It’s that day of the week again, y’all! (Sorry, all this talk about saturated fat and butter has me channeling Paula Deen and her southern accent.) It’s Tuesday, and you know what that means: a new post that drops a couple more knowledge bombs on the war zone that is the last sixty years of official government and medical community recommendations about dietary fats. Last week I introduced you to a paper whose author—a PhD professor of chemistry and biochemistry—concluded that maybe, perhaps, the fear mongering about saturated fat has been misguided, and heart disease might be caused by things other than butter, bacon, red meat, cheese, and similar delicious morsels.
I promised I would devote a couple of Mardi Gras/Fat Tuesday posts to dissecting the paper in more detail, so here goes.
Today’s focus is the issue of confounding. Specifically, as it relates to the paper we’re talking about, we’ll focus on carbohydrate consumption as a confounding factor when trying to implicate saturated fat—or any fat, for that matter—in heart disease. Regarding scientific studies, a confounding factor is something that skews or alters your results. Well, it doesn’t necessarily alter the results; what it does is make it difficult to identify the actual cause of the results. For example, if you eat a banana and a bunch of strawberries and then break out in hives, we can’t automatically conclude you’re allergic to bananas, because it could be the strawberries, and vice versa. We can’t know for sure which it is unless we isolate them and perform the experiments independently: one day we give you a banana and see what happens. The next day, we give you strawberries and see what happens. And then, of course, there’s a third option to consider: maybe we give you a banana and you have no reaction, and we give you strawberries and you have no reaction, so maybe it's the combination of a banana and strawberries that you’re allergic to. Pretty complicated for something that seems so straightforward, huh? Imagine how tough it gets when we talk about an entire subset of macronutrients, like saturated fat.
I gave some pretty good examples of confounding last week:
- Whenever people eat more ice cream, there are more shark attacks. Therefore, eating ice cream causes shark attacks. Right? No. There’s a third factor that hasn’t been considered here: people generally eat more ice cream in the summertime, and they also tend to go to the beach and swim in the ocean more in summertime. The increase in shark attacks has nothing to do with ice cream.
- A study on cancer of the mouth: researchers follow two groups of people, separated by height, for twenty years and see who ends up getting oral cancer. After those twenty years, more people in the tall group had cancer than in the short group, so they conclude that being tall is a risk factor for oral cancer. But what they didn’t account for was the fact that more people in the tall group used chewing tobacco than in the short group. The cancer is completely unrelated to height, and it was this third factor—the confounder—that was responsible for the outcome.
Researchers (responsible ones, anyway) are aware of confounding and usually try to correct for it. This is why we read about “age-matched controls,” or “controlling for age/weight/ethnicity/smoking status/etc.,” in scientific studies. They try to level the playing field by accounting for all the confounding factors they can think of. Great! That’s exactly what they should do. The thing is, they can’t possibly control for everything, particularly when there are variables they don’t even know about. I am no fan of Donald Rumsfeld, but the man was on to something when he talked about “known unknowns and unknown unknowns.” That is, there are things we know interfere with (or confound) study outcomes but can do nothing about, and there are things we don’t know about that will complicate outcomes. This is why we should always take sensationalist health headlines with a grain of salt. They’re not designed to educate and inform us; they’re designed to attract more viewers, readers, subscribers, more Facebook “likes” and more “shares.” (Speaking of inflammatory news flashes that should be taken with a grain of salt, don’t be scared about the salt. Not surprisingly, it isn’t the health-wrecker we’ve been led to believe.)
So when studies come out that seem to indicate a connection between a single dietary element and some chronic illness or another (like saturated fat and heart disease), we have to be very careful about believing the conclusions unquestioningly. It’s almost impossible to separate out every single factor and say conclusively that it was one of them that caused the observed outcome.
According to Dr. Lawrence:
“Over the years, data revealed that dietary saturated fatty acids are not associated with CAD (coronary artery disease) and other adverse health effects or at worst are weakly associated in some analyses when other contributing factors may be overlooked.”
And yet, the message we end up getting is:
Saturated Fat Causes Heart Disease!
Really? Are we able to definitively say it was the saturated fat and not, say, lack of exercise? Insufficient sleep? A low vitamin D level? Use of laundry detergent on days that end in Y? And most important, maybe some other factor in the diet that the researchers failed to control for, like consumption of sugar, wheat, trans fat, 100-calorie snack packs, fluoridated water, or a zillion other things that have nothing to do with saturated fat.
So what does all this really have to do with Dr. Lawrence’s paper? For starters:
“Human food preferences tend to favor foods with both fats and sugar, which complicates any attempts to correlate saturated fats with disease.”
Spot-on! Thanks, doc! I mentioned this briefly last week: very rarely do we eat saturated fats by themselves. We tend to prefer them coming to us on or in carbohydrate delivery vehicles: butter on toast, cheese on a sandwich (or in a burrito, wrapped in a huge flour tortilla), hamburger on a bun (and with fries), steak with a potato, etc. It’s almost impossible to completely separate saturated fat from any other dietary component and try to ascertain the effects of eating it. (Almost impossible in humans, that is. It’s easier to do with animal experiments, but then there’s the pesky issue of whether what happens in mice, worms, rabbits, monkeys, or any other species, can be extrapolated to humans.)
