What “Experiments of Nature” Teach Us About Exercise
To optimize your health or supercharge your training, you sometimes need to look beyond the lab
Heading out the door? Read this article on the new Outside+ app available now on iOS devices for members! Download the app.
In this column, we believe in randomized, double-blinded, placebo-controlled trials. Others may start popping a new supplement because their friend said it made them feel good, strap on the latest wearable device because logic suggests the information it provides should be useful, or start doing Norwegian double-threshold workouts because Jakob Ingebrigtsen is really fast. But we wait for solid scientific evidence, preferably from multi-year studies with large sample sizes synthesized in meta-analyses.
In truth, though, this approach inevitably leaves a lot of questions unanswered. Good luck running a trial in which half the participants are randomized to run 100 miles a week for the next 20 years, while the other half do no exercise whatsoever. As a result, many of our key insights about how to optimize health and improve performance come from other types of sources, including what a new article in Comprehensive Physiology calls “experiments of nature.” Mayo Clinic physiologist Michael Joyner and his colleagues surveyed some of the most important natural experiments in the history of exercise science, offering an important corrective to the cult of the randomized trial.
Joyner and his colleagues start with a subtle distinction. “Experiments of nature,” in their terminology, involve people with rare genetic or acquired conditions that shed new light on how a particular physiological system works. “Natural experiments,” in contrast, involve observing large populations who have been exposed to some sort of environmental or behavioral stimulus.
Experiments of Nature
In 1951, a British doctor named Brian McArdle described a 30-year-old patient who, for his entire life, had suffered from muscle pain and weakness after just a minute or two of light exercise. Even chewing food left his jaw muscles exhausted. McArdle figured out that the patient had a rare condition—now known as McArdle’s disease—that meant he was unable to break down glycogen, the form in which carbohydrates are stored in the muscle and liver, into lactate.
If you’re trying to understand the long-disputed concept of the lactate threshold, people who don’t produce any lactate at all turn out to be very useful. The initial concept of an “anaerobic threshold,” formulated in the 1960s, was that when your muscles can’t get enough oxygen, levels of lactate start accumulating in your blood, which (through a few intermediate steps) causes you to start breathing more heavily. But McArdle’s patients also showed a sharp increase in breathing rate beyond a certain threshold, even though they didn’t produce any lactate at all, which forced scientists to rethink the concept.
That original 1982 study had only four subjects—the kind of study that people like me might be tempted to dismiss as too small to be meaningful. “However,” lead researcher James Hagberg later pointed out, “at the time these four patients accounted for 10 percent of the total world McArdle’s disease population described in the medical literature.” These were insights that were only possible through small experiments of nature.
The same is true for many other topics. Joyner and his colleagues mention Eero Mäntyranta, the Finnish cross-country skiing champion who had a rare genetic variant leading to sky-high hemoglobin levels (whose story I first read about in David Epstein’s book The Sports Gene), as well as various studies of identical twins that have altered our understanding of muscle fiber types and the links between exercise and body composition. Even studies of world-class athletes fall into this category: they are freaks of nature (and nurture, of course) whose off-the-charts physiology sheds new light on how the body works. But you can’t randomize people to become Olympic champions, and you can’t recruit 100 of them to show up at your lab for testing.
Joyner’s paradigmatic example of a natural experiment is the London transport workers study, which is often cited as the starting point for modern research on physical activity and health. British epidemiologist Jeremy Morris collected data on 31,000 transport workers, comparing two nominally similar groups: those who drove London’s double-decker buses, and those who spent their workdays going up and down the buses’ stairs collecting fares. The results, published in 1953, showed that conductors were roughly half as likely as drivers to die of heart disease, providing some of the first large-scale data to show that exercise is good for your health.
Another famous natural experiment is Harvard nutrition researcher Jean Mayer’s 1956 study of hundreds of workers at a jute processing plant in India. He divided the workers into 13 categories ranging from sedentary clerks and supervisors to cutters, carriers, and blacksmiths doing very heavy physical labor. Then he assessed their weight and their daily calorie intake.
The results, which are shown below, require some explanation. The classes of workers are arranged from most sedentary (on the left) to most active (on the right). Bodyweight is plotted on the left axis, calorie intake on the right axis. If you look only at the right hand side of the graph, everything makes sense. The more physical the job, the more calories the workers eat, and their weights are all roughly the same, suggesting that the increased calorie intake is balancing out the increased workload.
But on the left side of the graph, things get wonky. The most sedentary employees actually eat more than anyone else, and as a consequence they also weigh more than anyone else. Here’s the data:
One way to interpret this data is that your appetite will naturally drive you to eat as much as your body needs—but only above a certain threshold of physical activity. If you’re sedentary, a situation unknown for most of human evolutionary history, then the appetite mechanism no longer works properly. That’s consistent with the idea that the link between exercise and body weight isn’t so much a question of calories burned (you’re probably familiar with the depressing stats on how many miles you’d have to run to burn off, say, a bowl of ice cream), but instead helps ensure that your appetite matches your expenditures.
Of course, weight loss and exercise are still contentious topics, almost 70 years after Mayer’s jute study. His findings didn’t settle the question once and for all, and that’s true for most of the natural experiments Joyner and his colleagues discuss. But their broader point is that these types of non-standard experiments add to our knowledge in ways that often wouldn’t otherwise be possible to test, and help generate hypotheses for subsequent lab experiments.
The value of considering different types of evidence may seem obvious, but the motivation for the paper was the frustration Joyner and others experienced trying to deploy convalescent plasma (antibody-rich blood from recovered patients) during the COVID pandemic. They ran into barriers with the National Institutes of Health’s treatment guidelines, which didn’t endorse its use. The dispute revolved in part around the NIH’s reliance on data from large clinical trials versus the data from smaller “experiments of nature” in patients with rare conditions that made them unable to make their own antibodies.
Joyner’s criticism of the NIH’s “bureaucratic rope-a-dope” got him suspended and threatened with firing by the Mayo Clinic (“Your use of idiomatic language has been problematic and reflects poorly on Mayo Clinic’s brand and reputation,” his boss wrote in the reprimand letter). So he’s now making his case in more academic language in the pages of Comprehensive Physiology—and it’s a message that’s relevant to anyone who’s trying to optimize their training or improve their health. Of course, I’m still a believer in clinical trials. If you travel too far down the “experiments of nature” road, you wind up concluding that, say, PowerBalance bracelets really did make Shaquille O’Neal a better basketball player. But you should evaluate each piece of evidence on its own merits, not merely on the category it falls into. Good science, it turns out, is an art.
For more Sweat Science, join me on Twitter and Facebook, sign up for the email newsletter, and check out my book Endure: Mind, Body, and the Curiously Elastic Limits of Human Performance.