Do your biomechanical quirks put you at risk? Can clues like running technique and flat feet predict who gets injured? And if a risk factor is known, does that mean it “causes” the injury? Everyone sure acts like it: fixing those risk factors is a major focus of rehab.
I recently published an injury risk factor update for my book about shin splints, and I’m embarrassed by it. 😳
It’s not that the update itself is embarrassing, but what it replaced, a simplistic and generic take. I feel like I owe an apology to all my past customers: I didn’t give you good enough information! Sorry!
In my defense, ladies and gentlemen of the jury, there were major mitigating considerations. “Phoning it in” on biomechanical risk factors wasn’t outrageous negligence, and I justified it with a song I often sing on PainScience.com: “structuralism,” the excessive focus on biomechanics and alignment in musculoskeletal medicine.
But it was the bare minimum. You can’t be falling asleep on guard duty because nothing exciting usually happens. I promise content that is wide and deep, and when I finally did dive deeper … surprise! I learned things! Interesting things! Things that might even have therapeutic implications.
And so the update became a bigger deal in a way I did not expect… and it cast a longer shadow over what it replaced.
What did I learn?
I learned that some people with a very specific running style are eight times more likely to develop shins splints. That’s not business as usual in the injury prediction business. It seemed like such a big deal that I got right into an exploration of the subtle but critical differences between injury “risk factors” and “causes,” and how easily obscure x-factors can explain both. And that was where things got much more interesting.
I’ll get into the specifics — what is this “dangerous” running style?! — but only for members. Yes, this is another one of those pesky paywalled posts. (Last week I learned a new term, which I had somehow missed: “gated” content! It’s not a pay “wall,” folks, it’s a pay “gate” — a gate you can pay to open.)
Most posts are free, but it costs money to keep reading this one. JOIN NOW (for USD $6/month or $60/year) to unlock this post and much else. (Already a member? Login to unlock.) Or see more information about membership. ❐
The rest of this post is about another 1800 words (6 minutes of reading). Headings preview:
- Another reason to run “[REDACTED]”
- Is it risky to “[REDACTED]”? Eight times the shin splints!
- Causality interlude: Epstein-Barr virus and multiple sclerosis
- Which came first, the risk or the the injury?
- Why we fail to realize that correlation isn’t causation when evaluating claims about “dangerous” biomechanics
- Risk versus cause: A, B, and X
- X-Factor example #1: [REDACTED]
- X-Factor example #2: [REDACTED]
- Structuralism still rules sports medicine
Another reason to run “softer”
There are already a variety of reasons why “reducing impact” is probably a good idea with most of the running injuries, especially with shin splints, which may be the most impact-sensitive of the lot. An intriguing 2018 study by Napier et al — done right here in Vancouver — has given us another reason.
This experiment flagged peak braking force as a novel new risk factor for medial tibial stress syndrome (one of the major types of shin pain). It’s an interesting and noteworthy data point, though not terribly consequential — it just boils down to being an extra reason to “soften” running style (a strategy that is discussed in considerable detail separately in my shin splints book).
Peak braking force (PBF) is a measure of how hard your feet push backward horizontally on impact. PBF spikes on the downhill, because forward movement is powered by both momentum and gravity (and resisted by friction). PBF contrasts with vertical impact forces, which is the much more conventional thing to worry about. Thanks to gravity, vertical forces in running are much larger than the horizontal forces.
But the horizontal forces are definitely still there.
Napier et al measured a half dozen running gait metrics in sixty-five healthy female recreational runners and then tracked them through a half-marathon training program. It was some fancy measuring, like motion-capture tech for the movies. The runners were festooned with reflectors and recorded while running on a treadmill.
About a third of the runners got hurt, which is typical, and shin splints was the most common injury (followed by iliotibial band syndrome), but there was no statistical correlation with most of the things they measured — most surprisingly “average vertical loading rate,” which was the prime suspect. The authors were looking for that specifically, and didn’t find it.
Instead, PBF was prominent in the injured! Extremely prominent.
Is it risky to “brake”? Eight times the shin splints
Runners with the highest PBF scores were eight times more likely to be counted among the injured than runners with the lowest PBF scores. Eight times! It’s a correlation, not proven causation — more on this below — but I also don’t think that it’s “just” a correlation. That’s a big enough number that it’s likely they are related in some way, and that way could be causal.
So this is an interesting result that certainly looks like PBF could cause shin splints.
But it’s also a bit of an outlier in the literature, reporting a novel risk factor for the first time, based on data from a smallish study of women only, and measuring gait only on a treadmill, and there are other ways to explain the result. So there is definitely no guarantee that this result is going to be replicated by anyone else — and, a few years later, no one has even tried yet.
The authors reasonably conclude that braking force might be a good thing to try to reduce, and that probably can be done by running slower, “softer,” and with shorter steps (higher cadence) — stuff I have already been recommending for many years now.
But this causality stuff is super tricky. Just eight times more likely to get injured? Pfft, that’s nothing. Wake me up when it’s thirty-two, like this…
Causality interlude: Epstein-Barr virus and multiple sclerosis
If you have been online this month, you’ve probably seen the headlines, because it is an actual Big Deal: the endemic Epstein-Barr virus has been strongly linked to multiple sclerosis (Bjornevik et al). Unlike countless other “links” in medical science, there is so much correlation smoke here that there is almost certainly a causality fire. EBV probably causes multiple sclerosis — an extremely important discovery. For instance:
Would a vaccine against EBV protect against MS? Can the B cells that dwell in the CSF be killed or inactivated with therapeutics? Would antivirals that target EBV provide effective therapy, especially when given early in the course of disease? Now that the initial trigger for MS has been identified, perhaps MS could be eradicated.
