Dear Science,
So my friends and I were out at a bar having a debate over what would be the worst imaginable lunch to have every day. Options thrown out included corn-syrup Coke with a large order of fries, doughnuts and a whole-milk latte, and fried mozzarella sticks. Some of this seems like common sense. Fried foods, sugary foods, and junk foods should be terrible for you and make you fat if eaten every day. We couldn’t figure out if anyone had bothered to actually study this and figure out what are the worst foods to eat every day. Help us out, Science! Which foods are most likely to make you fat if you eat them all the time?
Formulating a Fat Future
Science has been holding on to your question for a while, waiting for a good, scientifically rigorous answer to your question, which is surprisingly tricky. It’s not like scientists can go out into the world and force people to eat one food or another for years on end—randomizing some to foods we suspect are healthy and others to unhealthy foods. But finally a study—recently published in the New England Journal of Medicine by Dariush Mozaffarian and others from Harvard—figured out a way to answer your question: Which foods are most likely to cause one to gain weight?
This study group reanalyzed data from three huge surveys (involving 20 to 50,000 people each) that followed the participants over many years—tracking things like their weight and their habits (smoking, exercise, alcohol consumption, and so on), as well as what they ate (as reported by food diaries). All three were observational studies, where the participants reported what they ate, whether they smoked, how much they exercised. The Harvard group then took this data and ran it through a fancy statistical analysis to figure out which foods were associated with weight gain or loss.
Most of the results were as any reasonable person would suspect. Eating lots of fruits, vegetables, and nuts was associated with weight loss. Whole grain consumption was associated with weight loss, refined grain consumption with weight gain. The more fried foods one ate, the fatter one became over time. A few subtle details popped out, however. Eating potatoes—french fried, baked, or made into chips—was tremendously associated with gaining weight, more than just about any other food. Sugary juice was less associated with weight gain than sugary sodas. Processed meats and red meats were associated with weight gain. Low-fat dairy foods (like yogurt) were associated with weight loss, and even full-fat dairy foods (aside from butter), like whole milk and cheeses, were not associated with either weight gain or loss.
Healthily Yours,
Science
Send your science questions to
dearscience@thestranger.com.
Science, I am a little surprised that you have answered a question about causation with research reporting correlation. In particular, the meta-study did not correct for the relative wealth of the participants; I would think Science would at least briefly highlight the fact that what is being reported is association.
Second, Science seems to have fallen into the trap of equating “bad for you” with “fattening,” which is perhaps more understandable, as that’s the default assumption in the wider culture Science inhabits.
It is quite possible to eat a non-fattening diet that nonetheless causes something like, say, very high mercury levels, which is more “bad for you” than an extra ten pounds (and far, far more bad for your kids). This example is not hypothetical; it describes at least one population in Japan.
Food can do worse than make you fat. Toxins, parasites, and the many forms of malnutrition all tend to do a lot more damage than fried cheese.
With all the talk about food these days (Hello, Michael Pollan) you’d think this would be a no-brainer. Haven’t you seen Super Size Me?
@1: But the question asked was “Which foods are most likely to make you fat if you eat them all the time?”
So…Science answered, here is this study, and here are the foods it found most likely to make you fat.
@4, I respectfully disagree.
If you read the entire question, rather than the last line, you will find the words “what are the worst foods to eat every day?” The context given is a group of friends arguing about which foods are “worst for you”, and given that context, and the fact that the word “fattening” is not mentioned until the very end, it is entirely appropriate to note the misleading cultural equation of “bad for you” with weight gain.
And as I said, Science did not report which foods are most likely to make you fat, Science reported which foods correlate most closely with weight gain over time– which is not the same thing at all, particularly if there is no correction for the wealth or income level of the subjects.
An alternate explanation might go like this:
Suppose people who face periodic food scarcity tend to have, perhaps counter-intuitively, a higher average calorie intake over time. The most common cause of periodic food scarcity is poverty. Poorer people tend to buy foods that are cheaper per calorie, less perishable, and sold closer to their homes. Thus fewer fresh fruits, and more potatoes, potato chips, and fast-food french fries.
In this model, the particular foods consumed are no more fattening than any other food; they’re just the ones that happen to be most accessible to people who tend to consume more calories in excess of what they’d need to maintain their present weight and activity level.
Robotslave, perhaps the message from the question was that the asker WAS making the broader cultural assumption you blamed on the columnist. S/he said “worst” and later specified fattening. The answer is therefore fine.
