Stress fractures: is it weak bones or muscles?


As of September 2017, new Sweat Science columns are being published at Check out my bestselling new book on the science of endurance, ENDURE: Mind, Body, and the Curiously Elastic Limits of Human Performance, published in February 2018 with a foreword by Malcolm Gladwell.

- Alex Hutchinson (@sweatscience)


A new study from researchers at the University of Calgary, published in the November issue of Medicine & Science in Sports & Exercise, looks at bone quality and leg muscle strength in a group of 19 women who have suffered stress fractures in their legs, and compares them to a group of matched controls. The basic results:

  • the women who got stress fractures had thinner bones;
  • at certain key locations, the quality of the bone was lower in the stress fracture group;
  • the stress fracture group also had weaker leg muscles, particularly for knee extension (lower by 18.3%, statistically significant) and plantarflexion (lower by 17.3%, though not statistically significant).

Now, this sounds very similar to the results of a University of Minnesota study published a couple of years ago. Here‘s how I summed up the conclusions reached by those researchers:

What’s interesting, though, it that the bone differences were exactly in proportion to the size of the muscles in the same area, and there was no difference in bone mineral density. What this suggests is that the best way to avoid stress fractures is to make sure you have enough muscle on your legs — presumably by doing weights and (it goes without saying) eating enough.

What I don’t understand is that, in the new Calgary study, even though they mention the Minnesota study repeatedly in their discussion, they don’t discuss at all this idea that it’s the lower muscle strength that dictates the reduced bone size and thus the stress fracture risk — even though that was the primary conclusion of the Minnesota study. Instead, they say “the role of muscle weakness in [stress fractures] is unclear from previous studies,” and suggest that weaker knee extension might change running form to produce a “stiffer” running stride or somehow alter the direction of forces on the bone during running — both of which seem like unnecessarily complex and speculative ideas compared to the straightforward link between muscle strength and bone strength.

It’s entirely possible that I’m missing something here, because the paper is quite complex. But what I take away from it is, once again, that strengthening your legs is likely (though not yet proven in a prospective trial) to reduce your stress fracture risk.

7 Replies to “Stress fractures: is it weak bones or muscles?”

  1. “What I don’t understand is that, in the new Calgary study, (…) they don’t discuss at all this idea that it’s the lower muscle strength that dictates the reduced bone size and thus the stress fracture risk…”

    I can understand some of their hesitation. Apparently upper leg weakness combined with and low bone density in the lower leg are associated with fractures, while a typical measure of lower leg strength -plantarflexion- is not. If muscle strength would dictate bone strength directly, you would expect lower leg strength and lower leg bone density to be associated.

    Granted, “unclear from previous studies” doesn’t seem quite the right formulation.

  2. @RH: I’m fine with hesitation — heck, my usual criticism is that people read too much into their results rather than too little. What surprised me is that they didn’t even MENTION the possibility that muscle weakness could lead to bone weakness, despite citing heavily the Minnesota study that advanced that theory. To me, that makes the discussion of the study’s implications incomplete. (Again, unless I’m missing something, which certainly isn’t impossible.)

    It’s also worth noting that, while you’re correct that plantarflexion wasn’t significantly associated with fractures, it was 17.3% lower in the fracture group than the non-fracture group. (In comparison, the significant difference in knee extension was only 18.3%.) Because of the relatively small sample size and large scatter among values, the difference in plantarflexion wasn’t statistically significant. But that’s still quite different from finding little or no difference between the groups with a small range of uncertainty.

  3. On second thought (and reading) you may be right, especially about reading to much in one’s results.

    What is missing in this study is density data of the upper leg bone. It may not be the common fracture site, but it could show if muscle strength and bone density are really two sides of the same coin or that a more complex explanation -like the one they offer- is needed.

    It does seem a little odd not to mention an apparently well supported theory that would turn these results into a statistical fluke.

  4. And I should add that I don’t think that bone quality (i.e. “microarchitecture”), on its own, is unimportant. I would expect that it is an independent risk factor, and that was clearly the main focus of their study. I just thought that, in a study looking for effects of bone quality AND muscle strength, the potential for interaction between those two factors deserved a mention.

  5. A former colleague just told me about this blog and I am very happy to see that people are discussing this specific paper as I am one of its authors. It actually summarizes my Master’s thesis so I spent quite some time on the study. 😉
    Anyways, I am happy to comment on some of the mentioned points of criticism…
    First of all, you mentioned that we didn’t discuss the link of muscle strength and bone strength – this can actually be found in paragraph 3 of the introduction. That is also where we mention the theory of increased shear stresses caused by lower muscle strength, which is what most of the bone section of the discussion is all about. Therefore, we do discuss this aspect in great detail.

