Saturated Fat: Friend, Foe or Neutral?
by William R. Ware, Ph.D.Emeritus Professor of Chemistry, University of Western Ontario
The centerpiece of modern dietary recommendations concerning specific macronutrients is to decrease saturated fat intake, generally as a percentage of total energy intake. This recommendation appears in the context of both cardiovascular disease and diabetes, and is sometimes qualified by the recommendation that the caloric deficit be made up by unsaturated fats, but substitution with carbohydrates appears common. The central place this piece of advice occupies in mainstream medicine provides motivation for examining the actual evidence that saturated fatty acid intake per se presents a health risk, especially since the hypothesis that saturated fat is dangerous has evolved to the status of a widely accepted and unquestioned dogma. Furthermore, saturated fat intake becomes an issue when carbohydrate restriction is being used in connection with diabetes prevention or control, weight loss, atherogenic blood lipid profile modification etc. Carbohydrate restriction, which appears to be undergoing a rebirth, will be discussed in an upcoming Research Report.
Diet is a complex subject to study in humans, either in intervention trials or observational studies. There are only three macronutrients, fat, carbohydrate and protein, but macronutrients are mixtures. Fats can be saturated, monounsaturated, or polyunsaturated, and the latter can be further broken down into subclasses which include the so-called omega-3 and omega-6 types. Fats can also be natural or an industrial product, e.g. trans-fats. Carbohydrates vary markedly in their ability to elevate blood sugar and insulin levels and as well contain a wide range of fiber. If intervention studies reduce the amount of one class of macronutrient, the calorie intake will decrease unless the energy intake is topped back up with one or the other of two macronutrients. Thus there is the inherent problem of changing two or more variables at once while trying to study only one. The same problem exists when one type of fat is substituted for another. Also, if energy intake decreases, weight loss may occur which will confuse the issue since there may be a loss of body fat or a change in its distribution which impacts fat and carbohydrate metabolism as well as inflammation. These problems are fundamental to nutritional studies, but they appear to lead to the attitude that studies must still be done even if the interpretation is almost always going to be debatable. Armies of biostatisticians stand ready to help sort out the confusion. Probabilities are the real end result and in some cases, nonsense.
Observational studies are plagued by problems associated with measuring the intake of macronutrients, breaking down the results by subcategories, and correcting for total energy intake. Dietary habits change after baseline assessment and within the intervals between repeated assessments. Over long periods, the very nature of popular foods changes. Witness the low-fat revolution. Evidence-based medicine is happy only with randomized intervention trials and many of these have the same problems as any observational trial. Randomized trials may also not really recruit a realistic sample of the general population and in fact can involve significant bias. Individuals who participate in trials may be different, and these differences may be hard to account for in statistical adjustment of the data. Furthermore, large randomized intervention trials lasting several years or more suffer from poor compliance, and even the ability of the investigators to detect this problem is limited. Biostatistics play a central role in nutritional studies as they do in most medically related studies, and there is the unending quest for statistical significance no matter how small or possibly clinically insignificant an effect may be. Studies lead to publication which leads to progression up the academic ladder or reprints useful as hand-outs to physicians.
In addition, there is the problem of energy intake. To keep the energy intake from fat constant in studies where, for example, saturated fat is decreased, requires substitution with another fat. But this of necessity means that two fat variables are being changed at once. If benefit is seen it does not prove that the macronutrient increased was beneficial and/or the macronutrient decreased was detrimental. But this is frequently the impression some investigators attempt to establish, partly by demonizing the item decreased even if there is no or only very weak evidence that it is harmful.
To the above can be added the multitude of potential subject characterizes. One can study the overweight, the obese, the lean, hypertensives, premenopausal women, postmenopausal women, individuals with diabetes or prediabetes, patients who have heart disease, individuals with perceived elevated risk of heart disease, men only, women only, children, adolescents, young people, middle-aged people, old people, those who are physically active and those that are not, etc, etc. It is wishful thinking to assume that these characteristics do not really matter and in fact ignoring them is basic to studies that look at various endpoints based on the national consumption of, for example, fat derived from government statistics. Also, studies tend to focus on the impact of diet on one disorder, i.e. heart disease or diabetes, or a surrogate marker such as the blood lipid profile and in particular LDL cholesterol, or insulin resistance or the dyslipidemia associated with the metabolic syndrome. Issues then arise concerning the clinical relevance of small effects which, because they are statistically significant, permit a quantitative measure of risk reduction or elevation. There is also the tendency to accept as important and indicative of meaningful conclusions, numerical correlation coefficients which are so small that the associated scatter plots show no visual correlation at all. In some cases, while there may be a slight suggestion of correlation in a scatter plot, individuals trained in the physical sciences would laugh at any significance being attached to such correlations, given the huge scatter about the line the computer constructed through the points. In older literature, it was common to present such plots with the line, but not give the actual correlation coefficient, presumably because it was so small that any conclusion was wishful thinking. Many of these older studies are today the basis of the evolution of a mere hypothesis to a universal dogma taught to students in nutrition 101, the fat-heart disease connection.
