Critique 183: An unusual analysis of the association of alcohol consumption with mortality — 24 March 2016

Stockwell T, Zhao J, Panwar S, Roemer A, Naimi T, Chikritzhs T.  Do “Moderate” Drinkers Have Reduced Mortality Risk?  A Systematic Review and Meta-Analysis of Alcohol Consumption and All-Cause Mortality.  J Stud Alcohol Drugs 2016;77:185–198.

Authors’ Abstract

Objective: Previous meta-analyses of cohort studies indicate a J-shaped relationship between alcohol consumption and all-cause mortality, with reduced risk for low-volume drinkers. However, low-volume drinkers may appear healthy only because the “abstainers” with whom they are compared are biased toward ill health. The purpose of this study was to determine whether misclassifying former and occasional drinkers as abstainers and other potentially confounding study characteristics underlie observed positive health outcomes for low volume drinkers in prospective studies of all-cause mortality.

Method: A systematic review and meta-regression analysis of studies investigating alcohol use and mortality risk after controlling for quality-related study characteristics was conducted in a population of 3,998,626 individuals, among whom 367,103 deaths were recorded.

Results: Without adjustment, meta-analysis of all 87 included studies replicated the classic J-shaped curve, with low-volume drinkers (1.3–24.9 g ethanol per day) having reduced mortality risk (RR = 0.86, 95% CI [0.83, 0.90]). Occasional drinkers (<1.3 g per day) had similar mortality risk (RR = 0.84, 95% CI [0.79, 0.89]), and former drinkers had elevated risk (RR = 1.22, 95% CI [1.14, 1.31]). After adjustment for abstainer biases and quality-related study characteristics, no significant reduction in mortality risk was observed for low-volume drinkers (RR = 0.97, 95% CI [0.88, 1.07]). Analyses of higher-quality bias-free studies also failed to find reduced mortality risk for low-volume alcohol drinkers. Risk estimates for occasional drinkers were similar to those for low- and medium-volume drinkers.

Conclusions: Estimates of mortality risk from alcohol are significantly altered by study design and characteristics. Meta-analyses adjusting for these factors find that low-volume alcohol consumption has no net mortality benefit compared with lifetime abstention or occasional drinking.  These findings have implications for public policy, the formulation of low-risk drinking guidelines, and future research on alcohol and health.

Forum Comments

Prospective cohort studies for decades have shown that light-to-moderate drinkers have a lower risk of developing cardiovascular diseases (CVD), and lower total mortality rates.  This has led to what has become known as the “J-shaped curve,” in that, in comparison with no alcohol consumption, light-to-moderate drinking is associated with a decrease in the risk of disease, while larger amounts tended to lead to an increase in risk over that of non-drinkers.

Through the years, epidemiologists have recognized that some of the favorable effects of alcohol intake relate to other lifestyle habits of moderate drinkers: in comparison with non-drinkers, they tend to be better educated, have higher incomes, smoke less, and have many other healthy lifestyle habits   The challenge has been to identify these potentially confounding factors and attempt to adjust for them so that the net effect of alcohol consumption can be better assessed.  Unfortunately, in the present paper, the authors have misrepresented the current scientific data in an attempt to prove their previously described points that alcohol does not relate to a reduced risk of CVD and mortality.

Forum member Stockley listed a number of key features that clearly demonstrate bias in the present paper:

1. “This is the second attempt by the same group of researchers to disprove the beneficial relationship between light to moderate alcohol consumption and cardiovascular health. As described below, the first attempt (Fillmore et al, in 2006) was discredited by reanalysis of the studies. Reanalysis showed it to be biased, as repeat studies adjusting for the claimed bias still showed a cardioprotective effect for regular light to moderate alcohol consumption.

2. “Numerous meta-analyses have been undertaken over the past 10 years that have adjusted for this proposed bias and they have consistently shown that there is a cardioprotective effect for regular light to moderate alcohol consumption.

3. “In this new paper, Stockwell et al have again biased their meta-analysis by ‘cherry picking’ a small number of studies for their meta-analysis – they discarded 2,575 studies and analysed only 87. The studies that they analysed related reported intake to disease, but they carefully avoided hundreds of validated studies that showed reduced disease among moderate drinkers.

