keil stated: “In the beginning of 2014 the Lancet published a series of 5 papers criticising the quality of medical research very impressively; the title of the Lancet series was: ‘Increasing value and reducing waste in biomedical research’.  The tenor of this important series was ‘We need less research, we need better research and we need research for the right reasons’.  I ask the Lancet to live up to its pledges, demands, and expectations for better research.  The present study does not contribute to our knowledge and falls short of many excellent previous studies.”

Overall, many Forum members were concerned that some key results of these analyses, as shown in the supplementary material, are not those discussed in the text and presented in the abstract. Some of the overall results appear to have been glossed over; yet, the media has accepted the authors’ conclusions and widely announced them to the public without question.  Stated Forum member Finkel: “The Lancet continues to add to its legend of editorial promiscuity.  Strictly speaking, it seems to me that, considering the disconnects between the data and the conclusions, this paper should not have been published, about which the media are neither equipped to understand, nor do they care or want to care.”

References from Forum review

Beulens JW, van der Schouw YT, Bergmann MM, et al. Alcohol consumption and risk of type 2 diabetes in European men and women: influence of beverage type and body size The EPIC-InterAct study.  J Intern Med. 2012;272:358-370. doi: 10.1111/j.1365-2796.2012.02532.x.

Chen WY, Rosner B, Hankinson SE, et al. Moderate Alcohol Consumption During Adult Life, Drinking Patterns, and Breast Cancer Risk. JAMA 2011;306:1884-1890. doi:10.1001/jama.2011.1590.

Di Castelnuovo A, Costanzo S, Bagnardi V, Donati MB, Iacoviello L, de Gaetano G. Alcohol dosing and total mortality in men and women: an updated meta-analysis of 34 prospective studies.  Arch Intern Med 2006;166:2437-2445.

Farchi G, Fidanza F, Giampaoli S, Mariotti S, Menotti A. Alcohol and survival in the Italian rural cohorts of the Seven Countries Study.  Int J Epidemiol 2000;29:667-671.

Harding R, Stockley CS. Communicating Through Government Agencies.  Ann Epidemiol 2007;17:S98–S102.

Høyer G, Nilssen O, Brenn T, Schirmer H. The Svalbard study 1988-89: a unique setting for validation of self-reported alcohol consumption.  Addiction 1995;90:539-544.

Klatsky AL, Gunderson EP, Kipp H, Udaltsova N, Friedman GD. Higher prevalence of systemic hypertension among moderate alcohol drinkers: an exploration of the role of underreporting.  J Stud Alcohol 2006;67:421-428.

Klatsky AL, Udaltsova N. Alcohol Drinking and Total Mortality Risk.  Ann Epidemiol 2007;17:S63–S67.

Klatsky AL, Udaltsova N, Li Y, Baer D, Nicole Tran H, Friedman GD. Moderate alcohol intake and cancer: the role of underreporting.  Cancer Causes Control 2014;25:693-699. doi: 10.1007/s10552-014-0372-8.

Ruidavets JB, Ducimetière P, Evans A, Montaye M, Haas B, Bingham A, Yarnell J, Amouyel P, Arveiler D, Kee F, Bongard V, Ferrières J. Patterns of alcohol consumption and ischaemic heart disease in culturally divergent countries: the Prospective Epidemiological Study of Myocardial Infarction (PRIME). BMJ 2010, 341:c6077.

Schwingshackl L, Schwedhelm C, Galbete C, Hoffmann G. Adherence to Mediterranean Diet and Risk of Cancer: An Updated Systematic Review and Meta-Analysis.  Nutrients 2017;9:1063; doi:10.3390/nu9101063

Thun MJ, Peto R, Lopez AD, Monaco JH, Henley J, Heath Jr. CW, Doll R. Alcohol consumption and mortality among middle-aged and elderly U.S. adults.  N Engl J Med 1997;337:1705-1714.

Forum Summary

Using individual-participant data from a number of large studies, predominantly from the UK and other European countries, the authors of the present paper have estimated the association between the reported level of alcohol intake among drinkers with cardiovascular outcomes and mortality. They conclude that the lowest risk of these outcomes is from the intake of less than 100 g of alcohol per week, and that guidelines for the public should reduce the upper limits for the amount of alcohol that could be safely consumed.

Forum members identified a number of flaws in these analyses, including failure to separate advice for men versus women or for older adults versus young people. Further, the authors do not point out the effects of under-reporting of intake, the marked differences in many health effects according to type of alcoholic beverage (such differences are shown only in their supplement but not mentioned in the abstract); also, there is little discussion of the importance of the pattern of drinking on health results.

Most important, by excluding never-drinkers as a reference group in their main analyses, the authors essentially eliminated the ability to evaluate for any potentially beneficial (or adverse) effects of light-to-moderate drinking as compared with non-drinking. This is problematic, as in essentially all previous large studies, moderate drinkers are those for whom significant and large benefits have been reported for risk of cardiovascular disease and total mortality.

