Critique 245: Effects of different patterns of alcohol consumption on risk of mortality, major cardiovascular events, cirrhosis, and cancer — 25 January 2021

Jani BD, McQueenie R, Nicholl BI, Field R, Hanlon P, Gallacher KI, Mair FS, Lewsey J.   Association between patterns of alcohol consumption (beverage type, frequency and consumption with food) and risk of adverse health outcomes: a prospective cohort study.   BMC Medicine 2021;19:8.

 Authors’ Abstract

Background: Alcohol consumption is a leading contributor to death and disability worldwide, but previous research has not examined the effects of different patterns of alcohol consumption. The study objective was to understand the relationship between different alcohol consumption patterns and adverse health outcomes risk, adjusting for average amount consumed among regular drinkers.

Methods: This was a prospective cohort study of UK Biobank (UKB) participants. Abstainers, infrequent alcohol consumers or those with previous cancer, myocardial infarction (MI), stroke or liver cirrhosis were excluded. We used beverage type, consumption with food and consumption frequency as exposures and adjusted for potential confounding. All-cause mortality, major cardiovascular events-MACE (MI/stroke/cardiovascular death), accidents/injuries, liver cirrhosis, all-cause and alcohol-related cancer incidence over 9-year median follow-up period were outcomes of interest.

Results: The final sample size for analysis was N = 309,123 (61.5% of UKB sample). Spirit drinking was associated with higher adjusted mortality (hazard ratio (HR) 1.25; 95% confidence intervals (CI) 1.14–1.38), MACE (HR 1.31; 95% CI 1.15–1.50), cirrhosis (HR 1.48; 95% CI 1.08–2.03) and accident/injuries (HR 1.10; 95% CI 1.03–1.19) risk compared to red wine drinking, after adjusting for the average weekly alcohol consumption amounts. Beer/cider  drinkers were also at a higher risk of mortality (HR 1.18; 95% CI 1.10–1.27), MACE (HR 1.16; 95% CI 1.05–1.27), cirrhosis (HR 1.36; 95% CI 1.06–1.74) and accidents/injuries (HR 1.11; 95% CI 1.06–1.17). Alcohol consumption without food was associated with higher adjusted mortality (HR 1.10; 95% CI 1.02–1.17) risk, compared to consumption with food. Alcohol consumption over 1–2 times/week had higher adjusted mortality (HR 1.09; 95% CI 1.03–1.16) and MACE (HR 1.14; 95% CI 1.06–1.23) risk, compared to 3–4 times/week, adjusting for the amount of alcohol consumed.

Conclusion: Red wine drinking, consumption with food and spreading alcohol intake over 3–4 days were associated with lower risk of mortality and vascular events among regular alcohol drinkers, after adjusting for the effects of average amount consumed. Selection bias and residual confounding are important possible limitations. These findings, if replicated and validated, have the potential to influence policy and practice advice on less harmful patterns of alcohol consumption.

Forum Comments

Forum member Skovenborg wrote: “I agree that this new paper, based on the large UK Biobank dataset, is very interesting and well done, as are the protocol, the dataset and the analysis leading to the authors’ cautious conclusions.  As a matter of fact, the paper is a good example of the kind of study that many of us have long been asking for: evaluating how factors other than just the weekly amount of alcohol consumption may contribute to the health consequences of drinking.  I have noted that even with non-drinkers excluded from the analysis, you are able to recognize a J-shaped curve for the relation of alcohol to disease outcomes.

“I have some questions about the reported increased total mortality and risk of major cardiovascular events (referred to as MACE and including MI/stroke/cardiovascular death) in people drinking 1-2 times/week (which we assume includes many binge-drinkers) and an increased risk of liver cirrhosis in people drinking daily or almost daily.  From the analysis you observe that the group of people drinking daily are not the same people that are drinking alcohol with food, so they are not ‘Mediterranean style drinkers’ who tend to consume wine with meals on a regular basis.  It would be very hard to understand why the same weekly amount of alcohol would be more toxic for the liver if the consumption split up on 7 days/week rather than 3-4 days/week, and not to mention 1-2 days/week.  Further, regarding alcohol consumption with or without food and risk of liver cirrhosis, the group of people drinking alcohol with food have 0.2% events (HR 1.0) and people that do not drink alcohol with food have 0.5% events, but the adjusted HR is given in the paper as 0.88 for non-food drinkers.  (Intuitively that seems wrong that more events should be associated with a lower HR, but this is not discussed.)

“It is noted further that in terms of the higher risk of MACE for spirits versus red wine (HR 1.31, CI 1.15-1.50), the authors report a total mediating effect of 5 factors in the analysis that explains 16.6% of the effect; smoking (7.8%) makes up almost half of the total effect. This leaves plenty of room for the proposed protective effects of red wine polyphenols.”

