SATURATED FAT ON TRIAL
It is now time to look at the controlled clinical trials on saturated fat. It bears repeating; clinical trials are the gold-standard of experimental science. They are expensive, time-consuming and definitely much more difficult to control than the conventional observational study. Controlled trials can
prove causality because variables have
Before we expound the studies we must define some key words. The term 'randomized' simply means that participants in the study have been randomly allocated to the intervention and control groups. This process helps immensely to remove both known and
unknown confounders such as self-selection bias. Because all subjects are randomized, either group may have considerably different CHD baseline risks.
The notions of 'blinding' also need explaining. A 'double-blinding' is an ideal measure of control that essentially means both the subjects and the researchers would be none the wiser as to who is in the control and intervention groups, a safeguard against researcher bias and any possible 'placebo effect' on study participants. Unfortunately, 'double-blinding' is extremely hard to achieve in dietary experimentation. Only a select few trials have been able to achieve the 'double-blind' status through incredibly fastidious study designs that involve specially prepared foodstuffs. For the greater majority of clinical trials, a 'semi-blinding' is used. This means study subjects are aware of their diets although researchers are not.
There have been over 25 CHD dietary intervention trials since the first was published in 1955. However many of these have been multifactoral in design. Essentially, this means that saturated fat restriction was also accompanied by, for example, increased fruit and vegetable intake, higher fish intake, smoking cessation and stress alleviation classes etc. This makes attributing any effect of the intervention to saturated fat alone impossible. Therefore, only clinical trials which have manipulated the sole variable of saturated fat will be analysed. This way it is possible to equate any benefit (or none) from the intervention to saturated fat restriction, completely
The first clinical study was conducted in 1965 by Rose et al. It was a randomized and semi-blinded trial that lasted 2 years involving 80 subjects229
. The intervention compared replacing animal fat with corn oil or olive oil on CHD. The results were unequivocally abysmal. By the end of the trial, over 48% and 43% of those in the corn oil and olive oil groups had a major cardiac event, respectively, compared to the 25% in the animal fat control group. The authors, with an obvious disapointed undertone concluded, "The likelihood that the trial failed by chance to detect a true and important benefit from corn oil was extremely remote. It is concluded that under the circumstances of this trial corn oil cannot be recommended in the treatment of ischaemic heart disease. It is most unlikely to be beneficial, and it is possibly harmful."
In the same year Bell et al. published in the Lancet
a semi-blinded, randomized trial that lasted 3 years and involved just over 250 subjects230
. The intervention was to simply reduce total and saturated fat. When results were calculated, there was no difference in the incidence or mortality of CHD and total mortality. The authors remarked, "A low-fat diet has no place in the treatment of myocardial infarction."
Again in 1965, a third consecutive trial was published, this time by Hood et al. involving 460 subjects and lasting 5-17 years231
. The study was neither blinded nor randomized. Researchers replaced animal fat with polyunsaturated vegetable fat and saw marked reductions in total mortality. Over 3 times more
people died in the animal fat control group. However there are a number of issues with this. Firstly, there were significantly more elderly in the control group, this alone is reason enough to explain the vast difference in mortality rates. But furthermore, we find in the midst of the study that researchers say themselves that only those in the intervention group were strictly followed. The animal fat control group was more or less left to their own devices and ignored. Further credence is added to this observation when one notices that CHD incidence was in fact the same between groups. Only the intervention group had a reduced CHD mortality rate, very much suggesting that the intervention group received closer medical attention than the control group. On these pretenses and also considering that the trial was non-blinded and non-randomized, it's hard to definitively assert that the replacement of animal fat with polyunsaturated vegetable oils were the direct cause of the reduced mortality.
A year later in 1966, another non-blind and non-randomized trial was published in the prestigious Journal of the American Medical Association
by Christakis et al.232
. The study involved over 800 men and lasted 4 years. Those in the intervention group were assigned the 'Prudent Diet', which required saturated fat to be severely restricted while being replaced by polyunsaturated oils. This diet was contrasted against a control group who continued eating their normal diets. Researchers found a reduction in non-fatal coronary events in the intervention group but most surprising was the fact that while the control group has 0 fatal coronary events, the intervention group had 9. Total mortality was also calculated to be 27 in the intervention group and only 6 in the control.
