From Abracadabra to Zombies
Mass Media Bunk
a commentary on news stories or articles in the mass media that provide false, misleading, or deceptive information regarding scientific matters or alleged paranormal or supernatural events.
Note: Mass Media Bunk is now Skeptimedia.
Mythbusting the Cholesterol Myths
April 14, 2006. Uffe Ravnskov's The Cholesterol Myths - Exposing the Fallacy that Saturated Fat and Cholesterol Cause Heart Disease (2000) is the bible of cholesterol contrarianism. (Henceforth, the author will be referred to as UR.) These contrarians call themselves skeptics but their goal is not to examine all the evidence and think critically about it in the large context of our medical knowledge. Their goal is to cherry pick data to support their contention that low cholesterol is bad for you and high cholesterol is positively good for you. A diet high in saturated fats is never unhealthy, in their opinion. Not everything they say is false. Much of it is true. But everything they say is taken out of context to support their contention.
Joel M. Kauffman, a cholesterol contrarian, exemplifies the claims and methods of these folks. He wrote in Skeptic [vol. 12, no. 2]: "Cholesterol is highly protective against cancer, infection and atherosclerosis" and "high TC [total cholesterol] and LDL levels are beneficial at all ages." These are extraordinary claims and could be deadly, if taken seriously. It is hoped that no reader of Skeptic will throw away his or her cholesterol-lowering medications after reading these claims. Kauffman's source for this claim is a paper by UR: "High Cholesterol May Protect Against Infections and Atherosclerosis." Kaufmann changes UR's "may protect" to "are beneficial" and adds "cancer" for good measure. This kind of looseness with the facts is something to look for when reading the cholesterol contrarian literature.
This is the first installment of a "review" of UR's book. I use scare quotes because I do not plan to point out the good and the bad in the book. I plan to focus only on the bad. Part 1 of this review will cover the foreword, the introduction, and chapter one. But first let me assert that I agree with the contrarians that there is not a strong body of peer-reviewed published research that shows that a person who eats a low-fat diet is guaranteed to have low cholesterol which will prevent that person from getting atherosclerosis which in turn will prevent that person from getting a heart attack. Nor is there strong evidence that a person who eats a high-fat diet is guaranteed to have high cholesterol and get atherosclerosis and die of a heart attack as a result.
From these facts and a few others we may reasonably conclude a few things. Cholesterol and atherosclerosis are not the only causes of heart disease; having low cholesterol does not mean you are exempt from atherosclerosis or heart disease. Not eating animal fat and cholesterol does not mean you are exempt from atherosclerosis or heart disease. And, eating a diet with a good amount of animal fat and cholesterol does not necessarily mean you are promoting atherosclerosis or heart disease. There are lots of other factors that need to be considered: your past health history and the current state of your health, your family history with cholesterol levels and heart disease, your genetic predisposition to high cholesterol and/or heart disease, do you smoke, are you grossly overweight, do you exercise, an so on.
One technique the cholesterol contrarians use is to ask disturbing and suggestive questions, such as: "Could attempts to reduce cardiovascular mortality by lowering blood cholesterol actually do harm?" This is asked by Michael Gurr in a foreword to UR's book, p. xiii. He answers that several people think that it might and they've cited evidence.
UR concludes it is a myth that high fat foods cause heart disease because studies do not show that a diet high in saturated fat is a sufficient condition to bring on a heart attack nor that a diet low in saturated fat is a sufficient condition to prevent a heart attack. Here he identifies "causes" with sufficient condition. Some causes, however, are necessary but not sufficient conditions. For example, some viruses must be present (are necessary conditions) for certain diseases to occur but they are not sufficient conditions. Thus, the virus may be present but not manifest itself in illness. Perhaps there isn't enough of the virus to cause illness, perhaps it is isolated in an area of the body where it can't do harm, or perhaps one's immune system has prevented the virus from doing harm. More important, many things are causal factors rather than single causes. A high fat diet by itself may not be a sufficient condition to cause heart disease but it can be a major contributing factor in some people, e.g., someone who is obese, has a family history of heart disease, smokes, and works as an air traffic controller or in some other stressful occupation.
UR notes that 20% of those who die from heart attacks don't have atherosclerosis and concludes that atherosclerosis can't be a cause of heart attacks. Again, he confuses "sufficient condition" with "cause". Another conclusion from the data is that there are other causes of heart attacks besides atherosclerosis. This confusion cannot be accidental since he clearly knows the difference between "risk factors" and sufficient conditions (see p. 22).
Another technique UR uses is to accuse those who disagree with him of "medical quackery" (5). He says they "routinely belittle, deny or explain away any scientific observations that contradict their idea" (9). He didn't need to be clairvoyant, however, to anticipate that he would be criticized for being unscientific and incompetent (12). In fact, he provides his own supportive evidence for this very criticism when he describes what should be the case if a "scientific hypothesis is sound" (12). Of course the hypothesis should agree with the data. However, to say that the hypothesis "must agree with all observations" is to belie a fundamental misunderstanding of the nature of causality and of medical research. His statement is correct if we are talking about a universal generalization. If I claim all crows are black all you need is one non-black crow to prove me wrong. In any case, the data UR is concerned with is mostly in the form of studies looking for statistically interesting correlations and differences. It is often the case, that different studies on the same topic produce different results. There are many reasons for this and I needn't go into them here. This is especially true when you are dealing with such complex organisms as human beings are trying to isolate significant factors that might be causally related to various outcomes such as coronary heart disease (CHD).
