Critique 219: A global overview of alcohol consumption and health – 3 September 2018

GBD 2016 Alcohol Collaborators. Alcohol use and burden for 195 countries and territories, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016. Published online August 23, 2018

Authors’ Summary

Background Alcohol use is a leading risk factor for death and disability, but its overall association with health remains complex given the possible protective effects of moderate alcohol consumption on some conditions. With our comprehensive approach to health accounting within the Global Burden of Diseases, Injuries, and Risk Factors Study 2016, we generated improved estimates of alcohol use and alcohol-attributable deaths and disability-adjusted lifeyears (DALYs) for 195 locations from 1990 to 2016, for both sexes and for 5-year age groups between the ages of 15 years and 95 years and older.

Methods Using 694 data sources of individual and population-level alcohol consumption, along with 592 prospective and retrospective studies on the risk of alcohol use, we produced estimates of the prevalence of current drinking, abstention, the distribution of alcohol consumption among current drinkers in standard drinks daily (defined as 10 g of pure ethyl alcohol), and alcohol-attributable deaths and DALYs. We made several methodological improvements compared with previous estimates: first, we adjusted alcohol sales estimates to take into account tourist and unrecorded consumption; second, we did a new meta-analysis of relative risks for 23 health outcomes associated with alcohol use; and third, we developed a new method to quantify the level of alcohol consumption that minimises the overall risk to individual health.

Findings Globally, alcohol use was the seventh leading risk factor for both deaths and DALYs in 2016, accounting for 2•2% (95% uncertainty interval [UI] 1•5–3•0) of age-standardised female deaths and 6•8% (5•8–8•0) of agestandardised male deaths. Among the population aged 15–49 years, alcohol use was the leading risk factor globally in 2016, with 3•8% (95% UI 3•2–4•3) of female deaths and 12•2% (10•8–13•6) of male deaths attributable to alcohol use. For the population aged 15–49 years, female attributable DALYs were 2•3% (95% UI 2•0–2•6) and male attributable DALYs were 8•9% (7•8–9•9). The three leading causes of attributable deaths in this age group were tuberculosis (1•4% [95% UI 1•0–1•7] of total deaths), road injuries (1•2% [0•7–1•9]), and self-harm (1•1% [0•6–1•5]). For populations aged 50 years and older, cancers accounted for a large proportion of total alcohol-attributable deaths in 2016, constituting 27•1% (95% UI 21•2–33•3) of total alcohol-attributable female deaths and 18•9% (15•3–22•6) of male deaths. The level of alcohol consumption that minimised harm across health outcomes was zero (95% UI 0•0–0•8) standard drinks per week.

Interpretation Alcohol use is a leading risk factor for global disease burden and causes substantial health loss. We found that the risk of all-cause mortality, and of cancers specifically, rises with increasing levels of consumption, and the level of consumption that minimises health loss is zero. These results suggest that alcohol control policies might need to be revised worldwide, refocusing on efforts to lower overall population-level consumption.

Forum Comments

Problems in combing data from very divergent cultures: From the title alone, Forum members realized that the data presented in this paper would be of limited value in setting drinking guidelines, and would not provide important data relating alcohol consumption to health that could be applied to individual cultures or people. The key problem in that by combining data from widely divergent populations into one analysis, the investigators make it impossible to consider the strong effects that social and cultural factors of individual populations have in modifying the effects on alcohol on health. For example, by combining data from countries where ischemic heart disease and stroke are minor causes of death (where there may be a 10% increase in total death rates from alcohol) with data from western industrialized countries where such ischemic diseases are the leading causes of death (where there may be a 10% decrease in total death rates from alcohol intake), you will end up with zero effect of alcohol on death rates. Such an estimate does not provide information relevant to either population: in fact, it applies to no one.

As Rehm & Room recently pointed out: “There are strong cultural norms guiding heavy drinking occasions and loss of control. These norms not only indicate what drinking behaviour is acceptable, but also whether certain behaviours can be reported or not. . . . This explains the multifold differences in incidence and prevalence of alcohol-use disorders, even between countries where the average drinking levels are similar.” These authors conclude: “Current practice to measure alcohol-use disorders based on a list of culture-specific diagnostic criteria results in incomparability in the incidence, prevalence or disease burden between countries. For epidemiological purposes, a more grounded definition of diagnostic criteria seems necessary, which could probably be given by using heavy drinking over time.”