There is also the fundamental problem that people eat meals that are best described in terms of food patterns, not isolated macronutrients and that the characterises of the mix of macronutrient content, e.g. types of fat and carbohydrate, vary from day to day. In addition, there is the matter of how fast the meal is eaten and the social context of the meal (gulp it down and run or spend 1-2 hours in pleasant social interaction with a glass or two of wine). These are issues that are frequently ignored and are in fact hard to measure and quantify.
If one wishes to address the question of the merits of carbohydrate restriction in the context of heart disease, diabetes and the metabolic syndrome, then the question of fat intake and the types of fat become a central issue. Papers regarding the debate concerning the merits of carbohydrate restriction are now appearing with increasing frequency; perhaps because the guidelines that focus on reducing fat in connection with heart disease or diabetes are now starting to be recognized as simply leading to pharmaceutical drug interventions, i.e. they do not work. An added complication in the carbohydrate vs. fat debate is that the terms low-fat diet and low-carbohydrate diet are poorly defined and in fact there is considerable overlap, misunderstanding and confusion. Carbohydrate restriction is a rather loose term and discussions must clarify exactly the extent of the restriction and the types of carbohydrate being restricted. The reader is referred to the research report Carbohydrate Restriction.
SATURATED FAT AND CARDIOVASCULAR DISEASE
Saturated fatty acids have no carbon-carbon double bonds and are thus fully saturated. Mono- or polyunsaturated fats have carbon-carbon double bonds. Examples of foods containing a high proportion of saturated fat include cream, cheese, butter, animal fats such as suet, tallow, lard and fatty meat, coconut oil, cottonseed oil, chocolate and some prepared foods. Eggs also contain significant amounts of saturated fat. During the last third of the 20th century, these foods occupied a unique position by being proscribed by governments, nutritionists, and mainstream medicine as being at the root of many of our health problems. Older generations who spread liberal quantities of lard on their bread were regarded as foolish in spite of the fact that coronary heart disease was not common for their generation.
There are a number of saturated fatty acids, the principal ones being lauric, myristic, palmitic and stearic acid, and these have in some cases quite different biological actions which complicates the interpretation of studies that lump all saturated fatty acids together when attempting to study their impact on health. This actually is quite important because there are some curious balancing effects associated with certain endpoints which yield negligible net risk associated with saturated fat intake. But ignoring this is operationally useful—the grease that keeps the nutritional study machine running smoothly.
At issue in this review are the alleged benefits and dangers associated with the dietary intake of saturated fat. In this context, one very important point must be established early on. This involves the correlation between dietary SFA intake and the SFA levels found in the fat carrying blood plasma components. It is quite common to see the comment that dietary intakes are difficult to measure and it is better to use plasma levels when investigating questions such as insulin resistance or the risk of CVD. In a review published in 2008, Volek et al1 point out that for SFAs, the statement that plasma fatty acid content reflects dietary intake is not true. They cite ecologic studies where total fat intake varied from 56.7% to 27.2% with virtually no variation in plasma SFAs. In another type of study, dietary SFA intake was decreased by 50% at constant total fat with no change in the plasma stearic or palmitic acid content. Furthermore, in a widely quoted study which according to the title showed that total fat intake modifies plasma fatty acid composition, the changes in plasma SFAs were insignificant. Similar results were reported by Qi et al 2 based on data from the Nurses' Health Study where no significant association was obtained between plasma or erythrocyte (red blood cell) FA content and SFA intake, even stratified by fatty acid type. Thus if one is concerned with the question of the risks or benefits of the dietary intake of saturated fat, studies that use plasma levels are not relevant and in fact relate instead to complex metabolic questions and do not indicate the increase or decrease of saturated fat intake.
In a lengthy and comprehensive review which appeared in 1998 in the Journal of Clinical Epidemiology, Ravnskov examined the evidence for and against the hypothesis that saturated fat intake was related to CVD incidence or mortality.3 A very large number of ecological (population) studies, as well as cross-sectional (snapshot), case-control and cohort follow-up studies were examined. In addition autopsy studies that examined the relationship between saturated fat (animal fat) intake and the extent of atherosclerosis were reviewed. The overall picture that emerged was one of inconsistency with more studies falsifying the hypothesis than supporting it, even when studies of comparable quality were compared in detail. The author concluded that the correlation found by Keys which launched the hypotheses suffered from selection bias (selecting only studies that agreed with the hypothesis from among a much larger number where no correlation existed), a point repeatedly made in the literature, and was not supported by his later Seven Countries study nor by more recent ecological studies. Ravnskov also points out that in fact, there is no real support at all since a hypothesis is unlikely to be correct when a very large number of studies falsify it. Stated simply, according the heart-diet idea, the intake of SFAs promotes CHD and therefore studies that find no difference between the intake of SFAs and CHD in patients and CHD-free individuals are obviously contradictory. It is important to recognize that hypotheses are made to be falsified, rather than proved. Science progresses by falsifying hypotheses and finding better ones. Scientific progress is hindered, on the other hand, by converting hypotheses into dogma even when they have in fact been falsified. Examples of this go back to ancient times. It is also important to recognize that when a large number of studies directed at a single question give opposing answers, i.e. gross inconsistency, there is a problem with the hypothesis that generated the studies and a high likelihood that bias and confounding are at work to produce the lack of concordance. Ravnskov also does not find convincing evidence up to 1998 that polyunsaturated fatty acids (PUFAs) were beneficial in the context of this review.
In the "Dissent" published in the same issue4, Golomb actually more or less agreed with Ravnskov. She concluded that the evidence for a beneficial effect of SFA reduction and PUFA augmentation is unconvincing. This was the main conclusion of Ravskov's review.
By 2005, there had been little change in the evidence picture, but the dogma was alive and well. Then during that year, Volek and Forsythe5 published a paper, the purpose of which was to make the case for not restricting saturated fat in a low-carbohydrate diets. They made 4 points with documentation.
The authors apparently considered that this was clear enough so that no additional comments were necessary. However, in another paper, Volek et al1 present a scatter plot from a frequently quoted paper that is used to demonstrate that saturated fat intake increases cholesterol and therefore represents a CVD risk. If one examines just the part of the plot ranging from 5% to 25% saturated fat intake as a percentage of total energy, a range that encompasses the majority of North Americans who on average consume only 14% of energy as SFA, the correlation is almost impossible to see, and if one looks at the range from 15% to 7%, the change recommended by the current guidelines, the scatter is so great that this decrease in SFA intake appears equally likely to raise as to lower total cholesterol. To quote Volek et al, "…the idea that SFA is inherently and unambiguously detrimental becomes fundamentally untenable and dietary recommendations for across-the-board reduction, fundamentally inappropriate." It thus becomes a source of constant amazement to see the exact opposite stated repeatedly in the literature as if it were on a par with the fact that earth goes around the sun.
Gary Taubes in his article on fat and heart disease in the journal Science6 also makes a point of discussing the various SFAs in our diet and concludes that from the CVD risk point of view, eating a nice juicy sirloin steak is a "wash" simply because of minor adverse effects if they exist, from some of the SFAs, are balanced by minor beneficial effects of other SFAs. Again, the issue disappears if one admits that LDL does not drive atherosclerosis.
In 2005 a 20-year update on the famous Nurses' Health Study also looked at the association between dietary fat intake and CHD.7 The abstract is interesting because it contains no mention of saturated fat, an omission which might strike the casual reader as odd given the almost universal recommendation by mainstream medicine, the American Heart Association, the National Cholesterol Education Program panel and most of the nutrition community to avoid this dangerous fat. The reason for the omission is evident upon examination of the tabulated results. When corrected for confounding, there was no significant relation between the relative risk of coronary heart disease and saturated fat intake when the lowest vs. the highest quintile were compared. The same in fact applied to total fat intake, but polyunsaturated fat was found to be significantly protective and trans-fats significantly harmful. This study involved the analysis of data from a lengthy follow-up study of over 78,000 female nurses.
This subject was again reviewed in 2008 by Accurso et al8 in a paper calling for a critical appraisal of dietary carbohydrate restriction in individuals with type 2 diabetes and/or the metabolic syndrome. In connection with saturated fat intake, they comment on the inconsistent results and cite several critical reviews that have pointed to the general failure to meet the kind of unambiguous outcomes that would justify the blanket condemnation of saturated fat per se. They note that during the current obesity and diabetes epidemic, the proportion of dietary saturated fat decreased and for men the decrease was 14%. They also point to the now famous result of the Woman's Health Initiative study which found no difference in CVD incidence for those who consumed < 10% saturated fat compared or those whose consumption was > 14% of total energy intake. They also comment on the point raised above that increased saturated fat consumption decreases the small, dense LDL, the LDL though to be atherogenic. Finally, they also mention that a greater intake in saturated fat reduced the progression of coronary atherosclerosis and greater carbohydrate intake increased the rate of progression. For the replacement of fat with carbohydrates, they point out that the result is almost always harmful.
The above discussion is not meant to suggest that this is other than a complex issue. It is true that some saturated fatty acids increase LDL, some decrease it. The same applies to HDL, but HDL is much more strongly related to CHD event risks. But as discussed above, saturated fat intake also dramatically modifies the atherogenic nature of LDL in a favourable direction. In addition, we do not eat individual saturated fatty acids in isolation, we eat mixtures, along with mono- and polyunsaturated fats, protein and carbohydrate and this mixture results in interactions. Finally, if one just focuses on the LDL elevating properties of some saturated fats, then there is still the little problem that LDL levels are totally uncorrelated with the calcified plaque load, the extent of atherosclerosis seen at autopsy or the total plaque load of the coronary arteries as seen by modern contrast enhanced CT scans. This was discussed in last month's Newsletter. And if one believes that atherosclerosis comes before symptomatic heart disease, then this appears to provide a good reason to ignore LDL and look elsewhere for why we get heart disease. This incentive to look elsewhere is reinforced by the recent study showing that over 50% of individuals admitted to hospital for CHD in general and heart attack in particular have low to very low LDL levels.9
Thus it seems that when skeptics and simply the curious or cautious students of this subject look for convincing or significant evidence behind the guidelines recommending limiting saturated fat, it is not out there, at least in the context of CVD.
FATTY ACIDS AND INSULIN SENSITIVITY
The reported connection between saturated fat and insulin sensitivity is presumably part of the justification for diabetes guidelines recommending limiting this nutrient. Risérus10 and Risérus, Willett and Hu11 have reviewed the impact of saturated fat on insulin resistance and also the risk of type 2 diabetes. In six short-term randomized controlled dietary intervention studies reported between 1995 and 2002, the overall results showed no significant association between insulin sensitivity and the intake of saturated fatty acids (SFAs), monounsaturated fatty acids (MUFA) or polyunsaturated fat (PUFA). Four of the studies used mainly healthy subjects whereas in two the participants were type 2 diabetics or individuals with impaired glucose tolerance.
Other randomized intervention trials have been somewhat inconsistent. The largest randomized trial, the so- called KANWU study, examined the effect of elevating SF or MUFA on insulin sensitivity in healthy men and women.12. Participants were randomized into two groups. One was given a diet with a high proportion of SF, the other a diet high in MUFA. Each group was also split randomly into two groups, one receiving fish oil containing 3.6 g per day of omega-3 fatty acids (n-3 FA), the other a placebo. There was no difference in insulin sensitivity observed related to n-3 FA intake with either diet. The SF diet reduced insulin sensitivity by about 10% whereas the MUFA diet had no significant effect. When the changes for the two diets were compared, the difference was only of borderline statistical significance by one measure, and insignificant by another. A second important finding was that subjects with the highest total fat intake with either diet did not show any difference in insulin resistance at all when the two diets were compared. This study is frequently held up as convincing evidence that saturated fat decreases insulin sensitivity and is therefore bad. In fact, the evidence is somewhere between very weak and non-existent.
Finally, these reviews make clear that there seems to be no evidence to suggest that neither increasing n-3 FAs nor decreasing the n-6/n-3 ratio improve insulin sensitivity. There is however some data from a study by Summers et al 13 which suggests that substituting SF with n-6 PUFA increases insulin sensitivity. For individuals with diabetes, few data are available on the effects of dietary fat quality and the optimal proportions of SF, MUFA and PUFA are uncertain. Studies that substitute PUFAs for SFs have also been observed to alter the amount of visceral fat and this complicates the interpretation. Whether or not trans-fatty acids impair insulin sensitivity remains an open question, although one trans form of conjugated linoleic acid was found to impair insulin sensitivity even an intake representing 1% of total energy. Since the evidence for the recommendation to avoid trans fats is fairly strong, widely recognized and even the subject of legislation, the danger associated with this type of fat does not seem to any longer be an issue. Its appeal should have always been limited, given that it came out of chemical factories and represented a class of fat almost entirely foreign to human biochemistry.
In spite of what appears to be very weak evidence, both reviews cited above conclude that substituting SFAs and trans-fatty acids (TFA) with MUFA or PUFA has a beneficial effect on insulin sensitivity. However, it would seem that if the benefit is real, it is very small since it did not turn up in a number of randomized trials. This conclusion also seems to rely heavily on the KANWU study12 and the n-6 PUFA study by summers13 and the adverse effect of on isomer of conjugated linoleic acid. The first two were at best of borderline statistical significance and the relevance of the latter is not obvious. Furthermore, most would agree that substituting trans- FAs with unsaturated FAs is a good idea, but why bunch TFAs and SFAs together in a concluding statement?
SATURATED FAT AND DIABETES RISK—EPIDEMIOLOGIC STUDIES
This question was addressed in both the Nurses' Health study14 and the Health Professionals Follow-up Study15, the latter restricted to men. Both studies assessed dietary fat intake and in the follow-up, determined the relative risk of diabetes using low intake as a reference. In both studies, when the data was adjusted for potential confounders, there was no significant relationship between dietary saturated fat intake and the risk of type 2 diabetes. Both studies ran for about 12 years prior to the reporting of the above results.
Attempts to establish a connection between saturated fat and diabetes frequently depend on serum markers of intake rather than directly measured intake, and as mentioned at the beginning of this review, for saturated fat this is not a reliable or even meaningful approach. Thus these studies will not be considered even though they appear to influence opinion on the question. Risérus, Willett and Hu11 reviewed epidemiologic studies that relate to dietary fats, insulin resistance and the risk of type 2 diabetes, but with respect to saturated fats, it seems only studies based on actual intake measurements merit consideration. Aside from the follow-up studies, they discuss cross sectional studies which are not designed to provide a definitive picture, either because of the absence of controls for comparison or because of the absence of any statistical analysis, or because animal fat was measured without regard to its non-saturated fat components. If the data to be considered is restricted as indicated, there is no significant evidence of an association between type 2 diabetes and the intake of saturated fat.
SATURATED FAT AND INFLAMMATION
A search of the literature reveals little by way of definitive studies that address this question. To attempt to look at dietary intake of saturated fat and markers of inflammation carries a high risk of oversimplification of a complex problem. Most of the few studies reported use serum markers for fat intake which presents a serious problem discussed at the beginning of this review, or they study one SFA in isolation where we all eat a mixture of at least 4 or 5 such macronutrients. In addition, inflammation markers are a strong function of the distribution of macronutrients in the diet and the metabolic state of the subjects. Again, many of the problems associated with dietary studies discussed in the introduction arise if one increases or decreases the saturated fat intake because of what else is changed. The question of inflammatory diets or anti-inflammatory diets has been addresses with some success, but to discuss this anticipates the research review on carbohydrate restriction and the association between a variety of dietary patterns and inflammation. To break down the dietary patterns and point to saturated fat as the agent responsible for increased or decreased inflammatory markers appears to not be justified. There are after all a large and interrelated number of inflammation markers which makes for a much more complex situation, especially with single end point studies using, for example, mortality. Thus the discussion of diet and inflammation will be postponed. However, it is informative in passing to point out that a recent invited commentary in the journal Current Atherosclerosis Reports16 titled Modulation of Inflammation by Nutritional Interventions contains no mention at all of saturated fats.
SATURATED FAT, RED MEAT AND CANCER
Associating fat intake with cancer has over the years been part of the "fat is bad" dogma. Only a short summary of the current status of this notion will be given and this by cancer type.
Thus there appears to be no evidence for prostate or colorectal cancer to suggest that fat is bad, and for breast cancer, the evidence is characterized by inconsistencies which suggest that these studies have simply represented a struggle to deal with confounding and that probably the risk is either small or non- existent.
A recent prospective study examined the impact of meat intake on mortality. While this study will not be discussed in detail, one interesting result was that not only was red meat found to carry a positive risk for all cause mortality, but the greatest risk was seen when deaths not attributed to cancer, cardiovascular disease, or injury or sudden death were examined. That is, removing major causes of mortality increased the residual mortality risk. This would seem to suggest that there may be something wrong with this study since this implies that red meat increases the risk of dying from almost anything. It is hard to imaging a mechanism that makes red meat so universally dangerous.23 The authors regard saturated fat as one of the culprits and point to what they believe is the connection between saturated fat and cancer. As discussed above, the evidence for this belief appears to be weak or nonexistent.
