Ice Cream is a Superfood?

[TLDR: Either nutrition "science" is an exercise in confirmation bias, or ice cream is good for you. If you'd like to go deeper and discover what really is good nutrition for your individual body, feel free to book a call.]

A few months ago The Atlantic ran a story on one of the most confusing results in nutrition science: people who consume more ice cream have better health outcomes.

The result is odd for those who take nutrition "science" seriously. Even for critics of mainstream nutritional epidemiology (like myself) it seems an unlikely outcome. The ice cream eaters are not likely to be people engaging in otherwise healthy behaviours. So the standard complaint - were people eating these foods and that's what made them healthy, or were they healthy people before who were eating foods that society associates with health? -  doesn’t apply. No one thinks ice cream is a health food. So what’s really going on here?

Before we explore the possibilities, let’s address the obvious. What the article unintentionally demonstrates is how biased nutrition researchers can be. They design a study to show that eating low-fat yogurt is healthy. When the study (that they already designed with that bias in mind) doesn’t show that, they try to bury the results. If the same study shows that people with diabetes who eat ice cream are healthier than people with diabetes who don’t eat ice cream, they bury that result as well. They have to do this, because such counter-intuitive results call into question the whole discipline of nutritional epidemiology. Of course it’s possible that people were lying about what they ate, or that people didn’t have complete recall over what they’ve eaten in the past year on one of these complicated FFQs (food frequency questionnaires), or that what they ate in the years after they filled out the FFQ changed from one year to the next.

(Most studies in nutritional epidemiology collect data in year 1 of the study, and then use that data to make conclusions about health outcomes that may have happened between years 5-7. For example, if I said I eat hamburgers once a week in 1996 and then I suffered a heart attack in 2002, the study might conclude that weekly hamburgers increase the risk of a heart attack. It would show this result even if the person in question went vegan in 1999.)

But accepting that there is something wrong with this methodology would call into question not only the ice cream result, but also every nutrition-related headline that has ever appeared in The New York Times (or The Atlantic). Everything you've ever read about "super foods" like broccoli, blueberries or kale comes from studies that suffer from these methodological flaws.

So, the discipline has its problems. But it does serve a purpose. Observational epidemiological studies shouldn’t be used the way they’re advertised - they can’t tell us what's healthy. But they can present us with a hypothesis for further testing. Based on the results presented in the article, there are two plausible hypotheses to be tested: 1) Yogurt is good for weight loss and blood sugar regulation, and 2) ice cream is good for weight loss and blood sugar regulation. If nutritional science is doing its job, we would expect to see both of those hypotheses tested in the peer reviewed literature by now. Since the ice cream signal is stronger, we would expect the ice cream hypothesis to have been tested more than the yogurt hypothesis. Has it been?

It probably won’t shock you to learn that I have searched high and low and have yet to find a single study testing the hypothesis that increased ice cream consumption promotes health, weight loss or blood sugar control. But what was surprising was just how many papers I could find testing the hypothesis that consuming yogurt is good for these things. This is despite the fact that the results from the yogurt trials are, well, abysmal. One 2014 meta-analysis states “the results of these studies are equivocal”. That’s science speak for “the results are an unmitigated disaster” and about the harshest thing that can be said in a meta-analysis of this kind. And indeed of the papers examined in that study, the association between yogurt consumption and weight loss was either weak or non-existent.

As to why no one’s looked into the hypothesis that eating ice cream promotes weight loss, I’ll leave the answer to your imagination. But for the purposes of playing devil’s advocate, let’s imagine that it is a plausible hypothesis. It’s certainly as plausible as the yogurt hypothesis, which seems to get tested over and over again. Is there some mechanism that might explain how it could be true?

Why would it be the case that what is essentially a combination of sugar and saturated fat might be associated with better health outcomes? These are probably the two most demonised ingredients in the universe of food.

It’s pretty hard to argue that dietary sugar is what would be making ice cream healthy. If that were the case then other sugary foods would also show up in the “unexplainable epidemiological results” pile. But, with the exception of pure butter, there is no food that contains as much saturated fat per gram as ice cream. Could it be that the saturated fat is what’s protecting ice cream eaters from the negative effects of sugar?

The answer to that question is “no one knows.” This is not a hypothesis that’s been tested in humans. What does seem pretty clear is that saturated fat has been unfairly demonised in the medical literature and is not universally associated with worse health outcomes. There is some observational data - including the famous “French paradox” - indicating that increased saturated fat intake might be associated with better health outcomes. There is also animal data showing that increased consumption of saturated fat (particularly stearic acid) protects mice from getting obese.

Does any of this mean that ice cream is good for you? Maybe. Like anything else, the answer is contextual. If your choice is between a scoop of ice cream and a donut, then yes, absolutely, the ice cream is the better choice. If your choice is between an ice cream and a sugar-free ice cream, then the sugar-free ice cream is the way to go (click here for a recent post on non-nutritive sweeteners). If your choice is between ice-cream and skipping dessert, then skip the dessert.

These recommendations are based on experience helping clients get healthy and lose weight rather than the peer reviewed literature (because there is no such literature when it comes to ice cream). 

That said, when it comes to actually knowing what foods are healthy for humans, the safe bet is to ignore everything you’ve ever read about a food being good or bad for you and ask a different question. What foods do you think humans evolved to eat? The answer to that from what we can currently make out from the data is that early humans were getting at least 70% of their food from meat and other animal products. They certainly had very limited access to sugar and no access to processed foods including seed oils. If you eat in a way that’s consistent with evolutionary appropriate human diets, chances are that you’ll do a lot better than by following any advice you read in The Atlantic or even in the journals of nutrition science.