The French Paradox is based on the notion that the French eat a diet rich in butter and other fatty sauces, foie gras, fatty cheese, foods high in cholesterol and saturated fat, and have traditionally shunned low-fat products, i.e. theirs is a diet conventional wisdom holds to be bad, and yet the incidence of heart disease is very low compared to many other countries. Another way of describing the paradox is that it describes a lower than expected coronary heart disease (CHD) mortality in a country where the classic CHD risks are no less prevalent than in other industrialized countries and in addition where the diet has been historically high in saturated animal fat . A related issue is illustrated by the title of Mireille Guiliano's amusing and entertaining book, French Women Don't Get Fat . The French Paradox was brought to center stage for North Americans in 1991 when the TV program 60 Minutes examined the evidence for its existence by interviewing doctors, cardiovascular specialists and epidemiologists from both France and the U.S. The conclusion 60 Minutes put forward was that wine was responsible. The impact this had on the image and sales of wine, and especially red wine, is now part of the lore of the American Wine Industry. It was about a decade earlier that the phrase French Paradox started appearing in the medical literature and what some might consider the landmark paper  did not appear until 1992.
In this mini-review we will examine the various issues associated with the French Paradox. In particular, was or is the rate of heart disease really lower among the French, i.e. is there really a paradox? If so, is this due to diet, wine, lifestyle or all of the above? Or is it due to something else? Does the Mediterranean diet have anything to do with the paradox?
The lower coronary heart disease (CHD) mortality seen in France, as compared to other countries is a critical part of the French Paradox. In fact it has been suggested that French physicians underreport CHD mortality. In a just published paper, Jean Ferrières presents data indicating that this is not the case . A comparison of three measures, the official death certificate based CHD mortality rate, a CHD mortality rate that allowed more cases, and the coronary event rate, was carried out for the cities of Glasgow, Scotland and Belfast, Ireland vs. three cities in France. It was revealed that large and significant differences persisted as the criteria were relaxed to permit the counting of more cases. For example, the death certificate based mortality and the broader CHD mortality figures for men living in Glasgow vs. Toulouse, France were 332 and 365 vs. 53 and 91 per 100,000, respectively. Clower  presents data based on World Health Organization numbers showing CHD mortality rates for American vs. French men and women to differ by factors of 2.7 and 3.1, respectively. From other data, a comparison between France and Britain finds mortality ratios of 1 to 4 for men and 1 to 6 for women when the endpoint is coronary heart disease . Thus the first part of the French Paradox stands up to scrutiny. Consistent with the lower CHD mortality rate Balkau et al  found that the absolute death rates from heart attacks in diabetic men in British and American studies as compared to a French study were higher by factors of about 3 and 2, respectively. Also, there is a lower mortality rate from diabetes in France as compared to the U.S.  and a lower prevalence of long-term complications in type 2 diabetes as compared to other countries . Thus the French Paradox would appear to extend to individuals with diabetes.
The other essential aspect of the paradox involves characterizing the French diet as high in fat and cholesterol and in general a bad diet. In a study of European countries, the dietary habits of the French were similar to those who resided in Belfast, Ireland, where CHD mortality is 4-5 times higher . A comparison between France and Finland, Norway, Denmark, Germany and the UK reveals a similar consumption of total meat, beef and butter . In general, the consumption of fruit and fiber by the French is low whereas the intake of saturated fat is very high at 16% of total energy. One survey found the mean fat intake of French adults to be about 39% of total energy intake, mainly because of high consumption of butter . There has also been a "diet quality" study based on conventional ideas about a healthy diet. In a sample from southern France (Languedoc area) only 10 out of 146 subjects had a wholesome diet as judged by these standards. Also, even in the south of France, the dietary habits were not in line with the main characteristics of the traditional Mediterranean diet . In fact, while it is true that among European populations other than the French, only those with a consistent classical Mediterranean–type diet still have a low CHD mortality, the French do not for the most part eat a Mediterranean diet . Thus it does not appear, based on the conventional wisdom, that the French are protected from CHD by the nature of their diet, a diet most North American dietitians would rate as very bad indeed. Thus there is evidence that the dietary aspect of the paradox is true in the sense that the French have a diet that, based on comparison with other countries, would lead to the prediction of much higher rate of CHD mortality than is observed. But this is based on the conventional dietary wisdom (cynics call it dogma) concerning fat and saturated fat. How well this wisdom holds up to critical examination will be discussed at the end of this review.
