You have probably read hundreds of recommendations to eat more polyunsaturated fatty acids (PUFAs), which in practice means omega-6 and omega-3. If you shop at Whole Foods, you may see Udo’s Blend, a blend of PUFAs which includes both omega-3 and omega-6, as if someone isn’t getting enough omega-6. It is unquestionable that omega-3 is beneficial but there is plenty of evidence that omega-6 is harmful, starting with the Israeli Paradox. Why are they lumped together?
A just-published paper in the British Journal of Nutrition makes several new points about the relation of PUFAs and heart disease. Its main point is a new look at experiments in which one group was given more PUFAs than another group. Those experiments — there are only about eight — can be divided into two groups: (a) experiments in which the treated group was given both omega-3 and omega-6 and (b) experiments in which the treated group was given only omega-6. The two groups of experiments seem to have different results. In the “both” experiments the treated group seems to benefit; in the “only omega-6″ experiments, the treated group seems to be worse off. Suggesting that omega-3 and omega-6 have different effects on heart disease. They have been lumped together because experiments have lumped them together (varied both at the same time).
Experiments that try to measure the effect of PUFAs usually say they are replacing saturated fats. More PUFAs, less butter. The paper points out that studies of the effect of PUFAs have at least sometimes confounded reduction in saturated fats with reduction in trans fats. Benefits of the change may be due to the reduction in trans fats, not the reduction in saturate fats.
The paper also makes several good points about the Finnish study, a classic in the fat/heart disease literature. Supposedly the Finnish study showed that PUFAs (replacing saturated fats) reduce heart disease. It had hundreds of subjects but they were not randomized separately. The subjects were divided by hospital. Everyone in one hospital got one diet, everyone in a second hospital got a different diet. This meant it was easy for there to be confoundings (i.e., the treatment wasn’t the only difference between the groups). Indeed, there were big differences in consumption of a certain dangerous medication and margarine between hospitals. (Margarine is high in trans fats.)
Perhaps the first author, Christopher Ramsden, who works at NIH, is responsible for the high quality of this paper.
Thanks to Susan Allport.
The “Israeli Paradox” is an ecologic study without much ability to separate the effect of omega-6. In fact, Israeli arabs who consume tons of olive oil and little omega-6 have higher rates of heart disease than Jewish Israelis who consume omega-6. Other studies have shown that countries that have shifted their cooking oil to omega-6 rich unsaturated fats tend to witness decreases in heart disease relative to countries that have not (e.g., in eastern europe). Also, rigorous epidemiologic studies show a robust inverse association between omega-6 consumption and heart disease, which is evident no matter how you slice the data, and is particular strong among the young (see the work from Willett’s lab at Harvard). Obviously, there may still be residual confounding, but this data certainly doesn’t support that omega-6 is particularly harmful.
Thanks, Vic, that’s good to know.
The group at Harvard has been confounding the effects of omega-3s and omega-6s for decades. We absolutely require omega-6s in our diet (on that we all agree) but that robust inverse association b/n omega-6s and heart disease would disappear if omega-3s and omega-6s were properly controlled for (as it is in the current study by Ramsden et al.)
Susan, thanks for your comment. I don’t know what you mean by confounding (“confounding the effects of omega-3s and omega-6s”). In experiments (such as the data analyzed by Ramsden et al.) confounding is changing two things at once. In surveys (such as what Willett et al. do), the only use of “confounding” I know about is the idea of unrecognized confounders — meaning something that wasn’t measured is correlated with something that was measured. But Willett et al. measured both omega-3 and omega-6 consumption so that didn’t happen.
How good is the evidence for the harmful effects of trans fats? I seem to remember an article (or interview) where Gary Taubes said that the evidence is pretty weak. But I just spent a few minutes Googling it, and I can’t find the Taubes reference.