In our Forum critique of the paper referred to above by Rehm and Room, we noted: “We know that the net effect of alcohol consumption relates to the amount of alcohol, the type of beverage, the rapidity of consumption, whether drinking with or without food, and surely a number of other cultural and genetic factors. What we are often unsure about is what the cultural context of drinking is for an individual subject or subjects in a certain population: different cultures seem to help control, or not control, the risk of drinking excessively. . . . these factors complicate the comparison of results of epidemiologic studies from different cultures” (the Forum critique is available at

Reviewer Van Velden stated: “Alcohol consumption is part of a healthy lifestyle, hence the adverse effect of low socio-economic groups with inadequate diet and social norms of responsible alcohol consumption. The effect of moderate exercise on cardiovascular disease also has to be taken into consideration. There are just too many variables in different cultural societies to group all of them together to come to a rational conclusion.” Added Forum member Goldfinger: “Again, looking at alcohol as a single across-the-globe entity fails to take into account the large difference of responsible moderate consumption versus abusive, excessive, periodic binge drinking, etc., which may be associated with cultural and socioeconomic differences of populations studied. The net effect of pooled data is irrelevant and uncontrolled for confounders.”

Importance of socio-economic status in modifying effects of alcohol: Even within the more industrialized countries, data increasingly point out how lower socio-economic strata (SES) subjects show many more adverse effects from alcohol intake than do higher SES subjects, even when the reported amount consumed is similar. An excellent example of the modification of health effects from alcohol that are related to SES factors is based on data from more than 50,000 subjects, representative of the population, in the Scottish Health Surveys by Katikireddi et al. These authors stated: “Low socioeconomic status was associated consistently with strikingly raised alcohol-attributable harms, including after adjustment for weekly consumption, binge drinking, BMI, and smoking. Evidence was noted of effect modification; for example, relative to light drinkers living in advantaged areas, the risk of alcohol-attributable admission or death for excessive drinkers was increased (hazard ratio 6.12, 95% CI 4.45–8.41 in advantaged areas; and 10.22, 7.73–13.53 in deprived areas).” These authors concluded: “Disadvantaged social groups have greater alcohol-attributable harms compared with individuals from advantaged areas for given levels of alcohol consumption, even after accounting for different drinking patterns, obesity, and smoking status at the individual level.” Unfortunately, in the presents paper data were compared only between countries with higher levels of income and lower levels of income, with no adjustments for SES differences within countries.

In our Forum critique of the Scottish paper (, Forum member de Gaetano noted: ““The Scottish findings appear to be in line with evidence from the “Moli-sani” study, an Italian cohort of 25,000 men and women aged >35 y randomized from the general population, showing a similar interaction between Mediterranean diet (MD) and SES factors in relation to risk of developing cardiovascular disease (Bonaccio et al). Basically, we found that adherence to a MD was significantly associated with lower CVD risk in higher but not in lower SES groups, with SES acting as an effect modifier of such association.” Very similar findings, of SES factors strongly modifying the health effects of drinking, has been reported in many other studies (Leyland et al, Mäkelä et al, McDonald et al, Probst et al, Mackenbach et al, Towers et al). Compared with differences in such factors within populations, the differences across the wide variety of cultures represented in the present paper are far greater.” As Forum member Ellison stated: “Unfortunately, the very large number of analyses presented in the current paper provide combined data from many diverse populations, but the results may apply to no one group of people.”

Forum member Skovenborg emphasized the importance of other lifestyle factors: “I find that the meta-analysis methods in this study have a high degree of complexity, and sometimes it is difficult to evaluate the assumptions of the authors. I just want to note that in a prospective study of English and Scottish cohorts (Perreault et al) the association between alcohol intake and cancer mortality was nearly nullified among individuals who met the physical activity recommendations, which (along with most other lifestyle habits) was not included in the present analyses. Further, in the PLCO Cancer Screening Trial (Kunzmann et al), the association of lifetime alcohol use with cancer risk in older adults was modified by beverage choice, with the lowest risk in wine drinkers and the highest risk in spirits drinkers; agina, the present study did not report findings by type of beverage.” Forum member Teissedre was also concerned that “The analyses failed to account for many other factors related to health, including the intake of fruits and vegetables, salt, meat, fish, sugar, etc. Further, there were no adjustments according to type of beverage or whether the alcohol was consumed with or without food. A lot of possible biases are ignored in this study. To take only one component separately (alcohol) in this study without considering many other important factors makes no sense.”

