Skip to content

We Don’t Know Anything about Diet and Exercise.

7 July 2016

Let me amend the title a bit: we know lots about diet and exercise. We know that eating fewer calories and getting more exercise will help us keep from becoming obese. We know that eating lots of vegetables and fruits provides a lot of nutrition – especially micronutrients (vitamins and minerals) – for not too many calories. We know that eating lean animal proteins helps us build muscle. We know that we need fats in our diets but that some fats are better for us than others.

What we don’t know is anything very specific about diet and exercise. We don’t know much about specific foods (eggs and coffee for example, have dueling studies every damn hour it seems like) except insofar as they contain specific macronutrients (protein, carbohydrates, fats) and micronutrients whose functions we (basically) understand. We don’t know how diet and exercise interact to influence weight beyond basic thermodynamic calculations.

There are at least three fundamental reasons for this. The first one is the human body. Calculating calorie intake and outflow are essentially impossible, in a general sense (meaning, no one equation works for a large group of people). If I eat exactly one hundred grams of peanuts, and you eat exactly one hundred grams of peanuts, and the peanuts are all perfect clones of one another, you and I will still absorb different numbers of calories from the handful. This is due to many things, including how much we chew, how acidic our gastric juices are, and how fast our peristalsis goes.

The same is true about exercise. Two people of similar fitness levels running the same distance in the same time will burn different numbers of calories. Maybe only slightly, but tiny daily differences in average calorie consumption/burn can be the difference between losing ten pounds in a year, and gaining ten pounds in a year. There are about 3900 calories in a pound of pure fat. So the difference between gaining ten pounds and losing ten pounds in a year is only about 200 calories a day.

Now, add to how variable the body is (meaning, your body might not have the same governing thermodynamic equations as mine), just how plastic the body is. Your body and my body not only start out different, they change differently in response to exercise and diet. Some people can become obese and not become diabetic. Others will become diabetic despite not ever being obese. Some people will easily lose weight by increasing exercise with an unchanging diet. Others cannot lose weight by exercising and must restrict caloric intake.

As we exercise, we get in better shape, but we also become more efficient: our bodies get better at using the calories they have, so to keep losing weight (if that’s your goal) as you exercise, eventually you will have to eat less too. Our bodies are very well made to accumulate and store energy for times when not much food is available. Some are better than others, and we can influence how good they are at it by training it. Overeating “trains” the body just as much as exercise does, and our bodies adapt to the environment they find themselves in.

These two inter-related problems  – body variability and body plasticity – confound studies of both exercise and diet. They make the data extremely noisy. What this means is that even if the studies are correct with regard to general trends (like say, diet being more effective than exercise at reducing weight in a cohort of obese individuals), they may have no relevance to whether diet or exercise is more effective for you personally.

Think of a similar problem: suppose a study showed that a group of 10,000 children taught algebra in third grade were 10% more likely to get college degrees than a group of 10,000 children who got ordinary third grade math. This might well be absolutely true at the cohort level. But it doesn’t mean that teaching those kids algebra did anything. It could be that it simply allowed us to identify that some kids were more academically capable than we thought they were and invest more resources in them. And it really doesn’t mean that teaching your child algebra in third grade is going to make them 10% more likely get a college degree. Kids are all different. You could be doing harm to your kid, by forcing something on them too early, making them think that academic achievement is too hard.

Cohort studies don’t really tell you anything about individuals unless the effect size of the intervention is much greater than the noise in the data. So for a lot of medical interventions, we can be very confident that things that work at the cohort level work at the individual level. Especially things like surgical interventions. People with burst appendices who have a surgical repair live a lot longer than those who don’t. Like, many years compared with a few days. But for lots of things, all we can really say is that “when lots of people like you take this medicine, they seem to live longer/healthier/happier than the people like you who don’t take this medicine.”

