Critique 250: Importance of a favorable pattern of drinking (regularly, and with food) on risk of mortality – 28 July 2021

Ma H, Li X, Zhou T, Sun D, Shai I, Heianza Y, Rimm EB, Manson JE, Qi L. 

Alcohol Consumption Levels as Compared With Drinking Habits in Predicting All-Cause Mortality and Cause-Specific Mortality in Current Drinkers.  Mayo Clin Proc 2021;96:1758-1769.

Authors’ Abstract

Objective: To investigate the joint associations of amounts of alcohol consumed and drinking habits with the risks of all-cause mortality and cause-specific mortality.

Patients and Methods: A total of 316,627 healthy current drinkers, with baseline measurements between March 13, 2006, and October 1, 2010, were included in this study. We newly created a drinking habit score (DHS) according to regular drinking (frequency of alcohol intake 3 times/wk) and whether consuming alcohol with meals (yes).

Results: During a median follow-up of 8.9 years, we documented 8,652 incident cases of all-cause death, including 1,702 cases of cardiovascular disease death, 4,960 cases of cancer death, and 1,990 cases of other-cause death.  After adjustment confounders and amount of alcohol consumed, higher DHS was significantly associated with a lower risk of all-cause mortality, cardiovascular disease mortality, cancer mortality, or other-cause mortality (Ptrend <0.001, Ptrend = 0.03, Ptrend <0.001, and Ptrend <0.001, respectively).  We observed that the amount of alcohol consumed have different relationships with the risks of all-cause mortality and cause-specific mortality among participants with distinct drinking habits, grouped by DHS. For example, in the joint analyses, a J-shaped association between the amount of alcohol consumed and all-cause mortality was observed in participants with unfavorable DHS (Pquadratic trend = 0.02) while the association appeared to be U-shaped in participants with favorable DHS (Pquadratic trend = 0.003), with lower risks in those consuming greater than or equal to 50 g/wk and less than 300 g/wk.

Conclusion: Our results indicate that alcohol consumption levels have different relationships with the risk of mortality among current drinkers, depending on their drinking habits.

*                *                      *                      *                      *                      *                      *

Accompanying editorial: Elagizi A, O’Keefe EL, O’Keefe JH.  Here’s to Your Health: Why a Drink With Dinner Might Improve Longevity.  Mayo Clin Proc 2021;96:1706-1709.

Forum Comments

This is an analysis based on a very large data base of alcohol consumers (subjects reporting > 0 g of alcohol/week) from the UK Biobank (analyses based on 316,627 subjects); median follow up of 8.9 years.  The investigators excluded subjects who reported large changes in intake in recent years.  They assessed frequency and amount of alcohol intake, with > 0 to < 50 g/week used as a reference category; they also asked if the subject usually consumed alcohol with meals (43.9 % of subjects answered yes).

The authors constructed categories of Drinking Habit Scores (DHS) for subjects; a “favorable” DHS was used for those subjects reporting a drinking frequency of 3 or more days/week and consuming alcohol with meals.  The investigators then related the DHS categories to all-cause and cause-specific mortality with adjustments for all common known confounders/modifiers of alcohol effect.

In addition to self-reported alcohol intake, the investigators carried out some analyses using a genetic risk score based on 90 single SNPs relating to alcohol intake from previous studies rather than the self-reported amount of alcohol consumed; this did not prove to be very informative.

Overall, Forum reviewers agreed with the main outcomes of the study, which indicate that subjects with a favorable DHS (in comparison with other drinkers) had significantly lower all-cause mortality (HR=0.82), CVD mortality (HR=0.84); cancer mortality (HR=0.82), and death from other causes (HR=0.77).  Also the authors report in the text that for all-cause, CVD, and other cause mortality, subjects with a favorable DHS showed a U-shaped or L-shaped curve (similar to the L-shape curve for mortality shown for wine drinkers by Gronbaek et al several decades ago).  Only for cancer mortality was there an increase in risk for subjects in the highest category of alcohol.  Supplemental data provided also show that subjects with a favorable DHS showed the same or lower mortality for all types of beverage, regardless of the amount of alcohol they reported.