How do you interpret disability-adjusted lifeyears (DALYs)? It is unfortunate that the authors of this paper do not present date on the effect of alcohol on total mortality for the two age groups; instead, they only give data on alcohol-related mortality and disability-adjusted years. The thrust of this paper is the strong adverse effects of heavy drinking among young people, which is associated with many “lost” years of productivity, hence changes in their DALYs. It is obvious that if a teenager dies from alcohol use, he or she will have a very large number of years missing, in comparison with a teenager surviving without disability, for being a productive citizen. Hence, the argument goes that if one takes the large amount of money this person would have made from working for so many years, this would result in a large figure for loss of income to the economy. When such calculations are done, however, they rarely include the costs to society of living: how much it costs the government, businesses, and the public in general to support years of life, especially when the person becomes aged and perhaps infirm. Some scientists argue that prolonging life of some subjects with terminal cancers, dementia, or certain other diseases greatly increases health care costs. The goal of many aging people is to live well as long as possible, but then have a short period of severe disability, thus have a very low number of DALYs.

How do you estimate the exposure to alcohol? Ellison noted further: “The only exposure variable considered in the present study was the average intake of alcohol, in grams/day of ethanol, with no data on the type of beverage, the pattern of drinking, or many unmeasured cultural differences among the countries that were included in their analyses. Also, the reported intake was altered by adjustments for unrecorded alcohol intake (as from home brewing, illicit production, local beverages, or alcohol sold as a non-alcohol product, as well as from sales data from each country) and estimates of the proportion of the total intake consumed by tourists in each country. Their individual estimates were thus based both on the reported intake and some way of combining such data with values from the population in general. The result was that, as the authors state, ‘For a given location, individually reported data on consumption were rescaled so that they aggregated to the estimates of population –level consumption.’ While this may be well-intentioned, it causes problems in comparing the exposure to alcohol in this study with previously reported studies.”

Combining data from young and older people when evaluating alcohol’s effects on mortality: In this paper, for some analyses the authors separate effects of alcohol among subjects less than and greater than the age of 50 years. But, they often compare all results, seeking to get a specific singular risk estimation in the population overall, by combining results for all ages. Given that many of the adverse effects in the young relate to binge drinking, drinking outside of meals, etc., it is especially important to know the pattern of drinking when relating alcohol to health. As an example, in a report of deaths in Canada related to alcohol, Rehm et al in 2007 reported a large number of deaths in the young among “moderate drinkers,” when all drinkers reporting up to a certain amount of alcohol, but not considering the pattern of drinking, were included. However, when he defined truly “moderate drinkers” by excluding those reporting binge drinking, the vast majority of deaths in the young attributed to moderate drinking were no longer present. As discussed below, guidelines for alcohol consumption must be directed at specific age-groups if they are to be effective in reducing abuse while not reducing light-to-moderate drinking that has been shown to improve health among older adults.

Need for specific data among typical drinkers, not heavy drinkers: While the very large number of figures in this paper are impressive, they all show a single curve summarizing results extending from 0 to 12.5 drinks per day (!), even though in no populations are there a large number of people consuming at this upper limit. (Further, in some instances there are no data available in their studies for subjects drinking above around 5 drinks/day or more.) We already know that high levels of drinking lead to severe health problems: what would be preferred is presenting the risk at the levels of intake usually seen in different populations, say, from none up to an average of two or three drinks per day; these are the levels of consumption for which data would be useful. Detailed information on this level of drinking (say, differences in outcome going from none to 1 drink/day) cannot be evaluated from data presented from these analyses. And, in the present paper, the authors include consumption up to 0.8 standard drinks per day in their “zero” intake category; Forum members contend that 0.8 drinks/day is close to the level considered “moderate” in many guidelines – e.g. a recommended level of “a drink a day” — and this causes a problem in using their data to look for any potential benefits of light drinking. No one advises people to “drink more,” and if the advice of these authors to abstain completely from alcohol was followed, it is likely that many light drinkers (who have been shown to have better health than abstainers) would stop their drinking.