Another year later in 1967, Bierenbaum et al. published the results from a 5 year long trial involving 100 men that was randomized but not blinded233
. The intervention consisted of a 28% fat, 50/50 combination of polyunsaturated vegetable oils that included corn and safflower oils. The other group was assigned a similar 28% fat, but instead receiving a 50/50 mix of polyunsaturated peanut oil and extremely rich in saturates, coconut oil. Altogether, the group with coconut oil had twice the saturated fat intake of the other group. After the results were tallied, the death rate between the two groups were the same.
In 1968 the National Diet-Heart study was published in Circulation
. The trial involved over 2000 participants and was both double-blind and randomized lasting 2 years234
. The intervention group received foods that were higher in polyunsaturated fats but lower in saturated fats. Interesting enough no mortality figures were published for neither all-cause nor CHD. However CHD incidence between groups were virtually identical.
In 1968, the Medical Research Council published the results of a secondary prevention, semi-blind, randomized trial involving almost 400 subjects where animal fat was replaced with soya-bean oil235
. After 2-7 years, total and CHD mortality were almost the same between the two groups.
One year later in 1969, perhaps the most important clinical trial was published. It is important for two reasons. Firstly, because it is the longest running (8 years), double-blinded and randomized trial to have ever been conducted. And secondly, because it was an excellently designed and controlled study. This was the Los Angeles Veterans Administration (LA VA) study which included nearly 850 institutionalized veterans and investigated the effects of substituting saturated fats with polyunsaturated vegetable fat236
. Because this study showed several different outcomes, it is absolutely imperative that the results be construed correctly. Patients were housed in a hospital where they ate every day for every meal in one of two dining halls. One hall served the exact same food, however in one it was cooked with a mix of vegetable oils and in the other the food was cooked with butter. When results were analysed, it was found that the polyunsaturated intervention group did have statistically significant lower CHD mortality. Although this is not without its explanations. Randomization was unsuccessful in evenly distributing the number of smokers between groups. The butter control group was significantly handicapped; including 60% more moderate smokers and 200% more heavy smokers. Furthermore, the control group was fed a Vitamin E deficient diet. The Vitamin E deficiency and the substantial number of smokers in the control group offer a perfectly plausible explanation for the unfavorable CHD mortality outcomes completely irrespective of butter intake. The most telling part of this study was that despite the disparity in the CHD mortality between groups, total mortality was the same. It had appeared to be a case of disease substitution; the polyunsaturated fat intervention group suffered statistically significant greater
non-cardiovascular deaths than the control group. If one looks at the survival curve for non-cardiovascular deaths, it seems to be approximately equal for the first 7 years. Between years 7 and 8, the curve take a steep plummet and continues to do so right till the end of the study. One can only imagine the curve to continue falling so dramatically if the study continued. And finally, cancer mortality was higher in the polyunsaturated fat intervention group, and incredibly showed no relationship with smoking. The LA VA study boasts qualitative superiority over other clinical trials and this certainly needs to be acknowledged. While a significantly higher CHD mortality was found in the control group, this can hardly be attributed to butter intake with such a handicapped baseline risk profile.
The next clinical trial was the Finnish Mental Hospital(s) (FMH) trial published in 1972. It was non-blind and non-randomized and is the longest running clinical trial at 12 years to investigate replacing animal fat with polyunsaturated oils, namely soybean oil and margarine237
. It is also the worst designed, controlled and conducted clinical trial to date. The FMH trial is the only clinical study ever to use a crossover design. Subjects were required to eat an intervention diet for 6 years before changing to a control diet. Such an absurd design has obvious implications. How is one to know whether results were caused from the diet 6 years ago, or the current diet? Such distinctions cannot be made. Furthermore, such was the deplorable level of control in the FMH trial that those who stayed even a month in the hospital and those who left the study and were re-admitted at a later date were all included in the results. The amount of sugar also fluctuated as much as 50% and total fat by as much as 25% between diet periods. Total carbohydrate intake also varied as much as 17% between diet periods. It is impossible to call the FMH study a controlled
trial, it almost entirely useless.
In 1978, the Sydney Diet-Heart study was published. The trial was not blinded although it was randomized and included some 458 men238
. The intervention was counselling a group to reduce saturated fat to 10% of daily caloric intake and to increase polyunsaturated fat to 15%. No particular instructions were given to the control group. After 5 years the intervention group suffered a mortality rate higher
than the control group.