Comments from specific Forum members

Forum member Skovenborg noted: “The study by Ma et al has several important merits like the focus on the benefits of regular drinking with meals instead of the harms of irregular binge drinking. Another merit is the study of a cohort of drinkers and thus evading the discussion of abstainer residual confounding issues. The problem of possible underreporting might be of minor importance in this study as you would suppose the underreporting to be non-differential; that is, the degree of underreporting would probably be the same in all the participants whether they are regular drinkers or not and would not affect the risk classification.

“The benefits of regular moderate drinking with meals have been shown in many studies, most of which are referenced by the authors. Several recent studies have confirmed the importance of drinking patterns. Rosoff et al found high educational attainment associated with reduced frequency of binge drinking and reduced risk of alcohol dependence in a Mendelian randomization study.  Probst et al examined the alcohol harm paradox (i.e., the observation that people of low socioeconomic status (SES) tend to experience greater alcohol-related harm than those of high SES, even when the amount of alcohol consumption is the same or less than for individuals of high SES) and found that the greatest difference in harms between low and high SES was for heavy episodic drinking (or risky single occasion drinking), rather than for the mean quantity consumed per month or week.

“In a study of alcohol and total mortality in 142,960 individuals from the MORGAN project, J-shaped association curves of alcohol intake with mortality were observed overall for European countries, however, the magnitude, nadir and persistence of the association somewhat varied, with stronger values observed in Italy and France.  This finding may be explained by the different patterns of alcohol consumption across countries. In Mediterranean countries alcohol is typically consumed during main meals and largely in the form of wine, a pattern that has been recognized as a healthy habit, whereas in non-Mediterranean countries alcohol is not usually consumed during meals, the preferred alcoholic beverage is not wine, and the practice of binge drinking, that is unequivocally an unhealthy habit, is more frequent (Di Castelnuovo et al).  Of note, intake of >20 g alcohol/day was not associated with cancer mortality among wine-preferring (>70%) participants in the MORGAN Project.  In the MORGAN Project, in fact, there was a completely different pattern seen for subjects in the cohorts from France and Italy (a U-shaped curve with even the highest intake of alcohol not exceeding the risk of abstainers) when compared with subjects from northern Europe (who showed a typical J-shaped curve).  In the present study by Ma et al, moderate alcohol intake (50-200 g/wk) was not associated with cancer mortality in participants with a favorable DHS.

“In the discussion of the mechanisms that might explain the study results, the authors did not include the important effects of first pass metabolism (FPM).  Ingestion of a coincident meal with the ethanol can reduce the peak blood level by about 4 fold at low doses (Levitt et al).  Due to the effects of FPM a considerable amount of alcohol molecules will not reach the systemic circulation.”

Reviewer de Gaetano noted: “I strongly agree with Dr. Skovenborg’s remarks.   Alcohol and especially moderate wine consumption is an integral part of the Mediterranean Diet; from the Moli-sani study, removing moderate alcohol consumption from the diet reduces the CV and total mortality benefit of the diet by 15-20%.”

Forum member Mattivi added: “Actually, people do not drink ‘alcohol’ per se, but as an ingredient of highly variable mixtures; the pattern of drinking is most important.  What makes the difference when a beverage is consumed with food could be in part the delayed and reduced peak of ethanol, but is this enough to provide an explanation?  In the case of wine, there is clear evidence that the polyphenols, especially the proanthocyanidins, are responsible of the capacity of red wine to scavenge peroxy radicals (see as an example, Rigo et al).  Red wine polyphenols can reach in the gut a millimolar concentration that is several orders of magnitudes higher than those achievable in the blood.  It is therefore expected (and it has demonstrated) that red wine taken with food can attenuate, and in some cases fully prevent, the post-prandial insult carried by several unhealthy components present in the meals (as in red meat, fried food, etc) (Natella et al).”