Are there implications for setting policy from these analyses? The authors spend much of their paper focusing on their contention that these analyses have large implications for setting policy around the world. However, their failure to consider specifics of the culture for which the policy is being formulated, such as other lifestyle factors, the pattern of drinking, underreporting of alcohol intake, etc., greatly limits their ability to do so. As pointed out so clearly by Harding and Stockley, “A comparison of world-wide 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 world-wide. Such a view, however, would be an over-simplification 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 evidence on the beneficial health effects of moderate consumption.”

As stated by reviewer MeEvoy, “The issues raised, including problems with equating measures of drinking across diverse cultures, failure to take into account patterns of drinking (regular intake versus binge drinking similar amounts), lack of control for critical covariates, such as smoking, exercise, and diet, and failure to take into account differences in disease risk across subpopulations, invalidate the study’s conclusions that any amount of drinking is bad. There is no question that heavy drinking is harmful to health, and that alcohol misuse contributes to an excess of deaths and disability globally.”

McEvoy continued: “I think that Aaron Carroll’s commentary on this paper in the New York Times (published on August 28, 2018), nicely summarizes many of limitations of this study and puts the results into perspective.” (While not a peer-reviewed scientific publication, the Op-Ed was by Aaron E. Carroll, MD; Professor of Pediatrics at Indiana School of Medicine.) Many Forum members considered that he succinctly summarized some of the problems of the present paper, when he wrote: “Observational data can be very confounded, meaning that unmeasured factors might be the actual cause of the harm. Perhaps people who drink also smoke tobacco. Perhaps people who drink are also poorer. Perhaps there are genetic differences, health differences or other factors that might be the real cause. There are techniques to analyze observational data in a more causal fashion, but none of them could be used here, because this analysis aggregated past studies — and those studies didn’t use them.” Adds reviewer Ellison: “When a doctor is advising a high- (or low-) SES patient in New England of a certain age and associated health conditions about the risks and benefits of moderate alcohol consumption, the physician needs to utilize data appropriate to that subject in his or her environment, not data that may be appropriate for a subject in a third-world country with a very different culture and lifestyle.”

Forum member Finkel provided a good overview of this paper: “Emerging from an avalanche of questionably relevant ‘supportive’ material and what might be a new record number of authors and their affiliations, I can easily say in brief that the overall impressions of other Forum members are well presented. There are a whole lot of assumptions and flaccid associations that try and fail to hold the center of this ‘study.’ Once again, the health benefits of moderate drinking are suppressed. It is too bad that so much busy work went to accomplish so little, and worse that the public and the public interest might be erroneously influenced by this publication.”

References from Forum critique:

Bonaccio M, Di Castelnuovo A, Pounis G, Costanzo S, Persichillo M, Donati MB, de Gaetano G, Iacoviello L. Cardiovascular Protection by the Mediterranean Diet Differs Across Socioeconomic Groups: Prospective Findings From The MOLISANI Study. Abstract, American Heart Association. Circulation 2016;133:AMP20

Carroll AE. Study Causes Splash, but Here’s Why You Should Stay Calm on Alcohol’s Risks:
Harms increase with each additional drink per day, yet they are much smaller than many other risks in our lives. Op-Ed Comment in New York Times, published August 28, 2018 (

Harding R, Stockley CS. Communicating Through Government Agencies. Annals Epidemiol 2007;17, Supplement 5S:S98-S102.

Katikireddi SV, Whitley E, Lewsey J, Gray L, Leyland AH. Socioeconomic status as an effect modifier of alcohol consumption and harm: analysis of linked cohort data. Lancet Public Health 2017. Online publication May 10, 2017.

Kunzmann AT, Coleman HG, Huang W-Y, Berndt SI. The association of lifetime alcohol use with mortality and cancer risk in older adults: A cohort study. PLoSMed 2018;15:

Leyland A, Dundas R, McLoone P, Boddy FA. Cause-specific inequalities in mortality in Scotland: two decades of change—a population-based study. BMC Public Health 2007; 7: 172.

Mackenbach JP, Kulhánová I, Bopp M, et al. Inequalities in alcohol-related mortality in 17 European countries: a retrospective analysis of mortality registers. PLoS Med 2015; 12: e1001909.

Mäkelä P, Paljärvi T. Do consequences of a given pattern of drinking vary by socioeconomic status? A mortality and hospitalisation follow-up for alcohol-related causes of the Finnish
Drinking Habits Surveys. J Epidemiol Community Health 2008; 62: 728–733.