The way to isolate single dietary factors is to sequester your test subjects in a hospital’s metabolic ward or some other place where you can monitor every iota of food they ingest. You have to know exactly what they’re eating, and how much. Studies like this are extremely expensive, not to mention no fun for the subjects. (In a study to determine the effects of saturated fat intake on heart health, for example, it might sound good at first—at first—that you can eat all the butter, heavy cream, coconut oil, bacon, and cheese you want, but I bet that gets pretty old when that’s literally all you’re allowed to eat. No bread for that butter, no potato for your sour cream. And let’s not forget—butter, bacon, and all other foods that contain saturated fats also contain unsaturated fats, so even then, we couldn't hang our hats on only the saturated fat causing whatever outcome we'd see.)
So what do researchers do instead? Often they rely on “food recall questionnaires.” Basically, these are surveys where they ask respondents to tell them what they usually eat. Questions might go something like this: “Do you consume red meat more than three times a week?” Or, “How often do you use butter?” Or they’ll list a food and ask you to indicate how often you’ve eaten it over the last ten years—every day, twice a week, once a month, never, etc. Are you kidding me? I can’t remember what I had for lunch yesterday, let alone estimate (estimate!) the amounts of certain foods I’ve eaten in the last decade.
Plus, there’s an even bigger issue: people try to make themselves look good when they answer these things. They know someone’s going to look at their answers, so they’ll downplay some of the behaviors that they assume researchers would think are “bad,” and play up the ones they think will make them look better. (They’ll claim to eat less fat or sugar than they really do, drink less alcohol, claim to exercise more, etc.) Unless we do put people in a metabolic ward, the fact is, we have no idea what they’ve really eaten during any period of time.
Not surprisingly, these types of questionnaires are notoriously unreliable. The only people who seem to think they’re the bee’s knees are the researchers who continue to use them. Obviously, not all food recall survey questions are as silly as the ones I made up here. I’m sure the researchers do their best to make them as detailed and specific as possible in order to isolate the dietary factors that are causing whatever outcome they’re studying. My point is only that even in a best case scenario, these things are shady, and connecting the dots on what really causes what is harder than getting one of the Real Housewives of Fantasyland to shop at Kmart.
|This is a great way to do research!|
But back to the main issue: carbohydrate confounding. Even if researchers do think saturated fat causes heart disease, the most accurate conclusion they can give us is that intake of saturated fat is associated with heart disease when it’s included in a diet that also contains protein and carbohydrate. (Especially when U.S. government dietary guidelines tell us the vast majority of our calories should come from carbohydrate...y'know, all those "healthy whole grains.") So in the context of a mixed diet, how can we say definitively that it’s the saturated fat that causes heart disease? What if it’s the protein? What if it’s the carbohydrate? What if it’s living in Florida and wearing black knee socks with white dress shoes like a 76-year old man? Unless the study isolates saturated fat from every other aspect of the diet and shows that it’s not just “associated with” but actually causes heart disease, we simply can’t know whether it’s time to put down the butter and use grape jelly on our toast instead. (‘Cuz hey, maybe it’s the toast!)
The good doc points out:
“There is no reason to believe that replacing fat in the diet with carbohydrate at a constant caloric intake will improve the serum lipid proﬁle signiﬁcantly. Indeed, a low-fat, high-carbohydrate diet causes an increase in serum triglycerides and small, dense LDL particles, which are more strongly associated with CAD than serum total cholesterol or LDL-C.”
In case you can’t decipher the geek-speak, he’s saying a high carbohydrate diet causes negative changes in markers associated with heart disease risk. Yes, associated. He’s not saying high carb diets directly cause heart disease, even though, based on the conclusions in his paper, he most assuredly thinks they do. See how hard this is? Truly responsible nutrition scientists can almost never say “If A, then B.” The best we can do is something like, “In some people, under some conditions, and with lots of other stuff going on, if A, then sometimes B.” But a conclusion like that doesn’t make the six o’clock news, the Huffington Post, Dr. Oz’s Twitter feed, or wherever people get their health news these days. It’s a total snoozer, so instead of admitting that we honestly don’t know what’s going on, journalists seize upon attention-grabbing, black-and-white sounding tidbits that often have nothing to do with the more complicated, very murky, and sometimes ho-hum conclusions the research actually draws.
Okay. Let’s get real for a sec. I’m not saying that all studies are worthless, or that we can’t possibly draw any valid conclusions from nutrition research. All I’m saying is that we can’t take the headlines at face value. We have to know who did the research, how it was conducted, who paid for it (the outcome of a study—or at least the outcome that gets publicity—could be very different depending on who funded it), and of course, what the confounding variables were. Too often in nutrition science, correlation is taken to mean causation. They are not the same things. (Remember ice cream and sharks.)
I can’t resist giving you one more bit of geek-speak to get you thinking. This one comes from a paper I mentioned last week, one that was published in none other than the American Journal of Clinical Nutrition:
“Replacement of (saturated fat) with a higher carbohydrate intake, particularly refined carbohydrate, can exacerbate the atherogenic dyslipidemia associated with insulin resistance and obesity that includes increased triglycerides, small LDL particles, and reduced HDL cholesterol.”
Translation: when we cut back on saturated fat and eat more carbohydrate instead (just like they’ve been telling us to for fifty years), markers for heart disease get worse.
Down the line, I’ll get into the actual mechanisms whereby high carb intakes influence markers for heart disease, but before we get there, we’ve got a few more issues to look at in this paper. So tune in next Tuesday, and until then—unless I write about something unrelated before then—I’ll leave you with another line that hammers it home:
“The adverse health effects that have been associated with saturated fats in the past are most likely due to factors other than SFAs…This review calls for a rational reevaluation of existing dietary recommendations that focus on minimizing dietary SFAs, for which mechanisms for adverse health effects are lacking.”
Preach it, brothah Lawrence, preach it!