Foot Strike, Running Shoes, Barefooting, Injuries, and Biomechanics: The Importance of Both Science and Anecdote
Posted on June 30 2011
Just before I left on a camping trip last week I received an email from Amby Burfoot pointing me to an abstract of a paper to be presented at the 2011 meeting of the International Society of Biomechanics. The study, titled “Foot Strike Does Not Predict Loading Rates During Shod or Barefoot Running,” was conducted by a group of researchers from Oregon headed up by James Becker. Amby wrote up a post about the study on his Peak Performance blog, and included this line: “I have no doubt biomechanists with a different perspective will find much to dissect and debate with the Oregon team. That’s what Ph.D.’s do for fun, after all.” Well, as a Ph.D. with an interest in biomechanics, I’ll take the bait and have my fun.
Basically, the researchers had 11 habitually shod subjects run across a force plate either in shoes or barefoot, and recorded the resulting ground reaction forces. In particular, they calculated vertical loading rate, which is essentially the speed at which the foot initially impacts the ground. Loading rate is of interest because some studies have suggested that it might be linked to a higher incidence of injuries like stress fractures. They also calculated a variable known as foot “strike index” (SI) for each individual – this is basically a value from zero to one that tells you at what point from heel to the tips of the toes the foot makes initial contact with the ground (a value of 0 would be far back on the heel, a value of 1 would be at the tips of the toes). Their goal was to determine if strike index was a good predictor of the vertical loading rate. In other words, does the location at which the foot makes initial contact with the ground influence the speed with which the collision occurs.
The study concluded that stride index is not a good predictor of loading rate for either shod or barefoot runners, though the latter relationship was nearly significant with a p-value of 0.06 (a p-value of 0.05 or less is typically considered a significant result). The authors also pointed out that most of the runners adopted a midfoot strike when switching to barefoot, and that those who continued to heel strike when barefoot experienced an increase in vertical loading rate. The latter is this is why I don’t agree with Amby’s statement that “the University of Oregon team will report that their lab testing doesn’t support the notion that forefoot striking lessens the “instantaneous load rate” vs. rearfoot running.” What it shows is that rearfoot running in cushioned shoes doesn’t differ from midfoot/forefoot running when barefoot – those who continued to rearfoot strike when barefoot experienced increased loading (as was also shown by Lieberman et al., 2010). In other words, shoes allow us to rearfoot strike with a loading rate similar to that seen when we forefoot strike when barefoot, and loading rates of both are lot less than if we were heel striking when barefoot.
The study also failed to find a significant difference in vertical loading rate between the shod and unshod conditions, and indicated that this could be due to a large amount of inter-individual variation in response. I’d also suggest that having habitually shod runners abruptly switch to running barefoot also might be involved here, as Lieberman et al., 2010 indicated that a majority of their habitually unshod runners had lower loading rates than shod heel strikers when running barefoot (this despite the fact that the difference between the two groups was non-significant). In other words, a learning process might be involved when it comes to barefoot running and reduction of impact (see this abstract from the same ISB meeting for more on this).
The authors conclude their abstract with the following:
“Based on this small sample, it appears foot strike pattern does not predict VILR under shod or barefoot conditions, and changes in VILR when one switches from shod to barefoot, or from RFS to FFS, are highly variable between individuals. Thus, generalizations regarding the benefits of one foot strike pattern compared to another should be interpreted with caution.”
My take is that this abstract seems to provide support for a number of observations that exist in the literature. In particular, Lieberman et al, 2010 also showed that heel striking in a cushioned shoe results in a vertical loading rate that does not differ from that of individuals who forefoot strike when barefoot (look at Figure 2b from their Nature paper). However, what does differ is the fact that limbs in the different conditions likely handle the impact loading in different ways. A number of kinematic studies have shown differences in things like joint angles and limb positioning at impact, including an abstract of a paper by a group from South Africa to be presented at the very same meeting as the study under discussion here. It’s important to remember that ground reaction forces simply describe the reaction of the ground to forces applied by the body via the foot. These are external forces, and thus don’t describe how those forces are applied internally to individual muscles, bones, joints, etc. Thus, the same loading rate might have a very different effect if the limb is positioned in a different way upon impact. In other words, we might not be comparing apples to apples here. What’s more, as the authors point out, footstrike is just one variable to consider, and that “other variables such as joint kinematics, joint stiffness, or muscle activity, may play a more dominant role in determining the VILR (loading rate).” For more on this, see this great post by Jay Dicharry from the UVA Speed Lab.
What I find interesting about this study is that it shows yet again that individual responses are highly variable when it comes to running mechanics – it’s why generalizations are in fact so hard to make. People respond differently to different interventions, and thus Amby is right when he suggests that generalizations about the most appropriate shoe and/or foot strike should be avoided. Different things will work for different people, but that’s also why I disagree with Amby when he expresses his lack of faith in anecdotal reports and refers to them as “bad science.” One of the first things we teach in elementary biology classes is the scientific method, and one of the foundations of that method is observation. Anecdotes about people getting hurt or overcoming injury through running barefoot or in minimal shoes are observations, and they allow scientists to generate and test hypotheses, just as the authors of the studies referenced here have done.
It’s also worth pointing out that science also does have limitations, and one of those is that we are very good at describing majority patterns, but studies are often very bad at telling the individual what is best for them. Non-systematic results in biomechanics studies, as well as anecdotal reports from runners who experience different outcomes when experimenting on themselves, all seem to argue that we will never have one single answer that applies to all runners when it comes to form and shoes. That’s why I view the discussion around barefooting and minimalism to be so important – it is helping to increase variety in footwear options for a highly variable and oft-injured population of runners. The “bell curve” that Amby refers to in his post has long been skewed to one type of shoe, and what I hope is that through individual experiments an additional research we will gain a better idea of where the tip of that bell should be.
At the end of the day, scientific studies can be a useful guide, but when you are a chronically injured runner, sometimes waiting for an answer in a scientific paper can take too long. Sometimes taking a risk and trying something dramatically different can be of great benefit. Sometimes it might break your foot. Without anecdotal reports from individual experiments though, we wouldn’t have the basis upon which to generate and test hypotheses about why some people get one result and some get the other.