The standard guidelines regarding risk and prevention of CVD and diabetes contain the recommendation to decrease fat intake and in particular the intake of saturated fat. Resérus, Willett and Hu11 conclude that more controlled long-term studies with sufficient power are needed to identify the optimal dietary fatty acid composition to reduce the risk of type 2 diabetes. But since type 2 diabetes is strongly related to insulin resistance, the small or absent effect of saturated fat on insulin resistance appears to agree with the epidemiologic studies which failed to find an association with diabetes. One can in fact argue without much risk of appearing dull that it might be better to focus on carbohydrates rather than fats. Diets high in refined carbohydrates are disastrous if judged by their effect on HDL and triglycerides, insulin resistance and inflammation.
The recommendation to reduce saturated fat intake in the context of CVD risk or prevention does not appear to be evidence-based at all. Furthermore, the quantitative suggestions for the amount reduction based on % of total energy given in guidelines appear to never have been tested in a convincing manner if at all and by themselves without regard for the patient characteristics and overall diet appear meaningless. If one accepts the proposition that there is no evidence suggesting that saturated fat intake, at least up to the intakes encountered in most North American diets, pose a significant health risk, then how can one come up with a recommended intake as a percentage of total energy that has any basis in reality? A recent paper quoted statistics that the recent daily intake of SFA in the U.S. was 27.7 g/day whereas the recommended intake was 22 g/day.24 Do the authors really believe that this difference, which is 9.8% vs. 8% of total energy for a 2500-calorie diet, is really clinically significant? Finally, it appears widely recognized that the fat reduction intervention in practice is ineffective as is the attempt to achieve significant weight loss. It in fact appears that dietary recommendations in connection with reducing the risk of either CVD or diabetes frequently fail, and sooner rather than later the patient will be on a pharmaceutical, either a statin or glucose control drug or both.
There seems to be general agreement that eating omega-3 fatty acids is beneficial in the context of CVD. However, now the American Heart Association has come out with recommendations that focus on the importance of omega-6 fatty acids and that it is unwise to severely limit their intake.25 Thus perhaps the most rational view appears to be that one should be sure the intake of omega-3 fatty acids is adequate if not high, the intake of TFAs nil and then forget about fats except for the fact that they are calorie dense. In other words, they appear neutral.
Dietary advice, which focuses on limiting saturated fat or decreasing total fat calories, also distracts from other dietary interventions that may well be vastly more beneficial. Carbohydrate restriction is one candidate and will be the subject of an upcoming review. Such diets generally involve increased fat intake. The advent of the "fat is bad" dogma might in fact be described as the beginning of the dark ages in nutritional science from which we are now just beginning to emerge. Critics of the dogma claim that the damage viewed in terms of public health has been phenomenal. Thus if one is restricting carbohydrates or undertaking a low- carbohydrate diet, there is little evidence to suggest a need for concern regarding replacing carbohydrates with saturated fat. However, this in no way diminishes the importance of certain polyunsaturated fats as potential candidates for making up the caloric deficit due to reduced carbohydrates.
The reader is referred to the book by Gary Taubes titled Good Calories, Bad Calories (Alfred A. Knopf, 2007). Part I contains a comprehensive documented discussion of the fat-heart disease--diabetes hypothesis. If one reads Part I, then the justification for Part II, which focuses on carbohydrates, will be very clear—the problem lies with carbohydrates, not dietary fat. In addition his Science article cited above is highly relevant to this discussion as is his article in the New York Times Magazine, July 7, 2002 titled What if it's all Been a Big Fat Lie? The New York Times article is in the public domain. The article in Science is also on the Internet—just Google the cited article title. The early history of diet and health discussed by Taubes is of particular interest, since it of course predates the fat is bad dogma, randomized trials, the armies of biostatisticians and the politics of nutrition. It simply concentrated on what was generally observed to work in practice. In fact by the early 20th century, the approach to treating obesity and diabetes may well have been more successful than that recommended today, and it bears a striking resemblance to carbohydrate restriction and the recognition that refined carbohydrates such as sugar and refined flour are not healthy.
The bottom line associated with this review and the review on cholesterol and atherosclerosis appears to be that modern medicine has a disturbing component that might be aptly described as mythology.