Chapter 2 The French Paradox was controversial but this was just the s – Thompson Estate

Chapter 2 The French Paradox was controversial but this was just the start.

In my first Chapter in this series, I outlined the background to the “French Paradox”. The paradox that alcohol may have counteracted the ill effects of high fat consumption in France captured the public imagination, but was controversial from the start, and debate among the expert epidemiologists has been brisk.

Early claims were made that that the paradox was not real, and the explanation for the paradox was simply an under-reporting of coronary heart disease by French doctors on death certificates ([i]).  Comparison of the official death statistics and the more detailed MONICA coronary heart disease mortality rates showed that this was no more than a slur on the reputation of French medicine. There was no convincing evidence of under-reporting in the national statistics, and this seems very unlikely as the sole explanation. ([ii])  A novel explanation, suggested a time lag effect, that the high fat intake in France was relatively recent phenomenon, and that recent dietary changes would eventually catch up with the French, ([iii]) but follow-up research showed the opposite   and cardiovascular mortality rates continue to fall in France. ([iv])  Recent re-appraisal of the role saturated fat in the diet ([v], [vi]) has raised the question the roles assigned in the French Paradox with  saturated fat as the villain and alcohol as the hero, may have been simplistic. Gathering evidence of the beneficial effects of the Mediterranean diet with its co-consumption of a low-moderate intake of wine ([vii])  raises the question that diet rather than wine, may explain the paradox, at least in southern regions of  France.  Nevertheless,  the French Paradox remains central to the wine and the heart debate and a strong driver of research.. ([viii])   A 2019 article entitled “The French Paradox: Was it Really the Wine?” by the American Society of Nutrition commenced with “Here we go again. I’m reading the billionth newspaper article referring to the “French paradox” ([ix])  sums up the public fascination with Renaud’s idea. Thirty years after it was first formulated, and a decade after Renaud’s death, it continues to stimulate research to discover the mechanism of the paradox. ([x],[xi])

Yet another Paradox:  The J curve

While it has been difficult to get to the truth with the French Paradox, the J curve is an even more challenging and controversial paradox. In this Chapter, I will present the data on the relationship between alcohol and mortality, and in subsequent chapters discuss the relationship with cardiovascular mortality and later, the multiple attacks on it. We will carefully examine claims that the J curve may be an illusion and try to reach an objective conclusion to satisfy the sceptical critics as well those who enjoy an occasional glass of wine. The reality is that evidence has evolved since Renaud identified the French Paradox. Justifying light to moderate alcohol intake of alcohol solely on the French Paradox may rely on more than a hint of wishful thinking than solid evidence. The J Curve is yet another paradox, with an equally fascinating story.  Understanding it is crucial to the alcohol-health debate and it warrants detailed examination. I hope the curves I have chosen are self-explanatory.

The J curve refers to the observation that persons with low-moderate intake of alcohol appear to have a lower risk of death, particularly from heart attack, than persons who do not consume alcohol at all.  Instead of a straight line indicating a steady increase in risk with increased consumption, the curve dips downwards indicating reduced risk with low-moderate consumption, before heading upwards in a straight line, following the shape of a J (sometimes referred to as a U-shaped curve). ([i]) A typical depiction of the J Curve in a publication from 2017 ([ii])  is shown in Figure 2.1 below. The figure compares the relationship of smoking to mortality ([iii]) comparing the mortality of persons who consume alcohol compared with non-consumers. ([iv]) The strikingly different linear relationship with smoking and the J shaped curvilinear relationship with alcohol is clearly seen.

The apparent benefit of low-moderate intake of alcohol on mortality was first recognised in 1926 by Raymond Pearl. ([i]).

Facsimile from Alcohol and Longevity. published in 1926 by Raymond Pearl. Pearl was Professor of Biology at Johns Hopkins Medical School in Baltimore. He had a sound training in biostatistics, having studied in London with Karl Pearson, one of the founders of biostatistics and inventor of Pearson’s coefficient of correlation. Pearl dedicated his book to his friends in The Saturday Night Club, a group which included prolific essayist and satirist HL Mencken. They appear to have enjoyed liberal alcohol consumption during the Prohibition era.  The data presented and views expressed in Pearl’s book generated considerable debate and controversy.

