Get access to everything we publish when you sign up for Outside+.
This article contains a discussion of topics related to fueling and body composition that may be triggering for some readers. You are amazing and you are enough, as you are, always. If you or a friend are struggling with mental health or disordered eating, you can call or text the National Eating Disorder Hotline.
Let’s start by acknowledging the elephant in the room. Training for performance at the edge of human capabilities is extremely complicated. Different things work for everyone, and that also applies to approaches with nutrition and body weight. While I am a fan of throwing out the scale entirely, other individuals may take their own approaches and have long-term success.
The keyword there is “individuals.” Individuals can do what works for them, ideally while working with an expert nutritionist. But when it comes to programs like high schools, colleges, and post-collegiate teams? There is absolutely no place in this sport for program-wide bodyweight and bodyfat monitoring that dictates subsequent interventions to hit arbitrarily low numbers. For performance, it’s questionable at best and disastrous at worst.
And for health, it’s dangerous for everyone.
University of Oregon Allegations
On Monday, October 25, journalist Ken Goe published an article in The Oregonian that interviewed six women track athletes at the University of Oregon who detailed a pattern of bodyweight and bodyfat monitoring that made them feel “devalued as individuals and at risk for eating disorders.” At least three times per year, athletes underwent DEXA scans, used to measure bone density and body composition, among other things. Allegedly, athletes that did not meet the coaching staff’s standards were told to add cross training volume and faced body judgment as well.
The coach has a quote in the article that seems to confirm at least some of the reporting. “When we get the numbers from our DEXA scans, we have an Excel spreadsheet that we can plug the numbers into, hit a button and it gives us a starting value for a training program.”
Given that DEXA scans measure bone density and body composition, one possible reading of that quote is that athletes at greater risk for bone stress injuries were put on different training schedules. But based on the other sources cited in the article, it seems likely that the coach is admitting to guiding training at least in part based on minor variations in body composition in a team-wide way. And all of us–coaches, athletes, nutritionists, doctors, fans–must step up against this sort of broadly-applied standard. Lives are at stake, and we can’t just stand on the sidelines. The performance science is unequivocal: eat enough always, and find your strong.
There are two major overarching problems with the type of monitoring and intervention that happened at Oregon. The first is bad data analysis and the second is simply bad data.
Bad Data Analysis: Performance and Health
Running training at the limits of human performance exists in the blurry space between science and art. On the science end, we start with baseline physiological attributes (i.e. genetics) and then introduce interventions (i.e. training) that produce certain outcomes (in this case at least, track PRs). Art is usually reserved for things like psychology, unique workout designs, or self-belief. But there’s an important truth in exercise physiology–the science itself acts more like an art if we are using good data analysis.
Adaptation from baseline to intervention to outcome is weakly understood. Often, even for world champions, we’re only partially sure why something works. Given that uncertainty, it’s tempting to try to quantify everything and solve for the outcomes we desire. For the Oregon staff, it seems that body composition was one of those variables they tried to quantify. I imagine they even had helpful charts–this Olympian had these attributes, this NCAA champion saw this progress–guiding their decisions with those DEXA scans and excel sheets. The problem is that body composition is not the independent variable in almost any case. Changing body composition isn’t what drives performance variation, it’s a dependent variable responding to dozens of other things, from training to nutrition to stress to genetics. Sometimes, individual health interventions may be needed, but that is a medical question, not a coaching one.
So yes, I imagine that the Oregon models validated these training decisions. But there is a fundamental flaw. By interpolating from outliers and focusing on dependent variables that are easier to measure (rather than acknowledging individual variability and the uncertainty of adaptation), the model tried to work backwards, like using a wood chipper to turn mulch into a log.
Bad data analysis makes bad decisions. And when those bad decisions impact how athletes think about their bodies, lives can be harmed. Here is the toughest problem, though, and why every single one of us needs to step up. There are people out there who will read about the DEXA scans and retort: “that’s just what it takes to win 14 NCAA Championships since 2012.” As someone tweeted at me, “College athletics is not always healthy.”
Bullshit. This is what the good data analysis on running performance actually says: long-term top performance is almost always driven from a place of health. And that’s where bad data risks mucking up the picture.
long-term top performance is almost always driven from a place of health. And that’s where bad data risks mucking up the picture.
RELATED: The Importance Of Eating Enough Food
Bad Data: Short-Term and Long-Term Interventions
Throw a dozen eggs at the wall, and one might not break, but that doesn’t mean that the broken eggs didn’t try hard enough, or ran into the wall the wrong way. It just means they’re eggs, and we should probably avoid throwing them against the wall.
All too often, though, the metric used to guide decisions is the number of wall-proof eggs that a coach has ever had in their program. Quantity of NCAA champions is a horrific way to determine a coach’s value, just as the number of Western States champions would be. It’s a bad data problem for two main reasons: the misinterpretation of short-term incentive structures and the ignorance of long-term physiological issues.
Short-term data problems
Short-term, body composition interventions may reveal enough unbroken eggs in a program to have something approximating “success,” at least when evaluated by trophies on a school’s mantel. With body composition, for example, some women can have a healthy menstrual cycle at relatively low body fat levels. Perhaps an outlier or two can use the DEXA scan info to help their performance, all while maintaining overall health and regular periods. But what about everyone else? That’s why applying standards across an entire program is so dangerous.