Cultural differences in the patterns of eating and drinking

Forum member Lanzmann-Petithory wrote: “I regret that the authors did not verify the association between wine preference, drinking pattern and end-points, because they had access to the data (for red wine, white wine, beer/cider, fortified wine, and spirits) and the population is large enough.  They could have mentioned the Arif & Rohrer  report from NHANES III on 8,236 men and women, in which the RR for obesity of binge versus no binge drinking was 1.77 (CI 1.18-2.65).  They could have mentioned the PRIME study (Ruidavets et al) that showed that ‘the hazard ratio for hard coronary events for binge drinkers, compared with regular drinkers, was 1.97 (95% CI 1.21 to 3.22).  The hazard ratio for hard coronary events in Belfast compared with in France was 1.76 (95% CI 1.37 to 2.67) before adjustment, and 1.09 (95% CI 0.79 to 1.50) after adjustment for alcohol patterns and wine drinking. Only wine drinking was associated with a lower risk of hard coronary events, irrespective of the country.’   The favorable drinking habit score in the present paper is more or less the French traditional drinking pattern, maybe partly responsible for the French paradox (that does exist, contrary to what some people claim). They could have mentioned the seminal paper of Renaud & deLorgeril which indicated that the traditional French drinking pattern is generally wine with meals, whereas binge drinkers tend to prefer cocktails of soda or beer and vodka or other spirits, and not necessarily with meals.

“The CREDOC in France (Mathe et al) have made a comparison of eating and drinking habits between Americans and the French people.  For Americans, a glass of wine is like a snack that they can take separately from mealtimes.  CREDOC concludes the comparison of French and American approaches to food and alcohol as follows: a substance for the former (the French), and a tool for the latter (the Americans).  As with the food they eat, the French drink wine for pleasure, because they have a taste for it, rather than as a means to a desired effect.  It is not about ‘filling up,’ viewing food as a means to achieve something else, but to perceive it as a substance for its own sake, and as an excuse to be sociable.  This in turn encourages enjoyment of the present.”

Lanzmann-Petithory continued: “In 1957, in ‘Wine and Milk,’ the French writer Roland Barthes wrote: ‘What is characteristic in France is that the converting power of wine is never openly presented as an end: other countries drink to get drunk, and this is accepted by everyone; in France, drunkenness is a consequence, never an aim.  A drink is felt as the spinning out of a pleasure, not as the necessary because of an effect which is sought: wine is not only a philter, it is also the leisurely act of drinking.  The gesture has here a decorative value, and the power of wine is never separated from its modes of existence, unlike whisky, for example, which is drunk for its type of drunkenness the most agreeable, with the least painful after-effects which one gulps down repeatedly, and the drinking of which is reduced to a causal act.’  Unfortunately, the young French generation is losing this traditional attitude and has a troubling trend for binge drinking too, especially with spirits, as in Anglo-Saxon countries (Lanzmann-Petithory)”

Form member Parente added to this discussion:  “While these welcome findings support a Mediterranean pattern of drinking that tilts risk reduction toward health-positive outcomes, their relevance to a U.S. population are somewhat limited by lower adoption of the Mediterranean pattern of eating stateside, i.e., meals built around varied, moderate portions of minimally processed plant-based dishes, seafood, nuts, whole grains, small and infrequent servings of meat, olive oil as the primary fat and sweets as only occasional treats in modest amounts (gelato the delicious exception in warm weather), all consumed in a setting of connectedness over an extended period and generally eschewing sugar-sweetened beverages. (Dietary habits were not described in the present study by Ma et al.)

“In the present study in a UK population, the mean BMI ranged from 26.6 to 27.5 kg/m^2 across all four groups, considered low- to mid-range overweight (BMI 25–29.9).  According to 2017–2018 statistics from CDC, the prevalence of obesity (BMI ≥ 30) among Americans was 42.4% with severe obesity (BMI ≥ 40%) prevalence increasing to 9.2%.  The average U.S. BMI at last count was close to the 30 kg/m^2 obesity cutoff with about 73% of the population overweight or obese.  Obesity-related conditions such as heart disease, stroke, diabetes mellitus and cancer are likely to confound a future study of drinking with meals in a U.S. population, reflecting this upward shift in BMI.

“To the excellent accompanying editorial by Elagizi, O’Keefe and O’Keefe and their comment that ‘Most people do not drink alcohol for its purported health benefits but rather to reduce stress and/or promote positive feelings’, it is worth mentioning that in the Mediterranean pattern of drinking and eating, wine is considered an integral part of the meal that does not require a reason for its inclusion or enjoyment.  Wine is part of the oeno-gastronomic pleasure of the Mediterranean meal and the connected meal experience itself.  When all goes well, the wine makes the food taste better and the food makes the wine taste better, positive feelings both.