Detailed descriptions of this unexpected relationship between alcohol and mortality began to appear in the late 1970’s. ([i],[ii])  The authors of the a 1977 study stated  “There was a strong negative association between moderate alcohol consumption (up to 60 ml per day).”

The most widely recognised early studies were both from 1981. Michael Marmot and colleagues in the UK described a “U shaped” curve on 10 years of follow-up in 1422 male British civil servants. ([iii]) Arthur Klatsky and his long-standing colleague Gary Friedman  in the US who described a J shaped curve in 10 year follow-up of 2015 persons in the US Kaiser Permanente health care system. ([iv]) Since then, multiple reports in different countries and populations have consistently confirmed the J-shaped relationship. Before we examine the J curve and the relationship with mortality from heart attack, we will look at the relationship between alcohol consumption and mortality from all causes. 

Arthur L. Klatsky, a graduate of Yale and Harvard, is a clinical cardiologist previously Chief of Cardiology and Adjunct Investigator at the Division of Research at the Kaiser Permanente Medical Center, Oakland, California. He has been researching the effects of alcohol since 1974 utilising the detailed health records of the Kaiser Permanente Health Maintenance Organisation.


Michael Marmot (now Professor Sir Michael) is one of the world’s most respected epidemiologists. Born in London but raised in Sydney, and studied medicine at Sydney University. He led the UK Whitehall studies of British Civil Servants. beginning in 1967 and has devoted his career to action on improving poor health caused by social inequities.

The J curve and all-cause mortality

Individual studies

There are too many studies  to report individually, but the  J-shaped curve identified by Klatsky and Marmot has been consistently found in populations as diverse as British civil servants, ([i]) physicians in the US, ([ii]) Britain, ([iii]) and Japan, ([iv]) women as well as men, ([v]) US female nurses, ([vi]) German construction workers,  ([vii]) US patients in a managed care population, ([viii]) an urban population in Denmark, ([ix]) young,  ([x]) middle aged ([xi],[xii]) and elderly, ([xiii]) and in populations  in Russia, ([xiv]) China, ([xv]) Japan, ([xvi]) Scotland, ([xvii]) France, ([xviii]) Italy, ([xix]) Poland, ([xx]) Australia. ([xxi])  

The usual methodology to obtain data to construct these curves is to follow up a cohort of subjects whose pattern of alcohol consumption has been recorded at baseline and check their mortality at follow-up years later. This is far more reliable than trying to draw conclusions from associations at the time of the survey.  An excellent example of a high quality cohort study was published in 2000 by Gaziano and colleagues. They followed 89,299 U.S. men from the Physicians’ Health Study enrolment cohort. (22) There were 3,216 deaths over 5.5 years of follow-up of male physicians who were 40 to 84 years old in 1982 and free of known myocardial infarction, stroke, cancer or liver disease at baseline.  The pattern of the curve again confirmed earlier observations that a low-moderate intake of alcohol is associated with a lower all-cause mortality.

A major study from US-wide national health surveys was published in 2017 by Xi and colleagues.  Impressive for the huge number of participants, they followed over one third of a million Americans and tackled the issue with advanced statistical techniques including sequentially adjusting for confounding factors. Their analysis is the most complete to date, and confirms previous studies. ([xxii])   The relationship from this relatively recent data  is shown in Figure 2.2. The J curve relationship at low-moderate levels of consumption is clear. The flattening of the curve at the upper end of consumption reflects the relatively small number of participants with hugely excessive weekly intakes, and this is confirmed with the widely spaced dotted lines, definig the statistical “confidence interval” of the estimate.

