Blog | Friday, August 31, 2018

Milking dairy data for more than they are worth


Another week, so naturally, another round of breathless headlines about the latest dietary study, this one about dairy fat, purportedly telling us what's what, but really just, well, curdling the way.

The headlines in question say things like Full Fat Dairy is Probably Better for You than Skim, and Forget the Fat-Free Dairy: Just Eat the Full Fat Stuff Already. Even an eminently sensible, well-informed, physician colleague at The Atlantic published a piece entitled The Vindication of Cheese, Butter, and Full-Fat Milk. Knowing how these things work, I suspect an editor chose this headline for a commentary that, predictably, reaches quite sensible conclusions overall. So, by all means read Dr. Hamblin's article, but whatever you do, don't stop at the headline.

Is dairy fat exonerated, and good for us now? No more so after the new study than after a very similar study published two years ago.

Just like its predecessor, the current study, in the American Journal of Clinical Nutrition, looked at health outcomes over time in people who had submitted blood samples that were analyzed for, among other things, levels of various fatty acids associated with consumption of dairy fat. Participants also underwent assessments for various health measures, such as blood pressure, body-mass index, and waist circumference, and completed food frequency questionnaires to gauge their overall dietary intake patterns.

The findings generating all the effusive expostulations were that higher versus lower blood levels of several fatty acids linked to dairy fat were associated with neither higher, nor lower, levels of cardiovascular disease or death from all causes. Higher levels of one particular fatty acid, heptadecanoic acid for those who care about particulars, was associated with reduced risk of death from stroke, and increased risk of death from all non-cardiovascular causes. Make of that what you will.

To begin with methods, this was an observational study of association only. Such studies never prove cause and effect, they merely show that A and B seem to happen together. In this case, A is intake of some full-fat dairy, and B is some level of risk for death and disease. I say “some level of risk,” because the association reported here, overall, was “no difference” in disease and death rates, not any benefit. Why don't observational studies prove cause and effect? Because A and B may well be “true, true, and unrelated.” In other words, there may well be an unmeasured “C” that is the real reason for “B,” and “A” is just along for the ride.

My objection here is non-denominational. I find that all sides in the diet wars point out the limitations of study methods when they DON’T like the conclusions, and blithely re-tweet ad infinitum with no mention of those same limitations whey they do like the conclusions. This is true of vegans, carnivores, and all factions in between. On behalf of a public that deserves better: I protest. This time, and every time. What's good for the vegan goose is good for the Paleo gander, or should be. This study proved nothing about cause and effect, because it is not the kind of study that can.

As for the findings, they are rather different than those headlines might cause you to believe. There were, of course, lots of data and details compiled, but the gist distilled to this, in the authors' own words: “In summary, our prospective investigation of older adults suggests no major associations between long-term exposure to circulating dairy fatty acids and total mortality …” Was there a protective effect, overall, of dairy fat? No. Was dairy fat consumed “instead of” low-fat dairy, as headlines imply? No. In fact, those with higher dairy fatty acid levels in their blood reported slightly higher intake of dairy fat than their counterparts, and also reported the same, or slightly higher, intake of low-fat dairy, too.

Let's capture that: apparently, intake of full-fat dairy was in addition to, not instead of, low-fat dairy. So, for instance, these could be people who eat some cheese, and use fat-free milk in their cereal. Let's hold that thought for now.

One more useful comparison of those with higher, and lower, fatty acid levels in their blood is overall health status at baseline. Those with the lower fatty acid levels had in general a higher BMI, a higher waist circumference, and were more likely to be on medication for hypertension, diabetes, or both. (They had less education on average, and drank more alcohol on average, too.) Does this mean that illness, or medication, or alcohol intake might affect fatty acid levels in the blood, or that those somewhat less healthy at baseline were less likely to consume full-fat dairy, or more inclined to cut it out of their diet? We don't know, but the questions are all valid.

While dietary intake was assessed in the current study, the word “sugar” does not appear, not once, in the paper (I ran a full text search for the word). The word “refined,” as in “refined carbohydrate,” does not appear once either. And while the word carbohydrate appears, the analysis was not controlled for refined carbohydrate intake; it just appears in the methods, and in the bibliography.

