Misunderstanding the Data on Diet, Exercise and Mortality

It's been a while since we've covered a nutrition or exercise study on the channel, so today we're covering one that looks at both. It made quite a media splash, and as is so often the case, we were less than impressed. That's the topic of this week's healthcare trio. You can't outrun your fork, or so they say, which basically means that you can't eat a bunch of junk food, followed by a bunch of exercise, and assume the exercise will make up for the junk food. Variations of this phrase exploded in media headlines last week for coverage of a recently published study on the effects of diet and exercise on mortality risk. Essentially, the media coverage went something like this. According to new study, you need both exercise and a quality diet to live longer. This sounds good, and there are little kernels of truth buried in there, but coverage of the study was, as usual, less than appropriate, so we're here to dig a little bit deeper. To the research! The study used data from the UK Biobank, a very large biomedical database that gathered health information from around a half million individuals across the United Kingdom between 2007 and 2010. Specifically, the researchers took one-time self-report questionnaires about diet and exercise from this biobank and grouped people based on their answers, followed by examination of who later died and how. They reported that individuals who exercised more and ate better had the lowest overall risk of mortality and the lowest risk of mortality related to things like specific cancers and cardiovascular disease. Let's first note that the diet quality questionnaire took into account only three categories of food items, fruit and vegetables, fish, and red and processed meats. This was a blunt measurement, as one author put it, and didn't take into account other food categories like sweetened foods and drinks. This isn't necessarily a deal-breaker on its own, but if you want to understand the relationship between diet and health, getting a picture of the whole diet is a good first step. But speaking of getting the whole picture, let's talk about self-reports. This is a common complaint, but worth repeating. The likelihood of people correctly remembering what they've eaten in the past, let alone how many servings of it they ate, is not great. Our memory is worse the further back we try and remember, and the same goes for exercise. It's a tough task to remember that much past information correctly, so studies like these that rely entirely on participants' memories of what they've eaten and how much are hard to draw firm conclusions from. On top of that, diet and exercise patterns are not fixed throughout people's lifetimes. The particular snapshot captured by a one-time self-report questionnaire isn't likely to be an accurate representation across a long period. Remember also the sheer number of confounders in a data set like this one. People who exercise regularly and eat things like fish and lots of fresh produce tend to have other things in common as well. Things like education, money, and extra time, all of which have huge impacts on health. While these factors can be controlled for to some extent with statistics, there are so many variables that the data set often becomes too noisy to hope for picking up reliable signals. And beyond only taking into account those three food categories we mentioned, participants were coded into just three diet quality groups, low, medium, and high quality. High quality diets were characterized as having at least 4.5 cups of fruit and vegetables per day, at least two weekly servings of fish, and less than two weekly servings of processed meats and less than five weekly servings of red meat per week. I've said a hundred times that the human diet has so many components that it would be nearly impossible to pinpoint a single one as bad or good. In that same vein, distilling any particular human diet down into low, medium, or high quality based on three self-reported factors while specifically leaving other factors like sugar out of the equation just isn't going to cut it. And lastly, the authors report that the association between diet and mortality from all causes and cardiovascular disease was not even significant. Look, we aren't getting down on this particular set of researchers. We've got thousands upon thousands of studies examining diets and exercise and mortality and they all run into the same problems. We don't really need more of them and we definitely don't need more non-nuanced media coverage of them. Am I saying fruits and veggies and fish aren't good for you? No! Am I saying that exercise isn't good for you? Absolutely not! Am I saying that red meat isn't necessarily bad for you? Well, actually, yes, but you'll have to go watch that video for more detail. The point is that the evidence doesn't back up these weekly headlines we see about diet, exercise, and mortality studies. This one included. Hey, did you enjoy this episode? You might enjoy this previous episode on how nutrition studies are just terrible. We'd appreciate it if you'd like the video, subscribe to the channel down below, and consider going to patreon.com slash healthcare triage where you can help support the show and make it bigger and better. We'd especially like to thank our research associates, James Glasgow, Joe Sevitz, Edward Lilleholm, and Brian Nam, and of course, our surgeon admiral, Sam.