4. “Stockwell et al seem to have deliberately pretended that the many animal and human studies over the past four decades that have provided extensive evidence for the biological mechanisms supporting the findings that light to moderate alcohol consumption is cardioprotective do not exist.”

The use of biased previous research in the present paper: One of the earliest confounders recognized by epidemiologists was that some of the “non-drinkers” in their studies were former heavy drinkers, and had stopped drinking due to adverse health effects. Scientists then began to collect precise data on previous drinking, to better control for drinking pattern (previous intake, regular versus binge drinking), smoking, physical activity, obesity, and other risk factors for disease and mortality. In almost every well-conceived and controlled study done, it was found that when ex-drinkers were not included in the referent group (and the group consisted only of lifetime abstainers) and other known confounders were also adjusted for, there was still a strong “J-shaped curve” for CVD and mortality for moderate drinkers. This pattern has been found consistently in studies from North and South America, Europe, and Asia, in cultures where alcohol consumption varies from the occasional subject to the large majority of people.

In the present study, the authors have resurrected the arguments that they and others raised decades ago. They express concern about, for example, the inclusion of sick quitters in the non-drinking population in an effort to demonstrate that epidemiologists have not properly adjusted for confounders in determining the presence of the J-shaped curve for CVD and mortality. In fact, the issues raised by Fillmore et al had been recognized by others in the field as early as 1995 and suitable corrections have been taken since then. The authors of this paper repeat themselves by quoting their previous and discredited paper by Fillmore et al. In 2006 Dr. Fillmore presented her findings at a large international symposium in Boston of leading scientists working in this field; these discussions led to a strong consensus of the scientists at this meeting that the so-called “errors” in previous studies, which Fillmore used to exclude specific results (in the same manner as this report) were not appropriate. These exclusions led her paper to have inaccurate results, as described in a complete supplement of the Annals of Epidemiology in 2007 (Ellison et al). Forum members consider it inappropriate that the authors of the present study continue to rely on discredited reports, as well as demonstrably flawed techniques in their new analysis.

Reviewer de Gaetano had further information on this topic: “In our ‘historical’ meta-analyses published in 2002 and 2006 by Di Castelnuovo et al, we had carefully addressed the problems raised by Stockwell and colleagues in this paper: inclusion or not of former drinkers in the control groups, adjustment for any possible variables including socio-economic status, year of publication and so on. In each analysis, we observed the classical J curve both on fatal and non fatal CV events and total mortality. Sometimes the adjustment resulted in even improved benefit by wine or beer moderate consumption.

“For example, in our 2006 pape we wrote: ‘A subgroup analysis restricted to studies that excluded either ex-drinkers or very light drinkers from the reference group generated a pooled curve that indeed predicted a lower (though statistically significant) protection, confirming the importance of properly selecting the reference group in studies on alcohol and health. The degree of association was lower in adjusted studies, as might be expected in view of several confounding factors characterizing observational studies on drinking habits; however, the benefit of light to moderate drinking remained in a range of undoubted public health value (15%-18%). Although residual confounding cannot be excluded, it would be very unlikely to modify the scenario in a substantial manner. We found indeed that when adjusted and unadjusted data derived from the same studies were compared, the maximum protection conferred by light to moderate drinking only decreased from 19% to 16%; we can thus presume that, even in the pessimistic hypothesis that residual confounding would have the same strength in lowering the protection as that of known confounding, the ‘real’ (maximum) protection against total mortality associated with low levels of alcohol consumption would still be higher than 10%. A similar reasoning would also apply to the harm associated with heavier drinking.’”

Forum members found it interesting that the authors of the present paper did not comment on the conclusions of Roerecke and Rehm from a recent meta-analysis on the role of confounders in explaining the observed association of alcohol with health outcomes. In that paper, these scientists stated: “Results from our quantitative meta-analysis showed that drinkers with average intake of < 30 g/day and no episodic heavy drinking had the lowest IHD (ischemic heart disease) risk (relative risk = 0.64, 95% confidence interval 0.53 to 0.71). Drinkers with episodic heavy drinking occasions had a risk similar to lifetime abstainers (relative risk = 1.12, 95% confidence interval 0.91 to 1.37).”  Further, in the same paper, these two authors, who have traditionally exhibited real skepticism about a beneficial effect of drinking on disease, concluded: “For drinkers having one to two drinks per drinking day without episodic heavy drinking, there is substantial and consistent evidence from epidemiological and short-term experimental studies for a beneficial association with IHD risk when compared to lifetime abstainers. The alcohol-IHD relationship fulfills all criteria for a causal association proposed by Hill.”

Ignoring a vast amount of experimental data:  There have been literally thousands of animal experiments that demonstrate that the administration of alcohol, as well as polyphenols present in wine and beer, decreases not only risk factors for CVD (abnormalities in lipids, coagulation factors, glucose metabolism, endothelial function, inflammatory factors, etc.) but also decreases the development of coronary and aortic atherosclerosis.  An increasing number of human clinical trials have shown similar favorable effects on many such risk factors (as well summarized by Brien et al).  The question is not does the moderate intake of alcohol beverages lower the risk of cardiovascular disease; we know that is the case from experiments and clinical trials.

Forum members realize that in observational studies, epidemiologists have to rely on their human subjects to tell them what they actually drink, and there is always the chance that there will be mistakes in their reporting, especially under-reporting of alcohol. However, over many decades, we have become better able to identify and adjust for such potential factors, and are increasingly able to recognize under-reporting of alcohol (Klatsky et al). Currently, data from essentially all cohort studies, including those that have been done exceptionally well and with control for all known confounders, are consistent with the animal data. They support what we have seen from experiments: light to moderate drinking decreases the risk of atherosclerosis and is one cause of a lower risk of developing cardiovascular disease.

Forum member Mattivi had some cogent remarks: “Science should serve the truth and to do it needs to take due account of all of the evidence, including controlled clinical trials in humans and animal models. What I find more disturbing in this paper is the lack of an overall vision, missing the opportunity (that is at the heart of the meta-analysis) to learn and build an overall view from the accumulation of collective scientific expertise. This setting, incompatible with the selection of convenience (cherry picking) leads unavoidably to repeat, with monotonic obstinacy, the same mistakes and the same misdeeds.” In addition, Reviewer Lanzmann-Petithory noted: “The authors do not mention wine in their paper, despite much scientific data indicating that wine is different from other beverages in terms of health effects.” Added reviewer Finkel: “We are all tired of having to counter the same self-serving polemics over and over again with unperturbed demeanor.”

Forum member Skovenborg provided a recent notation in the BMJ relating to this topic, in reference to a pre-publication article by Trinquart et al. “Science happens when you examine a hypothesis and use experiments to contest it. Non-science is what happens when you believe a hypothesis and quote all the evidence you can to support it, while ignoring the rest. A superbly illustrated bibliometric study shows how people believing and disbelieving in the connection between salt intake and cardiovascular disease cite different sources: the lines of non-science radiate from competing clusters of evidence.”

“We need less research, we need better research, we need research for the right reasons:”   Reviewer Keil quoted the above from Ioannidis at Stanford (Ioannidis, Ioannidis et al). Added Keil: “My plea is to apply this statement to the important field of research on the health aspects of alcohol consumption. In this field the integrity of science is of paramount importance, because alcohol is loaded with emotions and obviously many scientists, or so-called scientists, have great problems to discern between their emotions and the scientific data.”

Several Forum members supported Keil’s statement that sometimes papers reflect the emotions and “political correctness” of their authors rather than science. Reviewer Lanzmann-Petithory reminded us of the dangers that such an approach can result in politics, rather than science, leading to policy. She especially recalled the 1997 paper by Seltzer that described how political influence and “correctness” were used by the National Institutes of Health to block the publication of early data from the Framingham Heart Study showing a large and significant decrease in the risk of coronary heart disease from alcohol consumption.