McDonald SA, Hutchinson SJ, Bird SM, et al. Association of self-reported alcohol use and hospitalization for an alcohol-related cause in Scotland: a record-linkage study of 23 183 individuals. Addiction 2009;104:593–602.

Perreault K, Bauman A, Johnson N, et al. Does physical activity moderate the association between alcohol drinking and all-cause, cancer and cardiovascular diseases mortality? A pooled analysis of eight British population cohorts. Br J Sports Med 2017;51:651-657.

Probst C, Roerecke M, Behrendt S, Rehm J. Socioeconomic differences in alcohol-attributable mortality compared with all-cause mortality: a systematic review and meta-analysis. Int J Epidemiol 2014; 43: 1314–27.

Rehm J, Patra J, Taylor B. Harm, Benefits, and Net Effects on Mortality of Moderate Drinking of Alcohol Among Adults in Canada in 2002. Ann Epidemiol 2007;17, Supplement 5S:S81-S86.

Rehm J, Room R. The cultural aspect: How to measure and interpret epidemiological data on alcohol-use disorders across cultures. Nordic Studies on Alcohol and Drugs 2017;34:330–341. DOI: 10.1177/1455072517704795.

Towers A, Philipp M, Dulin P, Allen J. The “Health Benefits” of Moderate Drinking in Older Adults may be Better Explained by Socioeconomic Status. J Gerontol B Psychol Sci Soc Sci 2018;73:649-654. doi: 10.1093/geronb/gbw152.

Forum Summary

Investigators working with data from the Global Burden of Diseases, Injuries, and Risk Factors Study 2016 have combined data from 195 locations around the world, from 1990 to 2016, for both sexes and for 5-year age groups between the ages of 15 years and 95 years and older. Their purpose was to determine how estimates of the prevalence of current drinking, abstention, the distribution of alcohol consumption among current drinkers in standard drinks daily (defined as 10 g of pure ethyl alcohol) relate to alcohol-attributable deaths and disability-adjusted lifeyears (DALYs). The paper does not present estimates of the effects of alcohol on a key outcome, total mortality.

The paper presents a huge number of analyses with data specific for each contributing center. However, the Forum considers that real problems emerge when they attempt to combine data from many divergent cultures to determine a single association between alcohol consumption and health. Many lifestyle and cultural factors strongly modify the relation of a given amount of alcohol to health and diseases. Such modifying factors include especially the socio-economic status of the individual, circumstances such as drinking with or without food, the pattern of drinking (regular moderate versus binge drinking), the type of beverage (e.g., wine versus spirits), the intention of the individual (drinking to get drunk versus drinking to enhance meals), level of physical activity, etc. When such modifying factors are not taken into consideration, the estimated intake of a given amount of alcohol (even if accurate), provides an incomplete assessment of the effects of alcohol on health.

The authors spend much of their paper focusing on their contention that these analyses have large implications for setting alcohol policy around the world. They then expand primarily on ways of decreasing alcohol intake world-wide through changes in guidelines. However, their failure to consider specifics of the culture for which the policy is being formulated, such as evaluating other lifestyle factors, the pattern of drinking, underreporting of alcohol intake, cultural factors, etc., negates their ability to provide useful information that is applicable to any single population: their guidelines end up applying to no one. And, their specific statement that zero consumption would be preferable everywhere is sharply contradicted by consistent reports from very large, well-done cohort studies (where individual data are available) which indicate that non-drinkers have higher risks of cardiovascular disease and total mortality than regular moderate drinkers who do not binge drink.

Setting guidelines for the public regarding alcohol consumption requires information not only on the reported intake but on other health conditions, as well as genetic, lifestyle, and cultural factors that may modify the effects of alcohol. Further, advice differs by age and the key problems within a country that the guidelines are directed toward improving. In the opinion of the Forum, despite the massive amount of work done by the investigators in the preparation of this paper, the overall combined results from such divergent populations have little applicability in setting guidelines that would lead to avoidance of alcohol abuse in any specific group of people around the world — they apply to no specific population.

* * * * *

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

Yuqing Zhang, MD, DSc, Clinical Epidemiology, Boston University School of Medicine; Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA.

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

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

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

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

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

Professor Andrzej Pająk, Epidemiology and Population Studies, Jagiellonian University Medical College, Kraków, Poland

Linda McEvoy, PhD, Department of Radiology, University of California at San Diego (UCSD), La Jolla, CA, USA

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

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

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

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