Guidelines for the public regarding alcohol intake relate to many factors, including the net overall physiologic effects on health as well as the particular needs of different countries and cultures. And, as in this study, the inclusion of results from so many different cultures is a problem for determining appropriate guidelines for alcohol intake.  Given that culturally specific drinking patterns, type of beverage, and many other lifestyle factors modify the health effects of alcohol, mixing data from markedly different cultures may give results that may not be appropriate for any specific population.

*                *                      *                      *                      *                      *                   *

Comments on this critique by the International Scientific Forum on Alcohol Research were provided by the following members:

Erik Skovenborg, MD, specialized in family medicine, member of the Scandinavian Medical Alcohol Board, Aarhus, Denmark

R. Curtis Ellison, MD, Professor of Medicine, Section of Preventive Medicine & Epidemiology, Boston University School of Medicine, Boston, MA, USA

Giovanni de Gaetano, MD, PhD, Department of Epidemiology and Prevention, IRCCS Istituto Neurologico Mediterraneo NEUROMED, Pozzilli, Italy

Tedd Goldfinger, DO, FACC, Desert Cardiology of Tucson Heart Center, University of Arizona School of Medicine, Tucson, AZ, USA

Ulrich Keil, MD, PhD, Professor Emeritus, Institute of Epidemiology & Social Medicine, University of Muenster, Germany

Dominique Lanzmann-Petithory, MD, PhD, Nutrition Geriatrics, Hôpital Emile Roux, APHP Paris, Limeil-Brévannes, France

Creina Stockley, PhD, MSc Clinical Pharmacology, MBA; Health and Regulatory Information Manager, Australian Wine Research Institute, Glen Osmond, South Australia, Australia

Pierre-Louis Teissedre, PhD, Faculty of Oenology–ISVV, University Victor Segalen Bordeaux 2, Bordeaux, France

Dag S. Thelle, MD, PhD, Department of Biostatistics, Institute of Basic Medical Sciences, University of Oslo, Norway; Section for Epidemiology and Social Medicine, Sahlgrenska Academy, University of Gothenburg, Sweden

Andrew L. Waterhouse, PhD, Department of Viticulture and Enology, University of California, Davis, USA

Harvey Finkel, MD, Hematology/Oncology, Retired (Formerly, Clinical Professor of Medicine, Boston University Medical Center, Boston, MA, USA)

David Van Velden, MD, Dept. of Pathology, Stellenbosch University, Stellenbosch, South Africa

Ramon Estruch, MD, PhD, Hospital Clinic, IDIBAPS, Associate Professor of Medicine, University of Barcelona, Spain.

Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway

In addition, the Forum included an invited comment from Richard Harding, PhD, experienced in providing advice to the public regarding alcohol consumption.  He was a co-author of previous British guidelines for drinking, has provided expert testimony for the House of Commons discussion on the topic, and was an invited participant last week in a discussion on a BBC2 TV news programme about the current paper.

Note on potential conflict of interest: Professor Elizabeth Barrett-Connor, one of the 120 listed authors of this paper, is a member of this Forum.  She had no input into the discussions by forum members in reviewing this paper, or in the final published critique.

*                   *                      *                      *                      *                      *                      *

Additional Note: Comments on the internet regarding this paper by other scientists: In addition to this review of the paper by Forum members, there has been considerable recent discussion of this paper on the internet that reflects similar concerns as the Forum has noted.  These include a comment from Professor of Epidemiology Cecile Janssen of Emory University in Atlanta, GA: “Yes, the numbers went a little up with a drink a day, but read what the authors wrote.  The problem with multi-study analyses is residual bias, the lack of detailed data on other factors that might explain the ‘slightly’ higher risk.  A slight increase might be no increase.” (Accessed on 4/19/18 from Professor Kenneth Rothman of Boston University commented: “Her point, with which I agree, is that the rates for nondrinkers show that there is a lot of selection bias and possibly uncontrolled confounding, and yet the authors make much of the small differences among levels of alcohol consumption.  Too much is being read into those small differences.” (Personal communication.)

Professor John Duffy, a statistician formerly from the University of Edinburgh and head of Statistics in the Department of Primary Care at the University of Birmingham, currently directs John C Duffy Statistical Consultants.  He provided Dr. Richard Harding with the following prior to the BBC2 programme last week: “The authors of this paper have essentially constructed average relationships across a range of countries.  The weight for each country is effectively the sample sizes in the studies for that country. So what population would this average relate to?  (Actually, the answer is a non-existent hypothetical population).  Different countries have different risk relationships, so what is the relevance of this average to the UK?”  He added: “The 100gms is not a threshold of acceptable risk – it is the minimum risk point – in a sense the best amount to drink if you want to minimise your chances of mortality.  Non-drinkers were not included in the study – so the question of at how much alcohol the risk gets up to the level of non-drinkers can’t be answered from this research.”