Do we need a randomized clinical trial (RCT) for alcohol and CVD? Forum member Estruch considered that it is important that we have a RCT of the effects of alcohol consumption on disease outcomes. “A large long-term randomized clinical trial comparing moderate drinkers versus abstainers on hard end-points (all-cause mortality, acute myocardial infarction, and stroke) is definitely needed. According to the hierarchy of quality of evidence-based medicine, the results of cohort studies have always some component of bias. The highest quality of the results is only achieved by randomized controlled trials. Without the results of these studies, we will continue having to base our knowledge on the results of cohort studies again and again.”

Member Ursini disagreed: “I would like to add a further comment on a possible RCT relating alcohol to health.  This is definitely a great analytical tool for prescription drugs, but not so good for foods.  Applying the pharmacological tools to food is extremely expensive and often ethically unacceptable, and the probability of getting a (false) negative result is extremely high.  Can we imagine a RCT showing that broccoli decreases the risk of cancer?  The risk is high of disregarding forever broccoli as a healthy food.  What we need is an integrated set of scientific information, from basic science, animal studies, and different epidemiological studies to produce a consensus on an acceptable ‘beneficial’ effect extending the consensus on the notion of GRAS. A ‘GRAB (generally rated as beneficial) would be most welcome….we need it for many substances, not only for alcohol and wine.”

Reviewer Keil also had comments on this subject: “RCTs are important, but please remember that we never had an RCT on smoking; instead we have had excellent prospective cohort studies and insight from excellent scientists such as Richard Doll and Richard Peto.  In public health, most decisions have to be made in the absence of RCTs.  Just think of the much debated field of air pollution.”

Implications of this paper for current advice on drinking:   Forum member Goldfinger commented: “This is a very disturbing paper, as meta-analyses are recognized as an important compilation of data sources to recognize important trends. The biased selection of studies that are included undermines the value of the paper, but more importantly promulgates misinformation in the name of appropriate scientific method. Failure to acknowledge the robust body of knowledge that supports the opposite conclusion, and disqualification of extensive animal and cell culture studies that offer plausible biologic explanation of observed benefits, is unconscionable.” A comment by Aaron A. Carroll, MD, in the New York Times of 3/23/16 echoes the conclusions of the Forum: “The evidence still says that a moderate amount of alcohol appears to be safe, and that it might even be healthy for many people. There’s nothing in this new analysis that would make me change my mind.”

Forum Summary

The stated purpose of this new analysis was to determine whether misclassifying former and occasional drinkers as abstainers and other potentially confounding study characteristics underlie observed positive health outcomes for low volume drinkers in prospective studies of mortality.  Unfortunately, the authors include in their analyses a number of old epidemiologic studies and do not acknowledge that when the “errors” that they have commented on in the past (such as including heavy ex-drinkers in the no-alcohol referent group) have been dealt with in the majority of studies over the past decade.  The authors still exclude the vast majority of these well-done studies in their new meta-analysis. Results of essentially all studies that adjust for their concerns continue to show a significant and meaningful reduction in the risk of cardiovascular disease (CVD) and total mortality from the moderate intake of wine and alcohol.

Forum members note how very selective the authors are in choosing which papers to include in their new analyses: they identified 2,575 studies on the subject, analyzed 87, but then they found some reason to exclude almost all of these studies to reach a conclusion that “ . . . there was no significant protection for low-volume drinkers (RR = 1.04, 95% CI [0.95, 1.15])” based on what is apparently only 6 remaining studies!

The paper ignores the comments by two other scientists (Roerecke & Rehm) who, like these authors, have in the past been very concerned that confounding and errors weaken the purported relation between alcohol and a lower risk of cardiovascular disease, but have more recently concluded: “For drinkers having one to two drinks per drinking day without episodic heavy drinking, there is substantial and consistent evidence from epidemiological and short-term experimental studies for a beneficial association with IHD risk when compared to lifetime abstainers. The alcohol-IHD relationship fulfills all criteria for a causal association proposed by Hill.”