Yes, these are just associations, and the columinist did carefully explain that we can’t do randomized controlled trials of food. However, this entire country is going obese, and very few fat people are actually food insecure. So I have to toss out your theory that periodic starvation is what’s making people fat and it’s merely a coincidence that the disgusting junk food we all know makes you fat also happened to turn out in the best data available to be associated with weight gain.
yonush18, where do you get your data suggesting that very few fat people in the US have experienced periodic food insecurity?
It’s interesting that you characterize the foods associated with weight gain as “disgusting junk food” when even in the meta-study cited, only specific types of packaged food were associated with weight gain; notice the notorious “junk food” ice cream is neither associated with weight gain nor weight loss. And of course not all of the foods associated with weight gain were packaged, ready-to-eat foods, either; e.g. baked potatoes, red meat, all-purpose flour.
I offered an alternate hypothesis to explain the findings of the study, not a test of that hypothesis. But that’s science– if you’ve got more than one viable hypothesis, you need to actually test them, rather than simply reject one or the other out of hand because they don’t conform to your biases (and by “bias” I refer to whatever prompted you to use the word “disgusting”, and no more– I don’t pretend to know what gives rise to your personal emotional responses to food).
I would suggest the following test of the hypothesis that the foods associated with higher weight gain in the meta-study actually caused that weight gain: Do studies from other cultures show the very same foods to be associated with higher weight gain?
Please note that the title of Science’s column this week is not “Which foods are most fattening?” but rather “Which are the Worst Foods to Eat?”
@3: I agree. It does sound like a no-brainer. I saw Super Size Me, too, and can’t believe what the fast-food industry constitutes as “Medium Sized” anymore.
Does anyone know if the highly publicized Heart Attack Cafe somewhere down in Arizona is still in business? The owner is supposedly a former nutritionist who soaks everything in lard before serving it, and diners weighing OVER 350 lbs. EAT FREE!!
Although, when it comes to spuds, I have to disagree slightly on baked potatoes being classified as fattening. It’s what goes ON them that can clog arteries and put on extra pounds (i.e.: butter, sour cream, shredded cheese, bacon bits, etc.).
@8: Point well taken.
And NO, I have no desire whatsoever to eat at the Heart Attack Cafe!
@9: Do you have any particular reasoning for what you say about baked potatoes? The study seems to clearly find more of a correlation of obesity with potatoes (including baked) than with cheese.
@1 the question “Which foods are most likely to make you fat if you eat them all the time?” is clearly a question regarding a CAUSAL relationship between food and fatness. Focus on the word “MAKE”. When “A” MAKES “B”, that is causal.
The problem with the study described by “Doctor” Golob is that is is an associative/correlative study…impossible to imply causation. So, i am afraid that roboslave is absolutely right.
Does that mean that you should ignore the study? Absolutely not! Associative studies have value….for example the link between lung cancer and cigarette smoking was associative for quite some time, and it turns out that not smoking is very good advice.
However, if Science (by this I mean the subject, not the columnist) is to be effective, there should be some discipline in what conclusions can be drawn from a study.
> So, i am afraid that roboslave is absolutely right.
No need to fear that at all.
Here is the abstract of the study: http://www.nejm.org/doi/full/10.1056/NEJ…
This is most likely some sort of combined longitudinal study with a random coefficient regression analysis.
As such, and even if it is not as sophisticated as that and is more of a straightforward linear regression analysis, it does in fact go a long way towards separating out the issues that robotslave raises.
I am not going to go into detail how or why that happens in such a study, as it is not statistics 101, but if you really want to know, I am sure google can help.
> The problem with the study described by “Doctor” Golob is that is is an associative/correlative study…impossible to imply causation.
Again, not really. Especially for a random coefficient analysis, the results CAN be predictive of other data outside the study. IOW, if you are statistically within the studies parameters, your diet and lifestyle will be predictive, to a known statistical certainty, of the weight you will gain or lose over time.
If you are worried that this study doesn’t specify the biochemical method by which eating potatoes is associated with weight gain, or similarly why your cells cause you to gain weight if you are an ex-smoker, well, those are for other studies. But for most people, it is fine to say that eating a certain amount of potatoes *causes* a certain amount of weight gain over time, independent of other dietary or lifestyle choices, and this is entirely predictable within a certain distribution over a large population.
@portervillenerd – actually, most legitimate scientists (such as myself) do not allow the implication of causation (which you, incorrectly, seem to use interchangeably with “predictive) without the biochemical methods that you so casually disregard. In order to show causation, in real science, you actually need to know the mechanism by which it is caused. The “how” question is critical, and the methods in the paper do not allow that question to be answered with anything other than conjecture.