Perhaps the scariest anecdote in the article involved an athlete who hadn’t had her cycle in a year when she tested at 16% body fat on her DEXA scan. She was allegedly told to reduce her body fat to 13%, an objectively dangerous level for almost all female athletes. Any athlete that makes that sort of change would see their performance crater short-term, while their health trajectory deteriorates long-term.
Still, a few athletes might be able to reduce their body fat to dangerously low levels and have success–that could be genetics or background (or if thinking of some Olympic champions in the past, possibly even doping). A short-term performance model that relies on that data may even indicate that interventions like those at Oregon are “data-driven.” In actuality, it’s bad data because it ignores the complexity that drives variability among athletes. One athlete might have success, by chance, while ten athletes get ground to dust, by physiological realities. Cherry-picking data from the outliers hurts everyone, at every level of the sport.
And even for the outliers, no trophy is worth the biggest bad data problem of all: potential long-term consequences that can tear down an athlete all the way to the cellular level.
Long-term data problems
In 2014, the British Journal of Sports Medicine published the IOC Consensus Statement on Relative Energy Deficiency in Sport (RED-S), defined as “impaired physiological function including, but not limited to, metabolic rate, menstrual function, bone health, immunity, protein synthesis, cardiovascular health caused by relative energy deficiency.” Low energy availability is when total energy intake doesn’t leave enough energy availability after considering energy expenditure during exercise (undereating or overexercising or both).
A 2019 review article in the Current Opinion in Endocrine and Metabolic Research journal described how failing to eat enough food can cause a cascade of negative physiological effects. Sex hormones plummet, disturbing sexual and reproductive function. Stress hormone cortisol skyrockets, messing with hormones, adaptation, and every other physiological function you can think of. The thyroid can go haywire and the metabolic rate can drop, with decreases in mental health too.
Low energy availability can impact health even in windows shorter than a day. A 2017 article in the Scandinavian Journal Of Sports Medicine found that female athletes with menstrual dysfunction and metabolic disturbances spent more time in a low-energy state, even when controlling for energy availability in a 24-hour period. A 2018 article in the International Journal of Sport Nutrition and Exercise Metabolism had a similar finding for male athletes, with decreased muscle breakdown, worse hormone balance and metabolic disturbances from within-day deficits.
Put it all together, and as outlined in this 2020 review in the journal Nutrients, failing to adequately fuel training can cause a cascade of effects that negatively impact just about every element of being a happy, functioning human being. For women, a key element is having a menstrual cycle, with long-term studies showing amenorrhea decreasing bone health and adaptation to training.
Even if someone didn’t care about the health of collegiate athletes that win trophies, encouraging them to tinker at the margins with body composition undermines their long-term potential based on every study. Over a month or even a couple years (like a college season or college career), maybe the data and equations used by the Oregon staff show some results for some outliers. But long-term, even those outliers are undercutting their potential if they try to force their body to hit numbers that are not natural for their genetics.
It all gets back to the mix of science and art when it comes to adaptation. We’re not exactly sure what leads from Point A through Intervention B and to Outcome C. It’s tempting to think short-term and to use the NCAA championships in a few months as outcome C. Maybe that means putting people in DEXA scans and telling them some arbitrary numbers to guide their interventions.
But what if we change the time horizon? If Outcome C is a few years down the line, even the initial outliers absolutely, unequivocally need their health to adapt to their maximum levels. This isn’t just about mental health and love of the sport, this is about stone-cold results.
Uncovering each athlete’s potential requires tough training for a long time layered onto the physiological health context to adapt over many years. It’s impossible to do that with the impaired physiological function that happens for most athletes when they pursue dangerously low bodyfat levels. Eating enough is imperative for performance, and athletes that are told to sacrifice health for performance will end up having neither.
Those performance realities are layered onto the psychology of body image and disordered eating. Everyone who is engaged with this sport knows some of the stories. Maybe it starts as a coach’s comment about appearance, a team culture, or pressure from mentors, before evolving into a psychological storm of self-judgment that can hurt them long after they quit track. It’s especially tragic knowing that it all stems from a fundamental misunderstanding of exercise physiology and statistics.
Athletic performance is not a math equation, and athletes are not calculus proofs with a pesky brain. Mental health can’t be summarized in an Excel sheet, and any workbook that isn’t charted out over multiple years is woefully incomplete. Athletes are humans, and coaches have a responsibility to support the full human, on and off the track.
And this moment is where every single one of us in the community needs to step up.
This isn’t the story of one program. I have no inside information and am relying on the sources in the article, but I don’t think this comes from a place of malice or scientific illiteracy. I think it likely comes from a place of bad data analysis, bad data, and pressure to satisfy demanding standards of program success.
Good data analysis requires accepting the uncertainty of adaptation, and not trying to control every little variable, deferring to subject-matter experts like nutritionists and doctors for individualized counseling. Good data incorporates everyone, looking past the unbroken eggs to acknowledge the experiences of all athletes. And it zooms out longer-term, looking at who is healthy and happy far past age 21.
People can be given the space to make mistakes–we are all learning in this process, and I know I’ve messed up too. This goes far beyond correcting some mistakes at one program, to an imperative need to change a culture. Changing bodies for performance is superstition masquerading as science, and the superstition must end now.
Let’s elevate good science focusing on adequate fueling, long-term health, and finding each athlete’s individual strength. Lives are counting on it.
David Roche partners with runners of all abilities through his coaching service, Some Work, All Play. With Megan Roche, M.D., he hosts the Some Work, All Play podcast on running (and other things), and they wrote a book called The Happy Runner.