“Unlike other patterns of drinking, especially binge drinking, imbibing too much or indiscreetly is frowned upon at the Mediterranean table, a disgrazia that typically falls outside of social acceptance.  This acknowledgment of wine’s place at the table, intended for consumption spaced out over various plates of food and shared enjoyment, contrasts with ‘doing shots’ unrelated to meals and the aspirational drinking to blackout that characterizes many binge drinking, youth drinking and pre-drinking behaviors.  The Mediterranean pattern of eating is making inroads inland and deserves continued scrutiny alongside that of drinking, even though two-hour meals are relegated to fantasy for most in the ‘go-go’ sector.  Martínez-González et al discuss problems that occur when one dietary pattern is being recommended as a change for a population from its previous usual pattern.”

Reviewer Van Velden noted:  “Alcohol consumption with meals, especially wine, is usually found to be associated with a responsible holistic lifestyle that includes physical, emotional and spiritual aspects.  This study supports the findings that a plant-based diet containing abundant anti-oxidants (including moderate red wine consumption) helps protect the body from the chronic and degenerative lifestyle-related diseases often associated with a high animal-protein diet.  We appreciate that a relaxing meal with friends promotes wellness beyond physical aspects alone, and that there are many confounders that must be taken into consideration before reaching a conclusion of what constitutes a healthy lifestyle.”

Forum member Finkel concluded: “We are finally getting beyond the bare bones of the amply documented J-shaped curve/French paradox complex.  This study, emanating from reliable investigators and composed of plenitude of subjects, adds credence to the growing evidence that the pattern of drinking is as important as the quantity of alcohol consumed.”

Comments on the editorial by Elagizi et al

The editorial in the journal accompanying this paper provides a good overview of the evidence  supporting the benefits of a healthy pattern of drinking.  It discusses why people drink alcohol in the first place (generally not for the health effects) and the role of regular red wine consumption with meals as part of the traditional Mediterranean diet.  It mentions a number of physiologic effects of wine and other alcoholic beverages (on insulin sensitivity, lipids, coagulation, etc.), but also describes potential health benefits from social bonding associated with drinking during meals.  It also warns that some people may have difficulty controlling the amount of alcohol they consume, and the associated dangers of excessive intake on health.

References from Forum critique

Arif AA, Rohrer JE.  Patterns of alcohol drinking and its association with obesity: data from the Third National Health and Nutrition Examination Survey, 1988-1994. BMC Public Health 2005; 5:126.  doi:10.1186/1471-2458-5-126.

Barthes R. Le vin et le lait. (Wine and Milk). In Barthes, Mythologies. Paris: Seuil 1957;74-77.

Di Castelnuovo A, Costanzo S, Bonaccio M, McElduff P, Linneberg A, Salomaa V, et al,  on behalf of the MORGAM Study Investigators.  Alcohol Intake and Total Mortality in 142,960 Individuals from the MORGAM Project: a population-based study.  Addiction 2021, in Press.  doi: 10.1111/add.15593.

Elagizi A, O’Keefe EL, O’Keefe JH.  Here’s to Your Health: Why a Drink With Dinner Might Improve Longevity.  Mayo Clin Proc 2021;96:1706-1709.

Gronbaek M, Deis A, Sorensen TIA, et al.  Mortality associated with moderate intakes of wine, beer, or spirits.  BMJ 1995;310:1165-1169.

Lanzmann-Petithory D.  Commentary on Kerr et al. (2013): The “French Paradox’ versus binge drinking. Addiction 2013;108:1049-1050. DOI:10.1111/add.12211

Levitt DG.  PKQuest: measurement of intestinal absorption and first pass metabolism – application to human ethanol pharmacokinetics.  BMC Clinical Pharmacology 2002;2:4.