The authors of the present paper also ignore the immense amount of experimental data, not only animal experiments but trials in humans, that have described the mechanisms by which moderate alcohol and wine intake have been shown to decrease essentially all of the risk factors for CVD, including low HDL-cholesterol, elevated LDL-cholesterol, endothelial dysfunction, coagulopathies, inflammation, abnormal glucose metabolism, and many others. The consistent finding of lower CVD risk among moderate drinkers in all well-done cohort studies is strongly supported by experimental evidence of the mechanisms.

In the opinion of Forum members, the present paper markedly distorts the accumulated scientific evidence on alcohol and CVD. As stated by one Forum member, “The biased selection of studies that are included undermines the value of the paper, but more importantly promulgates misinformation in the name of appropriate scientific method. Failure to acknowledge the robust body of knowledge that supports the opposite conclusion, and disqualification of extensive animal and cell culture studies that offer plausible biologic explanation of observed benefits, is unconscionable.”

The Forum concludes that the overwhelming body of observational scientific data, as well as an immense number of experimental studies, support the contention that, for most middle-aged and older men and women who choose to do so, the regular consumption of small amounts of an alcoholic beverage can be considered as one component of a “healthy lifestyle.” Such a habit has been shown to be associated with a lower risk of cardiovascular disease and of total mortality.

References from Forum Review

Brien SE, Ronksley PE,Turner BJ, Mukamal KJ, Ghali WA. Effect of alcohol consumption on biological markers associated with risk of coronary heart disease: systematic review and meta-analysis of interventional studies. BMJ 2011;342:d636; doi:10.1136/bmj.d636.

Carroll AA. In Defense of Moderate Drinking (Again). NY Times 3/23/16. Available at

Di Castelnuovo A, Rotondo S, Iacoviello L, Donati MB, de Gaetano G. Meta-analysis of wine and beer consumption in relation to vascular risk. Circulation 2002;105:2836-2844.

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. Archives of Internal Medicine 2006;l166:2437-2445.

Ellison RC, Editor. The harms and benefits of moderate drinking: summary of findings of an international symposium. Ann Epidemiol. 2007;17(Suppl):S1-S115.

Fillmore K, Kerr W, Stockwell T, Chikritzhs T, Bostrom A. Moderate alcohol use and reduced mortality risk: systematic error in prospective studies. Addiction Res Theory 2006;14:101–132.

Ioannidis JPA. Why Most Published Research Findings Are False. PLoS Med 2005;2(8):e124. doi:10.1371/journal.pmed.0020124.

Ioannidis JPA, Greenland S, Mlatky MA, Khoury MJ, Macleod MR, Moher D, Schulz KF, Tibshirani R. Increasing value and reducing waste in research design, conduct, and analysis. Lancet 2014;383:166–175.

Klatsky AL, Udaltsova N, Li Y, Baer D, Tran HN, Friedman GD. Moderate alcohol intake and cancer: the role of underreporting. Cancer Causes Control 2014;25;693-699.

Roerecke M, Rehm J. Alcohol consumption, drinking patterns, and ischemic heart disease: a narrative review of meta-analyses and a systematic review and meta-analysis of the impact of heavy drinking occasions on risk for moderate drinkers. BMC Medicine 2014;12:182.

Seltzer CC. Conflicts of Interest and Political Science. J Clin Epidemiol 1997;50;627-629.

Trinquart L, Johns DM, Galea S.  Why do we think we know what we know? A meta-knowledge analysis of the salt controversy.  Int J Epidemiol 2016: pre-publication doi: 10.1093/ije/dyv184.

*                *                      *                      *                      *                      *                *

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

Elizabeth Barrett-Connor, MD, Distinguished Professor, Division of Epidemiology, Department of Family Medicine and Public Health and Department of Medicine, University of California, San Diego, La Jolla, CA USA

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

R. Curtis Ellison, MD, Professor of Medicine & Public Health, Boston University School of Medicine, Boston, MA, USA

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

Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA

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/Cardiology, Praticien Hospitalier Hôpital Emile Roux, Paris, France

Fulvio Mattivi, MSc, Head of the Department of Food Quality and Nutrition, Research and Innovation Centre, Fondazione Edmund Mach, in San Michele all’Adige, Italy

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

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

Fulvio Ursini, MD, Dept. of Biological Chemistry, University of Padova, Padova, Italy

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