An even more recent study was reported in 2021 and also confirmed previous studies. A long term cohort follow-up of 142 960 individuals in 16 population cohorts (15 from Europe and 1 from Australia) was based on the long-running MONICA study.([i]) They used a detailed alcohol intake questionnaire at baseline  and took pains to identify life time abstainers (as distinct from ex-drinkers) as the reference group. In this well-done study with careful selection of the reference group, the J curve relationship was again confirmed with the most up to date analytical techniques available.


The role of meta-analysis.

The abundance of studies demonstrating the J curve is such that it is difficult to summarise the relationship between alcohol and mortality and arrive at the truth when a single study or selected studies are the only source of data.  So, medical scientists have resorted to the technique of meta-analysis. ([i]) This technique records data from multiple studies and calculates the overall risk of selected outcomes,  summarises the data and checks for consistency.  It can then estimate the average reduction of mortality from low-moderate intake of alcohol. There are many challenges to achieve a reliable, unbiased meta-analysis. Making sure that all relevant studies have been included, that poor quality studies are excluded and correcting for the duration of follow-up, are standard challenges with all meta-analyses. ([ii]) In studies of alcohol, there are unique challenges in standardising the level of  intake which may be reported in different measures in different reports and in trying to standardise alcohol intake from different beverage such as wine spirits or beer, and distorted by poor memory or intentional understatement. Despite these challenges, the results from the meta-analyses of alcohol intake and mortality are remarkably similar. When all studies are put together, the shape of the curve is affected to varying degrees by age, gender, inclusion of ex-drinkers in the reference group., and the duration of follow-up, but the general J shaped relationship is robustly retained. ([iii]) A particular challenge is in separating ex-drinkers from lifetime abstainers and we will examine the thorny issue of how this may distort the curve  in later chapters. Recent improvements in the techniques of meta-analysis based on published studies have made little difference to the conclusions about the shape of the J Curve. ([iv])

Meta-analyses based on published studies versus individual participants.

Standards for high quality meta-analyses have evolved over the years, and stricter standards now apply to studies published in first rank journals. ([v]) A key recommendation of these modern guidelines is that meta-analyses based on the individual data in the original studies rather than data collated from tables in the published reports are more likely to reflect the true relationship. Having access to the original data overcomes the impact on the results which may arise from publication bias, missing data or misinterpretation in published studies.

A recent meta-analysis in 2018 by Wood and colleagues is a prime example of an individual participant meta-analysis. It follows these stricter guidelines, and includes individual data from 83 studies on the relation between alcohol consumption and mortality, involving 599,912 subjects and 40,310 deaths. ([vi])  This meta-analysis, (Figure2.4)  using all the techniques required for a well-conducted meta-analysis and the data from individual participants in the studies included is the most stringent and authoritative analysis to date. This more complete study confirmed a flat relationship (i.e. no increase in risk) for low-moderate alcohol intake up to 10g or 10 standard drinks per week for all-cause mortality. It has been highly influential in identifying the threshold for increasing risk with increased alcohol intake. ([vii],[viii])  Of particular note,  this study focussed on current drinkers. Examination of the graph in Figure 2.4 shows no data point for zero consumption. and did not compare the risk with non-drinkers.  To date, there are no reports of similar quality to evaluate the complete span of alcohol consumption from zero to excessive.

Thus far in our search for the truth on the health effects of alcohol, low-moderate alcohol consumption appears to be consistent with a mortality benefit with mild to moderate consumption of alcohol. In coming chapters, I will report on the special case of the heart and cardiovascular system. the J curve looks like a solid and important piece of evidence

But lurking around the corner in this story is a storm of protest. Many believe the J Curve is too good to be true. How could a low to moderate intake of alcohol possibly be good when all are agreed that too much is obviously bad? Epidemiologists struggle with converting J shaped relationship to health policy. (13) New data on this controversy has appeared in recent years. It is difficult to see the truth, but let’s rely on the facts, not entrenched opinions.

With perspicacity still relevant a century later, Raymond Pearl in 1926 identified the limitations of his own observations and all others in this field.