In 1989, the Minnesota Coronary Survey was published. It was a double-blind and randomized study however it was short in duration at only 384 days239
. The intervention was a higher polyunsaturated fat, lower saturated fat diet. No difference was found for CHD incidence, mortality or total mortality between groups. Admittedly unsurprising given the length of the study.
The final study to involve specifically saturated fat restriction was the DART trial published as well in 1989. It was semi-blinded and randomized, lasting 2 years240
. Subjects in the intervention group were told to reduce total fat intake while also increasing the ratio of polyunsaturated fats to saturated fats. No differences were seen in CHD or total mortality between groups.
Since the '60s, a steady procession of clinical trials investigating the effect of saturated fat restriction on CHD mortality have been published in the medical journals much to the chagrin of the diet-heart hypothesis exponents. If saturated fat was the demonic plague on modern society as so readily portrayed by health authorities, this would have been clearly reproducible in controlled trials. But it hasn't. These trials illustrate that if we manipulate a single variable whilst keeping everything else controlled as best as we can, what we are really evaluating is cause and effect. Saturated fat restriction has never
been decisively shown to lower CHD mortality; there is no cause and effect observable and thus it is only logical to conclude that saturated fat does not cause CHD
The idea that saturated fat causes heart disease as a hypothesis per se, has graduated into one of the most hallowed precepts in modern science. And thus, the problem is not trying to invalidate the hypothesis, for we could have done this as early as the 1970s, the problem is trying to reverse an entire culture, trying to avert the masses from this entrenched medical mendacity that has nothing to do with public health, and everything to do with the appeasement of vested political and commercial interests. We can observe a shocking absence of evidence in both uncontrolled epidemiological research and controlled clinical trials that suggest a causal relationship between saturated fat and CHD. The weight of the literature is enormous, therefore making it absolutely necessary that one assesses it in its totality. It is an article of faith in our society that saturated fat is a deleterious substance; this has consequently led to the demonization of many nutrient-dense foods and food groups entirely. Our 'health authorities' insist that saturated fat should be consciously avoided to prevent disease. However, there is not an iota of evidence to prove this.* NOTESReference 7
: This study actually found a correlation between saturated fat and death rates however during the years 1951-1960, saturated fat as a percentage of total calories remained relatively constant (although there was an increase in PUFA consumption) while death rates from AHD continued to rise
. Whatever accounted for this increase, it could not have been saturated fat, thus nullifying any previous suggestion of a correlation. Reference 8
: Two analyses were presented both investigating 31 countries. One looked at CVD and the other at IHD.References 12,13,14
: No individual figures were given for saturated fat however total fat consumption was the same between CHD and CHD-free individuals.Reference A15
The 'A' stands for 'additions'. Since the sheer volume of research is so large, it becomes inevitable that some studies will be missed. Including them into the existing number sequence would make formatting and referencing a nightmare, so 'A' is used for any additions made after the original article. Reference 56
: Figures retrieved from 'Recorded adult per capita consumption, from 1961, Wine' tab. The most current statistics were not available for both countries, so the next most current for both countries were used (2003) to ensure accurate comparability. Reference 57
: CHD statistics can be found on page 39. Statistics for 2003 were used to appropriately correspond with Reference 56.Reference 85
: Data can be found by clicking on 'Food Balance Sheets', then selecting Japan and the preferred year. Click 'show data' to generate the table. Figures were taken from the 'Grand Total' row and the 'Fat supply quantity (g/capita/day)' column. Figures were 34.2 and 89.8 in 1961 and 2000, respectively.Reference 112
: The article is in French. 'Google translator' should suffice. FINAL WORDS
I am honestly stoked this is completed. Countless hours of reading paper after paper turned into a half-year obsession with the literature concerning saturated fat and heart disease. At nearly 7000 words, this is by far the largest piece I have ever written. But sadly, the depth of this article still remains superficial - the subject of saturated fat and heart disease is of immense proportions. For this reason, I have gone out of my way to specifically format citations for easy access to anyone that's interested. I urge you to consult them for yourself. Please make any criticisms and errors known to me and I will redress the issue. Thank you for reading.