Mathe T, Francou A, Colin J, Hebel P.  Comparaison des modèles alimentaires français et états-uniens. (French eating habits compared to Americans.) CREDOC, Research Center for the study and observation of living conditions. Consommation et modes de vie N° 255, ISSN 0295-9976, September 2012. Available at:

Martínez-González MA, Hershey MS, Zazpe I, Trichopoulou A.  Transferability of the Mediterranean Diet to Non-Mediterranean Countries. What Is and What Is Not the Mediterranean Diet. Nutrients 2017;9:1226.  doi:10.3390/nu911122

Natella F, Macone A, Ramberti A, Forte M, Mattivi F. Matarese RM, Scaccini C.  Red wine prevents the postprandial increase in plasma cholesterol oxidation products: A pilot study. British Journal of Nutrition, 2011;105:1718-1723. doi:10.1017/S0007114510005544.

Probst C, Killian C, Sanchez S, Lange S, Rehm J.  The role of alcohol use and drinking patterns in socioeconomic inequalities in mortality: A systematic review.  Lancet Public Health 2020;5:e324-e332.

Renaud S, deLorgeril M.  Wine, alcohol, platelets, and the French paradox for coronary heart disease. Lancet 1992;339:1523-1526.

Rigo A, Vionello F, Clementi G, Rossetto M, Scarpa M, Vrhovsek U, Mattivi F.  Contribution of Proanthocyanidins to the Peroxy Radical Scavenging Capacity of Some Italian Red Wines.  J Agric Food Chem 2000;48:1996–2002.

Rosoff DB, Clarke T-K, Adams MJ, McIntosh AM, Smith GD, Jung J, Lohoff FW.  Educational attainment impacts drinking behaviors and risk for alcohol dependence: results from a two-sample Mendelian randomization study with ~780,000 participants.  Molecular Psychiatry 2021; 26:1119–1132.

Ruidavets JB, Ducimetière P, Evans A, Montaye M, Haas B, Bingham A, et al.  Patterns of alcohol consumption and ischaemic heart disease in culturally divergent countries: the Prospective Epidemiological Study of Myocardial Infarction (PRIME). BMJ 2010;341:c6077. doi: 10.1136/bmj.c6077.

Forum Summary

This is an analysis based on a very large data base of alcohol consumers (subjects reporting > 0 g of alcohol/week) from the UK Biobank (analyses based on 316,627 subjects with a median follow up of 8.9 years).  The authors constructed categories of Drinking Habit Scores (DHS) for subjects; a “favorable” DHS was used for those subjects reporting a drinking frequency of 3 or more days/week and consuming alcohol with meals.  The investigators then related the DHS categories to subsequent all-cause and cause-specific mortality, with adjustments for essentially all of the common known confounders/modifiers of alcohol effect.

Overall, Forum reviewers agreed with the main outcomes of the study, which indicate that subjects with a favorable DHS (in comparison with other drinkers) had significantly lower all-cause mortality (HR=0.82), CVD mortality (HR=0.84); cancer mortality (HR=0.82), and death from other causes (HR=0.77).  In relation to the number of drinks consumed, the authors report that for all-cause, CVD, and other cause mortality, subjects with a favorable DHS showed a U-shaped or L-shaped curve (in that even for the heaviest drinkers, the mortality risk remained the same or lower than that of abstainers).  Only for cancer mortality was there an increase in risk for subjects in the highest category of alcohol intake, leading to a J-shaped curve.

While the authors did not report specifically on the types of beverage being consumed in their favorable drinking pattern, it is the typical pattern generally found in southern Europe, and typically it is wine with meals.  In these societies, wine is considered part of the meal, and is not consumed as “shots” that are usually being taken primarily as a way of getting drunk.  As described in this paper, it is important to separate the health effects of these opposite approaches for consuming a beverage containing alcohol; putting all subjects into groups based only on the grams of alcohol consumed over a period, usually a week, does not take into account the important differences associated with the pattern of drinking.

We are finally appreciating that we are much further along in our research on alcohol and health than the innumerable studies that have focused just the amount of alcohol consumed, or on the J-shaped curve.  We now realize the importance of the pattern of drinking.  This constitutes not only the type of beverage and frequency of drinking, but also the consumption of alcohol with food and the absence of binge drinking.  The pattern of drinking appears to be as important as, or even more important than, the quantity of alcohol consumed.

*                *               *               *              *               *               *                 *               *

Comments on this critique by the International Scientific Forum on Alcohol Research have been 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, MS (Epidemiology), Professor of Medicine, Section of Preventive Medicine & Epidemiology, Boston University School of Medicine, Boston, MA

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

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

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

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

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

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