This experiment was a triumph of modern data wrangling. Thanks to military records, researchers studied literally millions of subjects over twenty years, identifying a risk of developing multiple sclerosis thirty-two times greater after EBV infection.
And yet, even with that kind of statistical power, they still couldn’t prove causality. (Meanwhile, in sports medicine research, most studies involve only a few dozen subjects followed for a few months at most.)
The link could be explained by some common denominator, by “systematic differences between individuals,” and the authors were well aware of it; the whole enterprise was devoted to the causality question. As impressive as their number crunching was, the real artistry here was their careful ruling out of all possible ways for the signal to be due to something else. The result is technically still not proof of causality… but it’s as close as you can possibly get.
Compare and contrast with sports and rehab science. When risk factors for running injuries get studied, how much statistical power and “artistry” gets devoted to establishing causality? Usually we just get a weak and directionless correlation, and we don’t even know which came first! And even when we get a B follows A correlation, it never seems to occur to anyone to even ask if there could be another explanation for that relationship.
Let’s return to sports medicine.
Which came first, the risk or the the injury?
A “prospective trial” like Napier et al is the bare minimum required to figure out causality, but it also isn’t sufficient.
Usually when people gripe about correlation not equalling causation in science, we mean that two things have been found to occur together, but not in what order they occurred. Prospective trials fix that: they study the order of things, by testing predictions of what will happen to people in the future. Subjects are studied before they get injured, and then followed over time to see which of them get hurt.
A prospective trial figures out which came first, the chicken or the egg, by tracking chickens to see if they lay eggs. But that still doesn’t mean that the chicken caused the egg. Order isn’t everything. It’s just one clue.
Why we fail to realize that correlation isn’t causation when evaluating claims about “dangerous” biomechanics
But c’mon! If injury is eight times more likely in runners with high PBF… I mean… surely PBF must cause shin splints?! It just feels right! Right?
Wrong. A risk factor is still just a kind of correlation. It may not be a coincidental correlation, but it’s not necessarily a cause. Both the risk factor and the injury can often be explained by something else, and so many risk factors are not causes, just fallible “indicators.” And yet nearly everyone assumes that they are causes. Why? Two main reasons, I think…
The same general reason correlation and causation always get conflated: it’s just complicated and counterintuitive, and it’s easy to run with the simplest explanation and ignore the gnarly details. Human nature.
People specifically assume that sports injuries are simpler than they are, based mostly on an unjustified confidence that these injuries are basically “mechanical.” This illusion of knowledge keeps us from looking deeper. We fail to imagine other, messier, physiological, non-mechanical explanations for the data.
Risk versus cause: A, B, and X
Risk is a correlation suggesting that B will follow A.
X causes A.
X causes B.
In that scenario, the risk is not a coincidence. The correlation between A and B is there for a reason. But the cause is something else: the unknown X-factor.
So what the heck is X?! What are the non-biomechanical explanations that we fail to imagine, that no one ever seems to talk about, let alone study?
X-factor example #1: Fatigue
If higher peak braking forces don’t cause shin splints, but they are genuinely correlated, then what do they have in common? What could cause both of them?
No one knows this, so I am just speculating, but here’s a hypothetical example. What if the real explanation for this 8× odds-ratio was a subtle tendency to fatigue faster? Which could easily be subtle, and have many possible causes, everything from illness to just a bit of mild but chronic sleep deprivation.
Fatiguing a bit faster than other runners could easily explain a change in running style that causes PBF to spike.
And many causes of fatigue could also easily explain a higher vulnerability to injury, independently of PBF.
X-factor example #2: Connective tissue disease
Let’s consider another well-known risk factor for shin splints: there is plenty of evidence linking over-pronation (flat feet) to shin splints. Flat feet or fallen arches are indeed common in people who have shin splints. As with PBF, there’s an established link, which may or may not be explained by causality.
But it’s a weaker link, and there’s an excellent chance it’s not causal. People may not get shin splints because they over-pronate, but because they have something in common that makes them both likelier to get injured and likelier to have flat feet. And what could that be?
There’s at least one obvious candidate: one of the common and subtle connective tissue disorders. These conditions are routinely never diagnosed, so it’s highly plausible that they are a silent factor for many runners.
They are notorious for making people hypermobile in oddly specific ways, so they can also easily explain fallen arches. That explains A!
They are also notorious for making people prone to injury, so they could easily cause the injury. And that explains B.
So the flat feet and the injury could both be symptoms of that real cause — which is very subtle and basically never even considered. Sports medicine and research needs more imagination about neglected X-factors, and more savvy about the relevance of general medicine to “simple” injuries.
Structuralism still rules sports medicine
Eliminating or compensating for “causes” of injury like flat-feet is not merely prominent in sports medicine and rehab: it’s the overwhelmingly dominant paradigm.
This is why I am so reluctant to embrace structuralist explanations for shin splints (and much else). This is why the updated section on risk factors was rather simplistic originally: I just didn’t take it seriously. Based on twenty years of study, I have become convinced that biomechanical risk factors are almost always a red herring. But I am happy that the shin splints book is now much more specific and detailed about why.
And PBF could still be a cause. 🙂
(I wouldn’t bet on it though.)