An important question may not be how much fat the French eat but whether or not their diet in general results in being overweight or obese, conditions which would indirectly impact the risk of heart disease. Thus is the title of the above-mentioned book, French Women Don't Get Fat, based on fact or just a cute phrase to sell books? In their discussion of French eating habits, Paul Rozin et al provide the following data . The mean body mass index (BMI—weight in kilograms divided by the square of the height in meters) averaged across males and females is 24.4 (not overweight) for French adults and 26.6 (overweight) for American adults. Perhaps what is more important, 22.3 % of Americans are obese (BMI = 30) but only 7.4% of French qualify. These were numbers for 2002. Similar comparative figures for obesity of 30% and increasing vs. 8% and holding (!) are quoted in Will Clower's 2003 book The Fat Fallacy . Clower is a neurophysiologist at the University of Pittsburgh who spend two years in France as a research fellow at the Institute of Cognitive Sciences in Lyon, which gave him an opportunity to observe the French dietary habits and lifestyle. Clower's book is interesting in that it quotes what might be described as case histories where individuals who have come from North America to live for a few months to a few years in France generally lose weight while having in fact decided to "throw caution to the wind" and eat the typical French diet. Evidently, there was already an impression that the French diet was intrinsically bad. The same theme runs through Mireille Guiliano's book, including her own personal experience upon returning to France from the U.S. as an overweight teenager, and achieving normal weight by adopting the eating habits of France. Thus, in trying to explain the French Paradox, one of the important questions seems to be, why does the French diet not result in the expected frequency of overweight and obese individuals.
An interesting paper by Rozin et al  addresses this issue. The authors take the position that the reason the French are thinner than the Americans is that they eat less while spending a longer time eating. While this might seem somewhat simplistic, the authors go to some length to justify this view. Compared to the U.S., they document that portion sizes are smaller in French restaurants, the sizes of individual portions of items sold in supermarkets are smaller, the portions specified in French cook books are smaller, and the prevalence of all- you-can-eat restaurants is much lower in France than in the U.S. Both Clower and Guiliano also characterize the French diet as having smaller portions than those common in America. Also, direct observation in restaurants in the U.S. and France revealed that the French spend significantly longer eating a meal. Rosin et al also discuss studies showing that there is a delay before the brain signals that enough has been eaten, and slow eating allows this signal to be sent before overeating has occurred. Thus smaller portions are satisfying. A commonly made observation is that the French put their fork down between each bite, and while this is no doubt to some extent an exaggeration, it is consistent with the premium placed on savoring every bite. Lively conversation around the table and a glass or two of wine would also naturally encourage leisurely eating and more lengthy meals. Probably any North American who has lived in France for any length of time would confirm these observations. Exercise may also help explain the lack of overweight or obese individuals. As Guiliano points out, the French tend to walk more than the Americans as they go about their daily activities, walk to stores and frequently use stairs rather than elevators.