Form member Finkel noted: “I have complained often in the past of the absence of attention to beverage type, drinking pattern, and eating while drinking in studies reporting health effects associated with alcohol consumption. Well, here is a study (bless it!) that does pay attention to these factors. I hope it is the leader in a trend.  I view with favor its size and perspective. Its conclusions seem expected, but by having included these important factors it lends validation to many previous though less-complete reports.”

Forum member Ellison also considered this to be a well-done analysis:  “In these analyses, the pattern of drinking was estimated from type of beverage, frequency of any drinking, frequency of binge drinking, and whether or not the beverage was usually consumed with or without food (with adjustments for total weekly amount of alcohol).  Among the 502,536 subjects in the UK Biobank sample, excluded from the analyses were abstainers (n=40,648) and occasional drinkers (n=113,870), so this paper does not judge the effects of moderate drinking versus not drinking.  Instead, it looks at effects of the pattern of drinking among regular drinkers.  (Given the extensive analyses done, it seems unfortunate that the authors did not use this dataset to also judge outcomes related to abstinence as compared with various patterns of consumption.)

“As for the classification of drinkers into categories, I note that 47.8% of drinkers reported 1-14 average weekly units (8 gm of alcohol by British standards) and were classified as ‘low-risk drinkers.’  Another 43.3% reported what was called ‘increasing-risk drinkers’ (15-35 average weekly units for females and 15-50 average weekly units for males).  Thus, the classification in this study for ‘low-risk’ for women (up to 15 units per week or up to about 16 grams of alcohol/day) would be close to the recommended US guidelines for women of no more than 1 ‘typical drink’ of about 14 g/drink per day.  For males, however, the US guidelines are for up to 2 typical drinks/day; this would mean that some (many?) of the male subjects in the present study who consumed an average of between 15 and 50 units/week and were classified as ‘increasing-risk drinkers’ may still have been within the recommended guidelines for alcohol consumption by US standards.  Thus, classifying the males according to US guidelines may well have shown a lower risk for the ‘increasing risk’ category.

“Importantly, the major confounders or modifiers of effect of alcohol were included in their analyses; these included age, smoking, physical activity, SES (based on the Townsend score of deprivation), BMI, systolic blood pressure, total cholesterol, CRP levels, gamma glutamyltransferase levels, self-reported health, and a number of physical and mental health conditions (including hypertension and diabetes, which are often found to be mechanisms of alcohol effect).  The inclusion of factors related to socio-economic status and to smoking is especially needed in such analyses.

“Of the numerous comparisons of adverse outcomes associated with type of beverage consumed, the most striking differences were between red wine drinkers and spirits drinkers, with the latter having greater risk of most outcomes (25% greater for all-cause mortality, 31% greater for MACE, and 48% greater for liver cirrhosis).  In general, consumers of white wine had similar effects on risk as red wine drinkers, and none of the differences were statistically significant.  In comparison with red wine drinkers, consumers of beer or cider had increased risk of all-cause mortality (18%), MACE (16%), and liver cirrhosis (36%).   It is interesting that the risk of new cancer cases or alcohol-related cancer incidence were not affected significantly by type of beverage consumed.

Forum member Mattivi added: “I was not able to grasp why, apparently having enough subjects classified as preferential consumers of white wine (and champagne), the results associated with this subgroup were not further discussed in the main paper.  Looking at the very informative supplementary data, it seems that while the effects of red wine are clearly separated from the other beverages (spirits and the combination of beer and cider), the differences in the adverse health outcomes observed for the two groups red vs white wine were not statistically significant.  It would have been useful if the authors presented combined data from all types of wine.”

Reviewer Parente noted: “As follow on to Dr. Mattivi’s comments, the authors grouped together drinkers of white wine, champagne and fortified wines together in the ‘white wine’ category. Whether the inclusion of fortified wine drinkers in this group was appropriate is debatable.  In their analysis, the white wine category tracks closely with the red wine group, except for cirrhosis (HR 1.21 [0.91–1.61]).  Also, the p values were not significant for the white wine group across the board.  Clearly an analysis vis-à-vis white wine warrants a closer look in future studies.

“Also of interest is the BMI of participants compared to their American counterparts.  BMI in this UK cohort was 27 in the total sample (26.6, 27.2, 27.9 from lowest to highest risk groups). Americans’ BMI (2018 national statistics) has surged past 29 (29.1 men, 29.6 women), close to the obesity cutoff of 30.  This difference may have implications for cancer statistics reported in American studies on alcohol consumption.”

Forum member van Velden commented: “I agree with other members’ comments regarding the relation of wine to better outcomes.  However, it must be stressed that responsible alcohol (wine) consumption must not be seen in isolation from other lifestyle factors such as a good diet, exercise, weight management, no smoking, etc.  Moderate wine consumption is cardioprotective for most individuals without certain genetic factors such as alcohol-dehydrogenase deficiency, as such people cannot tolerate ethanol well.  In short, wine is not a ‘medicine’, but must be seen as part as part of a healthy lifestyle.  There are certainly health benefits in moderate red wine consumption!”

