Before doing so, I just want to clarify my point of view about this: I am not suggesting that butter is a health food, nor that you might not be better off eating less of it. The balance of the evidence, for example, strongly suggests that good quality olive oil is much healthier for you than butter. So please don’t read this post and walk away thinking that butter is unambiguously good for you and that the more of it you eat the better, as is promoted by some health influencers on the internet.
On the flip side, butter has been part of the human diet for at least several thousand years, and I am skeptical that, taken in moderation, eating butter is a form of suicide, or that it is worse for us than certain industrially produced oils (soybean, corn, etc) that have only existed for at best a century, and that cannot be produced without the use of toxic chemicals and sophisticated machinery. My critique of this study is not that it finds some possible health risks from eating butter, but rather that it’s a poorly done study that shows at most a very weak link between butter consumption and poor health, but that is nonetheless reported upon as being firm evidence that butter = death.
Without further ado then, here are my main critiques of the study:
This is a retrospective epidemiological study. In other words, the researchers mined data from a database, plotting the amount of butter people in the database reported consuming versus their risk for various diseases and death. These kinds of studies are inherently of poor quality (because they are prone to many confounding variables) and can at best find associations between certain variables, but never definitively prove causation. As the saying goes in science, “correlation does not prove causation.” (To understand this point better, consider for example that one could graph ice cream consumption v. drownings and show a correlation between the two. But of course, this wouldn’t be because ice cream causes drownings. Most of us would quickly grasp that the two are related only insomuch as people both go swimming and eat ice cream during the warmest months of the year).
The study compares butter to all plant oils, including olive oil. But olive oil is well known to be healthful, so this risks biasing the results. By way of analogy, if I were to place one major league ball player on the local little league team, that team’s batting average would most certainly go way up, but that would tell us nothing about how good or bad all of the other kids on the team are as hitters. Similarly, by placing the “all star” olive oil in the “plant oils” bucket, we might be getting an incorrectly rosy view of “plant oils”. And indeed, when the researchers separated out olive oil from other vegetable oils, only some of the non-olive oil plant oils appear to still show benefit.
A key flaw of these kinds of epidemiological studies is that they can never fully account for “healthy user bias.” What is healthy user bias? It’s the fact that people who care about their health make all sorts of decisions that lead to better health outcomes. Where this is relevant is that, since at least the 1970s, people have been told repeatedly to avoid eating excessive amounts of animal fats. Ergo, the folks in this study who ate the least butter are, almost by definition, the folks who care more about their health, whereas the folks who ate the most butter are the people who don’t.
Trying to isolate individual nutrients or foods as this study does is ultimately less helpful than looking at entire dietary patterns. A person who eats a lot of butter might do so because they are eating lots of sweets and pastries. Or they may eat in the style of a Frenchman (France famously has one of the highest rates of butter consumption in the world, and simultaneously a relatively healthy population with below average rates of heart disease), consuming a diet of salads, small portions of high quality meat cooked in butter laden sauces, vegetables, fresh fruit, and cheese. Do we really think that these are the same diets and will have the same health impact, even if they contain the same total amount of butter?
The study finds a small increased risk of death only in those participants who ate the most butter. It doesn’t find any health harms from more moderate butter consumption. So even if the results of the study are to be taken at face value, they still imply that moderate butter consumption is not a problem.
At worst the study shows a 17% increased risk of death in the heavy butter users, but this is a “relative” risk, meaning in real numbers the effect is much smaller. For example, if you had a 10% risk of dying over the next decade, a 17% increase in risk would mean that your risk of dying increased to 11.7%. So even if we assume the study conclusions are correct, the effects are fairly small.
In conclusion, I wouldn’t make much of this study, and certainly wouldn’t take it as seriously as the mainstream press seems to have. If the study results spook you and you choose to replace all of the butter in your diet with high quality olive oil, then there’s probably no harm done and you might be improving your health slightly. But since nutrition does not exist in a vacuum or a perfect reality, I would hate for people to over-react to this study, get rid of their butter, and start eating more processed carbohydrates and artificial fats (such as margarine) in exchange. That would almost certainly lead to worse health.