“The trouble with all this indirect evidence on alcohol and mortality is that from its inherent character it cannot possibly prove anything, no matter how much it may be multiplied in amount. It merely suggests or implies what may be so”

In the next Chapter, we examine the special case of the relationship between alcohol consumption and deaths from heart attack and stroke. 


[i] Sacks HS, Berrier J, Reitman D, Ancona-Berk VA, Chalmers TC. Meta-analysis of randomized controlled trials. N Engl J Med 1987;316:450-5.

[ii] Moher D, Liberati A, Tetzlaff J, Altman DG; PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ. 2009 Jul 21;339:b2535.

[iii] Gmel G, Gutjahr E, Rehm J. How stable is the risk curve between alcohol and all-cause mortality and what factors influence the shape? A precision-weighted hierarchical meta-analysis. Eur J Epidemiol. 2003;18(7):631-42. 

[iv] Colpani V, Baena CP, Jaspers L, van Dijk GM, Farajzadegan Z, Dhana K, Tielemans MJ, Voortman T, Freak-Poli R, Veloso GGV, Chowdhury R, Kavousi M, Muka T, Franco OH. Lifestyle factors, cardiovascular disease and all-cause mortality in middle-aged and elderly women: a systematic review and meta-analysis. Eur J Epidemiol. 2018 Sep;33(9):831-845

[v]  Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JP, Clarke M, Devereaux PJ, Kleijnen J, Moher D. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: explanation and elaboration. Ann Intern Med. 2009 Aug 18;151(4):W65-94.

[vi] Wood AM, Kaptoge S, Butterworth AS, Willeit P, Warnakula S, Bolton T,  Emerging Risk Factors Collaboration/EPIC-CVD/UK Biobank Alcohol Study Group. Risk thresholds for alcohol consumption: combined analysis of individual-participant data for 599 912 current drinkers in 83 prospective studies. Lancet. 2018 Apr14;391(10129):1513-1523 

[vii] UK alcohol unit guidance: CMOs' Low Risk Drinking Guidelines.

[viii] Australian guidelines to reduce health risks from drinking alcohol.2020  accessed Jun 2022

[i] Di Castelnuovo A, Costanzo S, Bonaccio M, McElduff P, Linneberg A, Salomaa V, Männistö S, Moitry M, Ferrières J, Dallongeville J, Thorand B, Brenner H, Ferrario M, Veronesi G, Pettenuzzo E, Tamosiunas A, Njølstad I, Drygas W, Nikitin Y, Söderberg S, Kee F, Grassi G, Westermann D, Schrage B, Dabboura S, Zeller T, Kuulasmaa K, Blankenberg S, Donati MB, de Gaetano G, Iacoviello L. Alcohol intake and total mortality in 142 960 individuals from the MORGAM Project: a population-based study. Addiction. 2021 Jun 9. doi: 10.1111/add.15593. Epub ahead of print.

[i] Britton A, Marmot M. Different measures of alcohol consumption and risk of coronary heart disease and all-cause mortality: 11-year follow-up of the Whitehall II Cohort Study. Addiction. 2004 Jan;99(1):109-16.

[ii] Gaziano JM, Gaziano TA, Glynn RJ  et al.  Light-to-moderate alcohol consumption and mortality in the Physicians' Health Study enrollment cohort.  J Am Coll Cardiol 2000;3596- 105 

[iii] Doll R, Peto R, Boreham J, Sutherland I.  Mortality in relation to alcohol consumption: a prospective study among male British doctors.  Int J Epidemiol 2005;34199- 204

[iv] Kono S, Ikeda M, Tokudome S, Nishizumi M, Kuratsune M. Alcohol and mortality: a cohort study of male Japanese physicians.  Int J Epidemiol 1986;15527- 532

[v] Wang C, Xue H, Wang Q, Hao Y, Li D, Gu D, et al. Effect of drinking on all-cause mortality in women compared with men: a metaanalysis. J Womens Health 2014;23:373e81.