The next, and probably bigger, problem with these longitudinal studies of diet and exercise are data reliability. The vast majority of studies – and all studies that last longer than a few weeks – are based on self-report of food consumption and exercise. They additionally require self-report of times of onset of other conditions that confound the studies (diabetes, heart disease, cancer, etc.). Self report is notoriously unreliable. I’ve worked with self-report data when I was doing diabetes research. It’s ghastly.

People do not weigh, measure, and record everything they eat. And even if they did, variations in the foods would mean that that wasn’t perfectly accurate. And even if they were, our bodies are so variable and plastic that it wouldn’t be very accurate. But nevermind all that because in these studies, we don’t actually know what people ate. We don’t actually know how much they exercised.

The only thing we know for sure is that the data is wrong. People exaggerate the healthy foods they eat. They exaggerate their exercise. Even when they’re just wrong, not lying. But people lie. Lots. Usually, I think, people lie to their doctors in ways that they think makes them look healthier. But they may not be correct about that. They may be making themselves look less healthy. We can’t ever know what these lies are, or by how much. We can’t know how much the honest people (not as many of us as we’d like) misestimate their consumption and exercise.

These problems: individual variability, body plasticity, cohort noise, data reliability, combine to undermine the credibility of all studies of diet and exercise. The general trends identified in these studies are probably usually accurate. But specific numbers cannot be reliably applied to you personally.

Here’s what we know about diet and exercise. Don’t eat too much. Get plenty of fruits and vegetables. Lean proteins. Unsaturated fats. And keep moving. Moderate exercise is good. More is better. Vigorous exercise is really good, but don’t go crazy because you can get injured. Especially don’t just jump right in to a lot of vigorous exercise if you’re in poor cardiovascular condition. Ramp up. Sudden changes can be lethal. Talk to your doctor about your diet and exercise. Find what works for you.

Use your diet and your exercise as tools to improve your life, and meet your goals. If that’s weightloss, great! If that’s being able to walk around the park without chest pain, great! If that’s finishing an Ironman in less than 10 hours, great! If that’s maintaining where you are, great! And you know what: you don’t have to have any diet and exercise goals. And anyone who judges you for that can go fuck themselves.

We know at the population level that people who are fitter and have basically good diets live longer and have higher reported quality of life. If you want those things, working toward them is not complicated. But it can be difficult. And it can be expensive. But there are no guarantees. Sometimes fit people with great diets die young. Sometimes obese, sedentary smokers live to be 100, happy as clams.

Nothing we learn about diet and exercise is going to be a magic pill for being fit and sexy and long-lived without discipline and effort. Ever. So make your decisions, and then work the best towards them that your circumstances allow. And don’t worry about the numbers* bullshit.



*Meaning, don’t worry about numbers like “Eating bananas twice a week makes you 15% less likely to have a stroke!” Absolutely do pay attention to numbers like your blood pressure and glucose and cholesterol. Those are your numbers, and they matter. Talk to your doctor.

3 Comments leave one →
  1. 7 July 2016 09:26

    A little over a year ago, I was presented with an A1c of 5.7. One tick above “normal.” So I am considered pre-diabetic, which I consider a shameful condition. And the latest research shows a link between diabesity and brain disorders such as dementia. No matter what I have done in the last year, it does not change. I had it tested again a month ago. 5.7. No matter what I do. My doc has told me that it just may not change. I have changed my diet drastically. I have lost weight and I have gained it back and lost it again. It doesn’t change.

    I just don’t believe in the calories in and calories out philosophy anymore.

    • 7 July 2016 09:31

      5.7 is a fine A1c, especially at your age. At that level, you are not going to have the complications associated with high blood sugars. Mine is 5.5-5.7 usually.

      Diabetes only results in damage – if you keep that A1c down below about 6.5 – incredibly slowly. Like, multiple decades. The complications tend to arise rapidly if your A1c rises about about 7.5. With diet and exercise, you’ll have little trouble maintaining that.

      Prediabetes onset at retirement age for an active person who keeps a decent diet? You have little to worry about in my opinion.

      (Please confirm everything with an endocrinologist…)

  2. Aimee permalink
    7 July 2016 14:45

    Great post. Thanks for breaking it down.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: