Butter is not back says study. Here’s why I’m ignoring it

“Butter is not back,” according to a new study by researchers at Harvard’s T.H. Chan School of Public Health.

The researchers raised the important point that the low-fat fads of the 80s and 90s led to the substitution of saturated fat with refined carbohydrates to our detriment, and so they set out to find what replacement foods were necessary to reduce heart disease risk.

They concluded that “saturated fat and refined carbohydrates appear to be similarly unhealthful.” and in the press release proclaimed that “limiting saturated fat is still best for heart health.” 

In the great heart health debate, the study is being cited as proof that saturated fats are unhealthy.

The researchers looked at the Nurses Health Study (84,628 women) and the Health Professionals Follow-up Study (42,908 men) over a period of 24-30 years. Participants were free of diabetes, cardiovascular disease, and cancer at baseline. The outcomes of interest were non-fatal heart attacks and deaths as a result of heart disease, identified through a review of medical records.

Participants completed a food frequency questionnaire once every 2- 4 years, and daily fat intake by type was calculated by multiplying the frequency of the food consumption with its nutrient content using US Department of Agriculture food composition data.

The researchers estimated that replacing 5% of calories from saturated fats with the equivalent of polyunsaturated fats was associated with a 25% lower risk of cardiovascular disease. Replacing saturated fats with monounsaturated fats was linked with a 15% lower risk, and replacing them with whole grains was linked with a 9% lower risk.

“On the other hand, swapping 5% of saturated fat calories for the same amount of refined carbohydrates and sugars did not change CHD risk.”

Does the raw data agree with their estimates?

Cardiologists like Dr. Axel Sigurdsson, in his article questioning the study, point out that the data tell a different story. “These numbers are not dealt with by the authors of the paper,” he writes.

He’s right and I did the number crunching myself. For the highest fat intake there was less coronary heart disease and vice versa for highest carbohydrate intake. This held true for saturated fats and whole grains.

For the highest level of saturated fat intake, there were 1,434 cases of CHD vs 1,599 for the lowest intake. For the highest level of whole grain intake there were cases of 1,520 cases of CHD vs 1,493 for the lowest level.

Overall, Sigurdsson writes, “There were 16% more cases of CHD in the group with the lowest fat intake (ca. 26% of calories) compared with the group with the highest intake (ca. 41% of calories).”

Saturated fat was generalized in this study; it’s much more complicated.

Sigurdsson raises the important point that dietary saturated fats “are derived from many different foods,” and “these have many other ingredients and characteristics that modify their health effects. Therefore judging the health effects of foods entirely from the amount of SFAs may be dreadfully misleading.”

Saturated fatty acids come in different forms. For example the lauric acid in coconuts is a saturated fatty acid that has been shown to be neutral for heart health. It’s just that Americans don’t usually get their saturated fat intake from foods like coconuts and grass fed butter, but from heavily processed, industrially produced animal products. How could the researchers control for all the crap in fried and fast food?

The researchers also didn’t address the differences in subtypes of the other fats in the study. For example, polyunsaturated fats like the omega-3s EPA and DHA from fatty fish have been shown to be more beneficial for heart health than other polyunsaturated fats. And monounsaturated fats like the omega-9s in extra virgin oil have been shown to be more beneficial than ones from other sources.

They also fall back on old knowledge on saturated fat’s effect on LDL cholesterol (LDL-C), the so called “bad kind.” They write, “a large body of evidence indicates that higher intake of most dietary (saturated fatty acids) increases blood levels of low-density lipoprotein cholesterol and the low density lipoprotein to high-density lipoprotein ratio, both of which are associated with a higher risk of CHD.”

First of all, while consumption of saturated fats tends to raise levels of LDL cholesterol in the blood, they also tend to raise the levels of HDL cholesterol, the so called “good kind,” providing a compensating effect. But the researchers don’t say this, they simply write that the LDL-C to HDL-C ratio is increased, shifting the balance toward CHD.

Second, using LDL-C to gauge CHD risk is past its prime. The focus needs to be on the particle number (LDL-P).

Third, saturated fat increases large buoyant LDL particles, which are not known to promote plaque formation and are possibly beneficial. It is carbohydrates that increase levels of plaque forming small dense LDL particles.

Observational or prospective studies should not guide public health.

There were more limitations than the authors dealt with or recognized. The multivariable model employed adjusted for known risk factors for coronary heart disease, including body mass index, family history of diabetes and myocardial infarction, menopausal status and hormone therapy use, regular use of aspirin, smoking status, physical activity, presence of hypertension and hypercholesterolemia, and percentage of energy from protein and cholesterol.

But their model was flawed, and this isn’t the first time the folks from Harvard got it wrong, according to Gary Taubes, author of Good Calories, Bad Calories.

A 2012 study claimed that red meat was deadly. However, looking more closely at the data as meat consumption went up among participants so did unhealthy behaviors: smoking, drinking, lack of physical activity, and body mass.

Similarly, as Sigurdsson also mentions, with this current Harvard study, the people with the highest level of saturated fat intake were more likely to have lower physical activity, higher BMIs, higher intakes of cholesterol and were more likely to smoke.

(As an aside, I’d like to point out that controlling for cholesterol intake in the multivariable model was not practical considering that it is now known, and finally reflected by the normally change-resistant USDA dietary guidelines, that dietary cholesterol doesn’t impact blood levels of cholesterol much at all, and doesn’t promote a risk for heart disease.)

The study’s reliance on the participants’ dietary recall through food frequency questionnaires spaced 2-4 years apart is also problematic. As Taubes writes, “The fact that they use questionnaires that are notoriously fallible is almost irrelevant here because the rest of the science is so flawed.”

An observational study is the first point on a path to more rigorous testing. It does not give the researchers a right to represent their findings in a way that lets the media run with a recycling of an unfairly worded press release.

But that is what happens every time a study like this is unveiled; every time skewing the public’s perception of what nutritional guidelines to live by.

As the researchers say themselves, “given its observational nature, our study cannot prove causality.”