[vi] Fuchs CS, Stampfer MJ, Colditz  GA  et al.  Alcohol consumption and mortality among women.  N Engl J Med 1995;3321245- 1250 

[vii] Brenner H, Arndt V, Rothenbacher D, Schuberth S, Fraisse E, Fliedner TM. The association between alcohol consumption and all-cause mortality in a cohort of male employees in the German construction industry.  Int J Epidemiol 1997;2685- 91

[viii] Klatsky  AL, Armstrong  MA, Friedman  GD Alcohol and mortality.  Ann Intern Med 1992;117: 646- 654

[ix] Grønbaek M, Deis A, Sørensen TI, Becker U, Borch-Johnsen K, Müller C, Schnohr  P, Jensen G. Influence of sex, age, body mass index, and smoking on alcohol intake and mortality. BMJ. 1994 Jan 29;308(6924):302-6.

[x]Andreasson S, Romelsjo A, Allebeck P. Alcohol, social factors and mortality among young men.  Br J Addict 1991;86877- 887

[xi] Lin Y, Kikuchi S, Tamakoshi A  et al.  Alcohol consumption and mortality among middle-aged and elderly Japanese men and women.  Ann Epidemiol 2005;15590- 597

[xii] Tolstrup JS, Jensen MK, Tjonneland A, Overvad K, Gronbaek M Drinking pattern and mortality in middle-aged men and women.  Addiction 2004;99323- 330

[xiii] Simons LA, Friedlander Y, McCallum J, Simons J. Alcohol intake and survival in the elderly: a 77month follow-up in the Dubbo study.  Aust N Z J Med 1996;26662- 670 

[xiv] Deev A, Shestov D, Abernathy J, Kapustina A, Muhina N, Irving  S. Association of alcohol consumption to mortality in middle-aged U.S. and Russian men and women.  Ann Epidemiol 1998;8147- 153

[xv] Yuan JM, Ross RK, Gao YTG, Henderson BE, Yu MC. Follow up study of moderate alcohol intake and mortality among middle-aged men in Shanghai, China.  BMJ 1997;314-18 

[xvi] Tsugane S, Fahey MT, Sasaki S, Baba  S. JPHC Study Group, Alcohol consumption and all-cause and cancer mortality among middle-aged Japanese men: seven-year follow-up of the JPHC study cohort I.  Am J Epidemiol 1999;1501201- 1207 

[xvii] Hart CL, Smith GD, Hole DJ, Haethorne VM. Alcohol consumption and mortality from all causes, coronary heart disease, and stroke: results from a prospective cohort study of Scottish men with 21 years of follow up.  BMJ 1999;3181725- 1729  

[xviii] Renaud SC, Gueguen R, Siest G , Salamon R. Wine, beer, and mortality in middle-aged men from Eastern France.  Arch Intern Med 1999;1591865- 1870

[xix] Farchi G, Fidanza F, Mariotti S, Menotti A. Alcohol and mortality in the Italian rural cohorts of the Seven Countries Study.  Int J Epidemiol 1992;2174- 81

[xx] Waskiewicz A, Sygnowska E, Drygas W. Relationship between alcohol consumption and cardiovascular mortality: the Warsaw Pol-MONICA Project.  Kardiol Pol 2004;60552- 557. 

[xxi] Cullen KJ, Knuiman MW, Ward NJ Alcohol and mortality in Busselton, Western Australia.  Am J Epidemiol 1993;137:242- 248

[xxii] Xi B., Veeranki S.P., Zhao M., Ma C., Yan Y., and Mi J.: Relationship of alcohol consumption to all-cause, cardiovascular, and cancer-related mortality in U.S. adults. J Am Coll Cardiol 2017; 70: pp. 913-922

[i] Yano K, Rhoads GG, Kagan A. Coffee, alcohol and risk of coronary heart disease among Japanese men living in Hawaii. N Engl J Med. 1977 Aug 25;297(8):405-9.