There are also significant differences between the French and the Americans in attitudes regarding food. In a study published in 1999, Rosin et al  examine this question. The categories compared included (a) Worry—the extent of worry as opposed to savoring food and also worry about weight gain; (b) Diet-Health Link—concern about the impact of food on health; (c) Pleasure and Importance—the pleasure of eating and the importance in life of food; (d) Culinary Associations—culinary as opposed to nutritional association of food; (e) Healthy Eater— Self perception as a healthy eater. Scores were assigned on the basis of a questionnaire. In the Worry and Diet-Health Link categories, the French exhibited much lower scores than the Americans. The reverse was true for Pleasure—Importance and Culinary Associations. Also, the French had a much higher score in the category of self-perception as a healthy eater. Thus as compared to Americans, the French are not as concerned about weight gain or eating what others perceive as unhealthy foods, but for them, eating is associated with great pleasure and this pleasure transcends nutritional considerations. Finally, the French consider themselves healthy eaters to a much greater extent than Americans. Rozin et al regard these observations as translating into differences in food-related or meal-time stress, and that this type of stress, which seems quite subtle, could account for part of the French Paradox since stress has negative health connotations, especially in the context of coronary heart disease. In other words, when comparing the French to the Americans, there are profound differences in psychological attitudes toward both food and its consumption that may have a bearing on health. These results probably underestimate the differences in lifestyle between France and North America. Hasty meals have become the norm in North America as individuals rush back to work after lunch or eat dinner almost on the run or at almost random times in order to meet other obligations including those associated with children's activities, which generally require parental taxi service. Couple this with two working parents and you have the ingredients for stress. Changes in the peaceful, leisurely way of life in France where one of the principal pleasures is associated with food may be occurring, but the French Paradox is not a recent phenomenon and CHD takes a while to develop.
Thus possible explanations for the paradox include lifestyle factors based on length of time spent eating, the pure pleasure and fascination associated with food, the disregard for the connection between what is eaten and health, and apparent calorie restriction due to smaller portions. This brings us to the question of alcohol and in particular wine consumed with meals and as well the level of wine and alcohol consumption in France compared to other countries. Let's look at the hypothesis, so dear to the wine industry, that the French Paradox is all about wine.
Two aspects characterize French wine drinking—quantity and regularity. De Lorgeril et al describe data from 1989 which found a mean consumption of alcohol of about 30 grams/day for men and only 10 grams/day for women . One glass of wine contains 12-15 grams of alcohol. Another study found roughly the same results expressed as percentage of energy intake from alcohol. For men it was 8% and for women 3.5%. Earlier studies based on grams of wine (not alcohol!) per person per day found France at 195, Italy at 144, Spain at 129, Greece at 84 and the U.K. at 33, obviously huge differences . These are all population-based estimates and thus will encompass wide ranges. As regards drinking patterns, the French tend to drink regularly and with meals rather than concentrate their drinking during weekends or simply binge drink. Drinking patterns appear to be an important issue. In a recent study based in New York, it was found that daily drinkers had a lower risk of heart attack than abstainers, drinking only on weekends increased rather than decreased the risk and those who drank with meals had lower risk than those who drank mainly without food . Similar results were reported recently in the New England Journal of Medicine  where drinking four or five to seven times a week revealed greater risk reduction than drinking once a week. Finally there is the observation made several times in the literature that heart attacks are more common early in the week. A study published in the journal Hypertension examined blood pressure the fluctuations throughout the week in subjects from France and Northern Ireland . In Northern Ireland, 66% of the total alcohol consumption occurs on Friday and Saturday, whereas in France it is typically spread throughout the week. Blood pressures for the Northern Ireland group peaked on Saturday whereas those of the French showed much less fluctuation with merely a small increase on the weekend. The authors concluded that the fluctuations observed in Northern Ireland could explain the higher incidence of heart attacks on Mondays in countries characterized by high alcohol intake on weekends. An alternative explanation is that many people hate their jobs and thus stress peaks on Monday. Nevertheless, the above results underscore the benefits associated with the French approach to alcohol consumption.