Reviewer de Gaetano wrote: “This is an interesting paper on different health effects of pattern of drinking. We have shown in the Moli-sani cohort of about 25,000 people from a general adult population that moderate alcohol (almost totally wine) drinking is an integral part of the Mediterranean diet, contributing to about 15% to the reduced death risk associated with the whole diet, both in the elderly and in subjects with type 2 diabetes (Bonaccio et al, 2016; Bonaccio et al, 2018).  Now, the UK biobank cohort shows that drinking wine with meals is better than drinking outside meals. I agree that wine is not a drug and its consumption should be considered in the context of different lifestyles, dietary habits first.”

Forum member Goldfinger added: “This is one of few, if any, papers that looked at beverage type and strategy/timing/culture of alcohol consumption in a large and diverse population, with respect to important clinical endpoints.  The prospective design and implementation of this project lend significant credibility to its findings.  I believe that it is a very important paper that reinforces that responsible wine consumption, with meals, and in moderation, is not, nor should be considered, in the same vein as irresponsible alcohol consumption, which is universally agreed to be potentially perilous from a health perspective.  I believe we will come to refer back to this paper frequently for perspective when reviewing future papers that generalize all consumption of alcoholic beverages with adverse outcomes.”

References from Forum critique

Bonaccio M, Di Castelnuovo A, Costanzo S, Persichillo M, De Curtis A, Donati MB, de Gaetano G, Iacoviello L; MOLI-SANI study Investigators.  Adherence to the traditional Mediterranean diet and mortality in subjects with diabetes. Prospective results from the MOLI-SANI study.  Eur J Prev Cardiol 2016;23:400-407. doi: 10.1177/2047487315569409.

Bonaccio M, Di Castelnuovo A, Costanzo S, Gialluisi A, Persichillo M, Cerletti C, Donati MB, de Gaetano G, Iacoviello L.  Mediterranean diet and mortality in the elderly: a prospective cohort study and a meta-analysis.  Br J Nutr 2018;120:841-854.  doi: 10.1017/S0007114518002179.

Forum Summary

This study was designed to determine factors that affect the pattern of alcohol consumption (including the frequency of consumption, type of beverage, with or without food, etc.), rather than just the reported average amount of alcohol, related to all-cause mortality, major cardiovascular events-MACE (MI/stroke/cardiovascular death), accidents/injuries, liver cirrhosis, all-cause and alcohol-related cancer incidence.  It was based on a large prospective cohort study of UK Biobank (UKB) participants and included more than 300,000 subjects; outcomes were ascertained over a 9-year median follow-up period.  Non-drinkers and only occasional drinkers were excluded from the analyses, so this paper does not judge the effects of light or moderate drinking versus not drinking, but only the effects of the pattern of reported alcohol consumption among regular drinkers.

The study showed that in comparison with subjects who consumed red wine more than 50% of the time (the reference group) there were lower risks of many adverse outcomes than among consumers of more than 50% of the time of spirits; the latter had 25% greater risk of all-cause mortality, 31% greater risk of MACE, and 48% higher risk of liver cirrhosis after controlling for total weekly alcohol consumption and relevant confounders.  In comparison with red-wine drinkers, consumers of beer/cider showed an 18% higher risk for all-cause mortality, 16% higher risk of MACE, and 36% higher risk of liver cirrhosis.   There were no statistically significant differences in outcomes between white wine drinkers and red wine drinkers.

Consumption of alcohol with food, versus not with food, showed a 10% (CI 2%-17%) lower risk of all-cause mortality.  While consumers reporting alcohol consumption 3-4 days/week had lower risk of some adverse outcomes than subjects reporting intake on only 1-2 days/week, for some reasons subjects reporting daily or near-daily intake had an increased risk of liver cirrhosis, probably related to under-reporting by some heavy drinkers.

Forum members agree with the conclusions of this study as reported in the Authors’ Abstract: “Red wine drinking, consumption with food and spreading alcohol intake over 3–4 days were associated with lower risk of mortality and vascular events among regular alcohol drinkers, after adjusting for the effects of average amount consumed.”  The results of this well-done study emphasize how inappropriate it is to use just the total average alcohol intake when relating the consumption of alcoholic beverages to the risk of adverse health outcomes.


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

Harvey Finkel, MD, Hematology/Oncology, Retired (Formerly, Clinical Professor of Medicine, 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

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

Fulvio Mattivi, MSc, CAFE – Center Agriculture Food Environment, University of Trento, via E. Mach 1,  San Michele all’Adige, Italy

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

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

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

Creina Stockley, PhD, MSc Clinical Pharmacology, MBA; Principal, Stockley Health and Regulatory Solutions; Adjunct Senior Lecturer, The University of Adelaide, Adelaide, Australia

Matilda Parente, MD, consultant in molecular pathology/genetics and emerging technologies, San Diego, CA, USA

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