[ii] Blackwelder WC, Yano K, Rhoads GG, Kagan A, Gordon T, Palesch Y. Alcohol and mortality: the Honolulu Heart Study. Am J Med 1980;68:164-9.

[iii] Marmot MG Rose G Shipley MJ Thomas BJ Alcohol and mortality: a U-shaped curve.  Lancet 1981;1580- 583

[iv] Klatsky A., Friedman G.D., Siegelaub A.B.: Alcohol and mortality: A ten-year Kaiser-Permanente experiment. Ann Intern Med 1981; 95: pp. 139-145.

[i] Pearl R.: Alcohol and longevity. New York, NY: Alfred A. Knopf Inc, 1926.

[i] Thompson PL. J-curve revisited: cardiovascular benefits of moderate alcohol use cannot be dismissed

Med J Aust 2013; 198 (8): 419-422 

[ii] Chokshi DA, El-Sayed AM, Stine NW. J-Shaped Curves and Public Health. JAMA.2015 Oct 6;314(13):1339-40.

[iii] Bjartveit  K, Tverdal  A.  Health consequences of smoking 1-4 cigarettes per day.  Tob Control. 2005;14(5):315-320.

[iv] Di Castelnuovo A, Costanzo S, Bagnardi V, Donati MB, Iacoviello L, de Gaetano  G. Alcohol dosing and total0 mortality in men and women: an updated meta-analysis  of 34 prospective studies. Arch Intern Med. 2006 Dec 11-25;166(22):2437-45

[i] McMichael J. French wine and death certificates. Lancet. 1979 Jun 2;1(8127):1186-7.

[ii] Ferrières J The French paradox: lessons for other countries Heart 2004;90:107-111.

[iii] Law M, Wald N. Why heart disease mortality is low in France: the time lag explanation. BMJ. 1999 May 29;318(7196):1471-6

[iv] Tunstall-Pedoe H, Kuulasmaa K, Mähönen M, Tolonen H, Ruokokoski E, Amouyel P. Contribution of trends in survival and coronary-event rates to changes in coronary heart disease mortality: 10-year results from 37 WHO MONICA project populations. Monitoring trends and determinants in cardiovascular disease. Lancet. 1999 May 8;353(9164):1547-57

[v] Nina Teicholz The Big Fat Surprise. Why Butter, Meat & Cheese Belong in a Healthy Diet. Simon and Shuster 2014

[vi] Hooper L, Martin N, Abdelhamid A, Davey Smith G. Reduction in saturated fat intake for cardiovascular disease. Cochrane Database of Systematic Reviews 2015, Issue 6. Art. No.: CD011737. DOI: 10.1002/14651858.CD011737

[vii] Estruch R, Ros E, Salas-Salvadó J, Covas MI, Corella D, Arós F, Gómez-Gracia E, Ruiz-Gutiérrez V, Fiol M, Lapetra J, Lamuela-Raventos RM, Serra-Majem L, Pintó X, Basora J, Muñoz MA, Sorlí JV, Martínez JA, Fitó M, Gea A, Hernán MA, Martínez-González MA; PREDIMED Study Investigators. Primary Prevention of Cardiovascular Disease with a Mediterranean Diet Supplemented with Extra-Virgin Olive Oil or Nuts. N Engl J Med. 2018 Jun 21;378(25):e34.

[viii] Ndlovu T, van Jaarsveld F, Caleb OJ. French and Mediterranean-style diets: Contradictions, misconceptions and scientific facts-A review. Food Res Int. 2019  Feb;116:840-858. 

[ix] Stefano Vendrame. The French Paradox: Was it Really the Wine?” American Society of Nutrition.  Accessed Sep 20th 2019 

[x]  Accessed Sep 20th 2019

[xi] Galinski CN, Zwicker JI, Kennedy DR. Revisiting the mechanistic basis of the French Paradox: Red wine inhibits the activity of protein disulfide isomerase in  vitro. Thromb Res. 2016 Jan;137:169-173. 

← Older Post Newer Post →

Leave a comment

Please note, comments must be approved before they are published