Why does this matter? Well consider a study by Li and colleagues at Harvard, published in 2015 in the Journal of the American College of Cardiology. Roughly 120,000 people were followed for 30 years. Those who reduced their saturated fat intake over time were compared to those who did not, but not just wholesale; the investigators identified what foods were replacing the sources of saturated fat in diets. Their findings are absolutely critical to making sense of this entire topic.

When saturated fat calories were replaced by trans-fat calories (e.g., replacing butter with stick margarine), heart disease risk, as we would expect, went up. When saturated fat calories were replaced by refined carbohydrate and sugar, rates of heart disease remained the same. This, obviously, does not mean that saturated fat is suddenly “good” for us, as so many have suggested. Rather, it means there is more than one way to eat badly (and, alas, the American public seems committed to exploring them all). Refined carbohydrate and sugar, and saturated fat from the usual sources, appear to have roughly comparable, and comparably bad, effects on overall health.

When, however, saturated fat calories were replaced by whole grain calories, rates of heart disease declined significantly. When saturated fat calories were replaced by unsaturated fat calories, from nuts, seeds, olives, avocados, and to a lesser extent fish and seafood, rates of heart disease declined significantly. These findings are entirely consistent with a vast aggregation of evidence suggesting that nuts, seeds, olives/olive oil, avocado, along with vegetables, fruits, beans, and lentils, and fish/seafood are consistently associated with better overall health and reduced risk of all chronic diseases and premature death.

Why weren't vegetables mentioned in the Li study as replacement calories? Probably because they tend to be so low in calories. Why weren't beans and lentils mentioned? Presumably because shifts in intake in this category were small, which is too bad. Replacing animal sources of protein with these foods is another way to reduce risk of all major chronic disease and premature death significantly, while doing the planet a big favor into the bargain.

So, this new study, like the one before, suggests that consuming a bit more, as compared to a bit less, full fat dairy is associated with no meaningful change in overall disease and death risk (leaving aside the subtle variations up and down). And, since those in the current study consumed low-fat dairy along with full fat dairy, that association can be expanded to: those consuming a bit more dairy have no meaningful difference in overall disease and death risk than those consuming a bit less.

But what if those consuming less dairy are drinking more soda instead of drinking milk (the word “soda” also does not appear in the new paper)? What if they are eating potato chips or French fries, while their counterparts are eating Greek yogurt? Based on overall population dietary patterns, this is not at all unlikely. Then, what the current study suggests is: dairy in general, and full fat dairy in particular, is neither better nor worse for overall health than soda and French fries.

In other words, the data in the current study could have supported headlines proclaiming: “full fat dairy apparently no better than soda, chips for health” and potentially have been just as consistent with the findings as the headlines we did get. We have seen this exact phenomenon before.

The reliable conclusions about dairy fat are thus: (1) it may be neither worse nor better than some other stuff people tend to eat and drink instead; and (2) that other stuff may be bad for us.

We know, reliably, that extra virgin olive oil, a mainstay in the Mediterranean diet, is not merely “possibly devoid of risk,” but actually and impressively good for us. We know that the same is true of unsaturated fats overall from nuts and seeds, olives and avocado, fish and seafood. We know the same is true of vegetables, fruits, whole grains, beans and legumes, too. And to that we may append: dairy, whether full fat or otherwise, may not affect risk very much against the backdrop of the prevailing American diet.

That last point is crucial. The prevailing American diet is terrible. Eating yogurt is apt to displace potato chips; cheese might substitute for cheese doodles; milk may be an alternative to Coca Cola. What would adding full fat dairy do to a diet that is optimal to begin with? The new study can tell us nothing about that, but the weight of prior evidence is certainly precautionary.

What, then, to do with all of this? Eat well overall. If you do, dairy will occupy a small space in your diet, or no space at all. In that context, the fat content of the dairy is maybe moot (or maybe not). Speaking of “moot,” there are reasons to limit dairy intake unrelated to health effects, including the treatment of cows, and as Dr. Hamblin concludes in The Atlantic, the significant implications for environmental impact of diet.

The new study may help you avoid freaking out about a bit of dairy fat, but it provides no rationale for expressly adding any to your diet. As for hyperbolic nutrition headlines: they are reliably bad for your health. Never swallow those.