The subject of the association between alcoholic beverages and CHD is too large to examine in this review. However, it is natural to look for a French-based study since a protective effect should be significant if wine is part of the paradox. There appears to be only one prospective (follow-up) study, published in 1999 . For wine as a protective factor in CHD mortality, 22-32 grams of alcohol a day roughly halved the risk and resulted in a 33% reduction of all-cause mortality, but as is frequently found, the benefits disappear with heavier consumption. When these results are compared with a very recent combined analysis of 34 prospective studies , it appears that the risk reduction in the French study was larger. While the preponderance of evidence seems to favor the protective hypothesis, critics point to the possibility of confounding. The position of the doubters was strengthened recently with the publication of a study based on telephone interviews with over 250,000 U.S. adults. Data was gathered as to prevalence of cardiovascular (CVD) risks factors among nondrinkers and moderate drinkers. Of the 30 CVD-related risk factors assessed, 90% were significantly more prevalent among nondrinkers. The authors comment that these findings suggest that some or all of the apparent protective effect of moderate drinking may be due to residual or unmeasured confounding . Thus when the studies showing benefit are summarized to indicate that one to three drinks a day are associated with about a 20% reduction of CHD risk , this perhaps is an overestimate. It appears that the connection between wine and the French Paradox, while still a hypothesis, is to some extent evidence based. Arguments regarding biological plausibility, while interesting, strengthen but do not prove the point. To settle the alcohol and wine question would require randomized trials which would almost certainly never get past ethics committees. However, if there is a protective effect, then studies are clear on one point, the French are maximizing the benefit by drinking for the most part in adequate but moderate amounts, with meals, and they spread the intake more or less evenly over the week.
Finally, there is the matter of fat. If fats and especially saturated fats are in fact not very dangerous or even neutral in the context of heart disease, then this weakens considerably one of the two pillars on which the French Paradox rests. The French Paradox had its origin at the height of the anti-fat era when many disorders, but especially cancer and heart disease were attributed to fat consumption. Thus it was easy to point a finger at the French diet and pronounce it heart-unfriendly simply because of the high content of fat, and especially saturated fat. In other parts of the world, and especially in North America, low-fat food and low-fat diets became the rage and fat calories were replaced by large amounts of refined carbohydrates. Low-fat diets had a tendency to reduce HDL cholesterol levels and elevate triglycerides levels, two trends not viewed as beneficial to heart health. Also, high intakes of refined carbohydrates increased the risk of developing insulin resistance and eventually type 2 diabetes, neither of which is favorable to heart health. The French did not buy into the low-fat revolution or low-fat foods, but held steady on the traditional course of a high fat diet. Part of the reason may simply be the attitudes about food and health discussed above. Space does not permit examining the heart disease—dietary fat issue in any detail (see the Archives for an extensive discussion of this subject in the research review titled Dietary Fat and Heart Disease, Is There a Connection? The review appeared in the November-December 2002 issue of International Health News and still seems current today). However, the following brief discussion appears relevant to the French Paradox question.
By the beginning of the new millennium, the view that saturated fat was bad in the context of CHD was firmly established as a dogma and few dared to raise questions that would embarrass the establishment. In 2001 Gary Taubes ruffled some feathers with an article in the journal Science titled "The Soft Science of Dietary Fat," in which he raised serious questions about the evidence for the fat/heart connection and the merits of low-fat diets  (available free—Google "Gary Taubes fat"). The establishment's reply to this "shot over the bow" was in a letter to Science  which was immediately criticized in the same journal . It was pointed out that the two reviews cited as refuting Taubes' position either had no references at all or had references that for the most part did not support the defense. While Taubes' Science article received considerable media attention, a much more high-profile article was published a year later in the New York Times Magazine  with the provocative title "What If It's All Been A Big Fat Lie?", which again raised serious question about the low-fat dogma and low-fat diets. This article contained quotes from respected experts (it is also free via the same Google search).