    You also seem to be very concentrated on that one study from Minnesota. I have spend months going through relevant literature and you have to see our study in context with all the other studies as well. You are right with it being the closest study there is to date to what we did, but there are certainly other studies that have to be considered. Fact is that researchers are in no way clear about what role the muscle-bone relationship plays in stress fracture occurrence. There are conflicting results on the topic of stress fractures in general and as mentioned in our paper this is not different for the muscle studies on this topic. A very important point to consider here is that this was the first study where muscle forces were measured directly (through plantarflexion torque) – other studies on runners (or athletes of running-related sports) have used surrogates of muscle strength (just like the Minnesota researchers used muscle cross-sectional area). Staying with that example, muscle cross-sectional area is known not to be a very reliable indicator of actual muscle strength. Yes, the study gives an INDICATION that there could be something there, but to use it interchangeably with muscle strength is outright wrong.

    Another point discussed here was that we didn’t really mention ‘the obvious’ (lower muscle strength -> smaller bone -> stress fractures) but instead started discussing more complex relationships. Well, as a researcher one has to look at the results in an objective way. Just because the Minnesota paper showed a smaller muscle cross-sectional area as a possible risk factor, we had to look at your own results closely and not just pick out what fits that and forget about the rest. In our case the plantarflexion torque (most related parameter to the muscle cross-sectional area measured in the Minnesota study) wasn’t significantly different between the groups and that’s just how it was. There was a trend, which we discussed, but nothing more. On the other hand the knee extension strength WAS significant and isn’t related to what the Minnesota study looked at. We had to discuss this finding even if it brings up a more complex view on the topic than the ‘obvious’ connection is. In an organism such as the human body nothing is as simple as it appears on first glance anyways…
    And RH is right – we didn’t want to read too much into results that weren’t statistically significant. We had small groups (not smaller than most non-military studies though) and you really have to be careful interpreting results like that. BTW, we do discuss in the paper how our trend of lower plantarflexion torque supports the other study’s findings.

    The last point I wanted to mention here is that the main focus of the study was on the bone itself. It was the novel part of the study as nobody had looked at the bone microarchitecture like that before. And it is important to emphasize here that it wasn’t the BONE SIZE that was smaller in SF subjects of our study… it was microarchitectural parameters that differed (which influence the bone strength drastically).

    I hope I could clarify some points for you here, but if you have more questions I am happy to discuss this further (or just read my thesis – I had a lot more space to explain things than in this paper!). 😉
    But in general you got the right message – leg muscle strengthening seems to help to prevent stress fractures!

  6. RH – I just noticed your comment about the bone density. I am not sure what location you mean with “upper leg bone”, but we did do DXA density measurements of the femoral neck (see start of section ‘bone outcomes’). I am not really sure what you mean with

    “it could show if muscle strength and bone density are really two sides of the same coin or that a more complex explanation -like the one they offer- is needed.”

    Maybe you could expand on that?


  7. @KS

    Thanks for taking the time to explain.

    As I assume to be the case for most readers of this blog, I don’t have access to the whole article, so I didn’t see the section you refer to.

    I reacted to the point made in the blog, that a rather complicated mechanism for lower leg fractures is suggested (roughly put: different running style), while there seems to be a simpler, more general explanation (low muscle strength –> low bone density –> fracture).

    From the picture and an admittedly too quick reading of the abstract I got the impression that only lower leg bone density was measured, while both upper and lower leg muscle strength were measured.

    A significant correlation was found between (I wrongly assumed lower leg) bone density and fractures, as well as between upper leg strength and lower leg fractures. Yet no significant relationship between lower leg muscle strength and lower leg fractures was found. In itself, this is a refutation of the “low muscle strength –> low bone density –> fracture” hypothesis, since in the lower leg we apparently have strong muscle, but weak bone.

    As Alex pointed out however, the relationship between lower leg muscle strength and lower leg fractures was close to being significant, so it could be possible that, in truth there is a simple relationship between weak muscle and weak bone, but by some coincidence, it didn’t show up as significant in the lower leg in this study. The obvious question to me was, if there was any relationship between upper leg strength and upper bone strength. If so, the “low muscle strength –> low bone density –> fracture” hypothesis is possibly saved. If not, it is wrong.

    So, to answer your question I had no specific location in mind, but simply a location that is somehow representative of bone density in the upper leg.

Comments are closed.