The straw man technique is another favorite of UR. For example, in setting up his opponents' argument he says they claim that the amount and the type of fat in our diet determines the level of cholesterol in our blood. They say that if we eat an atherogenic diet ["a diet containing too much cholesterol and saturated fat (found in animal products, such as meat, milk, eggs but also in palm oil and coconut oil) and too little polyunsaturated fat"], our blood cholesterol will be high and high blood cholesterol is the main cause of atherosclerosis, which causes CHD. In fact, it is well known and propagated in the literature that other things besides diet affect blood cholesterol and that other things besides high cholesterol can cause atherosclerosis and that other things besides atherosclerosis are causal factors in CHD.
At the conclusion of his Introduction, UR writes that "the diet-heart idea ... seriously conflicts with the laws of logic...." (13). I submit that even his straw man portrayal of the "diet-heart" idea does not conflict with the laws of logic. His straw man is based on a distortion of the data and the claims of medical professionals. His logic is flawed because he selectively identifies cause with sufficient condition when it suits his purposes.
In his first chapter, what UR thinks is good logic is revealed. He notes that since 1970 death from fatal heart attacks in Japan has declined while animal fat consumption has increased. He thinks this supports his belief that animal fat in the diet is not a major cause of heart disease. Yet, he notes that more people are surviving heart attacks today because of better medical treatment (31), though he doesn't mention that diet, aspirin, and statins are part of this better treatment. What he needs to support his hypothesis is data that show that as one's diet increases in animal fat one's chances of a fatal heart attack decline. He doesn't produce these data because they don't exist.
A similar kind of reasoning occurs in the writing of another cholesterol contrarian, Marshall E. Deutsch, who cites a study that found an increased mortality rate among elderly Japanese-American men with low cholesterol levels (Skeptic, vol. 12, no. 2, p. 20). Deutsch seems to think that the data support the notion that low cholesterol is bad for everybody. Given his logic, he may even think these data support Kauffman's hypothesis that high cholesterol is good for you. He doesn't seem to realize that the data are consistent with the much more reasonable hypothesis that when these men became chronically ill and malnourished their cholesterol levels dropped.
Chapter two is called "Myth 2 - High Cholesterol Causes Heart Disease." Here Ravnskov provides an example of how to misuse statistics to support your position. He cites the Framingham Heart Study, which found decreasing levels of cholesterol associated with increased mortality among older participants. He also compares death rates of those with very high and those with very low cholesterol readings. The death rate was higher in those with low cholesterol readings. He takes this to mean that either decreasing cholesterol is bad for everyone or that decreasing cholesterol is a significant causal factor for the mortality rate. He rejects the researchers' explanation that other (disease) factors were causing both the lowered cholesterol and the mortality.
Ravnskov knows that if there is a correlation between x and y there are at least four possibilities regarding causality. 1) There is no causality involved. The relationship is coincidental. 2) x causes y. 3) y causes x. Or, 4) some other factor, z, causes both x and y. A researcher can't just pick the one of the four possibilities that fits with his hypothesis and declare it to be the correct one!
He also cites statistics regarding those who had a second heart attack. Lowering cholesterol didn't help prevent the second attack. These data do not support his claim that high cholesterol doesn't cause heart disease. Other factors besides cholesterol levels have to be considered. Also, perhaps those who suffered a second heart attack did not lower their cholesterol levels enough to be beneficial. In any case, the more interesting data would be that which is relevant to first heart attacks and here the data supports the claim that lowering cholesterol significantly lowers the risk of coronary heart disease.*
Ravnskov claims that he's shown "that there is little or no evidence that blood cholesterol plays any role at all in coronary heart disease." This is an exaggeration typical of his bombastic approach to argumentation. One of his favorite tactics is to raise possibilities that favor his interpretation and ignore all other possibilities. For example, when he considers a study led by Dr. Jack Medalie, he raises the possibility that the group that showed the highest number of heart attacks and which had the highest LDL-cholesterol levels "may have included more stressed, overweight, inactive and smoking individuals" than the group which had the lowest incidence of heart attacks and also the lowest LDL-cholesterol levels. There were 78% more heart attacks in the high LDL group (220-500 mg/dl) than in the low group (77-189 mg/dl). True, but it is also possible that the group with the lowest LDL levels had more stressed, obese, inactive smokers. Neither possibility is more likely than the other. Medalie's study involved 10,000 male government and municipal employees. To make his point, Ravnskov needs to do more than raise the possibility that these other factors may account for the significant difference. He should show that when adjusted for stress, weight, activity, and smoking, the difference disappears. He can't just assume that when adjusted for these factors, his position will be supported.
* AmeriCares *