In connection with the French Paradox there are two issues: (a) the CHD danger of high fat diets; and especially (b) the CHD dangers of saturated fat. Already in 1998, Ravnskov published an extensive review of this question in the Journal of Clinical Epidemiology . He was not able to find conclusive evidence to justify the claim that saturated fats were bad. About the same time, Hu et al from Harvard published a paper on dietary fat intake and the risk of CHD in women which was based on a large prospective study . If one looks at their Table 3 in which results adjusted for confounding are presented, saturated fat intake up to almost 19% (quintile median) of total energy was found to not be statistically associated with CHD risk (heart attack or fatal CHD). Nor was a high intake of total fat significant, but a significant inverse association (beneficial) was found for polyunsaturated fat, and enhanced risk was seen for trans fat. No association was found with cholesterol intake up to 275 mg/day (quintile median). Hu et al point out that the results of prospective epidemiologic studies of fat and CHD have been inconsistent, with 2 finding a positive association and 6 finding no association. Saturated fat increases LDL levels, but it also increases HDL, the so-called good cholesterol. Obviously there is the potential for these two effects to compensate. Yet the increase in LDL levels caused by saturated fat is at the heart of the establishment case against this particular fat. Saturated fat, incidentally, is a mixture of fats, most of which do not raise LDL . The study by Hu et al has recently been updated with a total of 20 years follow-up . The conclusion was the same—"Intakes of total fat, saturated fat and monosaturated fat had no clear relation to CHD regardless of age."
Three recent studies are relevant. A study from Denmark reported in 2004 reached the same conclusion, this time for both men and women. No statistically significant associations were found for either men or women between the intake of saturated or total fat and CHD . This was in spite of not adjusting for confounding to the extent done by the Harvard researchers. The second study is the now famous dietary modification trial which was part of the Women's Health Initiative (WHI). Dietary intervention that reduced total fat intake and increased the intake of vegetables, fruits and grains did not significantly reduce the risk of CHD . Finally, in a multicenter study that included Harvard, it was reported in 2004 that for postmenopausal women with relatively low fat intake, a greater saturated fat intake was associated with less progression of coronary atherosclerosis . This result may not be universally applicable since many of the subjects had CHD, were hypertensive or had diabetes, but it is in the wrong direction if one believes the conventional wisdom and the study included women at high risk but free of CHD.
The current recommendation by the National Cholesterol Education Program for men and women at high risk for CHD is that the intake of saturated fat should be reduced to 7% of total energy intake. The American Heart Association (AHA), in its just-published 2007 guidelines for CHD prevention in women, calls for limiting saturated fat intake to 10% and if possible 7% . One might ask, where are the statistically significant data to back up these recommendations? Lots of interested individuals have looked for the data, but don't seem to be able to find it ! In the AHA guidelines, they appear to give only one reference directly related to saturated fat and women, a small British study of 2002 . It found no relationship for men, but found risk for women, although the data were not adjusted for confounding to the extent done in the much larger and longer Harvard study mentioned above, which found no connection between CHD risk and fat or saturated fat for women [21,22]. Aside from the WHI study mentioned above, the AHA document does not include in its bibliography any of the negative studies concerning saturated fat and CHD, and in particular the Harvard study, and the guidelines obviously ignore all of these results. Also, this AHA guideline is for women, and yet two of the studies cited concerning diet and CHD risk were done only on men and the British study of both men and women cited found no connection for CHD and saturated fat in men anyway! Thus important studies, including a highly relevant one from the premier center of nutritional epidemiology in the world, are ignored and more or less irrelevant ones cited. The guidelines are titled "evidence based."
Thus the French Paradox appears to be based on an exaggerated and probably even totally wrong view of the importance of dietary fat in general and saturated fat in particular in connection with the risk of CHD. Therefore the paradox may in fact be a fantasy. However, the mystery of the low CHD mortality would remain, and the possible explanations advanced above would still be relevant since they are not dependent on the fat hypothesis. Some may find the above discussion contains sufficient reasons to imitate selected features of the French lifestyle. For those for whom this notion has appeal, the two books cited above will provide valuable and detailed guidance.