A new study published in the Journal of Allergy and Clinical Immunology suggested there was a link between eczema and bone fractures. Specifically, the authors found that people with eczema were more likely to experience hip, pelvic, spinal, and wrist (but not arm) fractures. So…does eczema increase the risk of bone fractures, as articles like the one below suggest?

How to interpret these types of studies is always difficult. Firstly, it’s important to understand how this study was done. This was what is called a cohort study. Essentially researchers looked at the medical records of roughly 3 million people in the UK and identified people who did and did not have eczema. They then looked to see if the people with eczema were more likely to break something in the future.

Cohort studies are different from randomized trials. In randomized trials, you take a group of people divide them into two (or more) groups and randomly assign them to get one treatment or another. Since who gets what treatment is supposed to be completely random, if you have enough people in your study, then the two groups should be roughly similar in terms of their characteristics. This way you can be certain that any differences between the groups are due to the treatment you want to study.

With a cohort study, the situation is a bit different. This is what is called an observational study because you are simply observing what happens to people without intervening. In a cohort study you simply compare two groups, in this case people with eczema and people without eczema. But there are obvious differences between people with and without eczema. People with eczema are more likely to be asthmatic, they are more likely to take steroids as part of their treatment, and they are probably different in other ways you can’t immediately think of.

People with eczema are more likely to be asthmatic, they are more likely to take steroids as part of their treatment, and they are probably different in other ways you can’t immediately think of.

The authors in this study are by no means amateurs. They did their best to adjust for as many baseline differences as possible using statistics. But not matter how hard you try, you can never completely balance out these baseline differences. In statistics, this problem is referred to as residual confounding. It means that no matter how hard you try, there will be always some differences between groups you cannot account for.

A classic example of residual confounding occurred in the early 2000’s with hormone replacement therapy. Hormone replacement therapy used to be routinely given to women after menopause for its health benefits. There were in fact many observational studies that suggested that it reduced the risk of heart disease. However, in 2001 the Women’s Health Initiative randomized trial was published and showed that it did not. What essentially happened is that the earlier observational studies had residual confounding because the women who wanted to take hormone replacement therapy where in general healthier and more health conscious than the women who did not. The fact they were more health conscious actually led them to seek out hormone replacement therapy because it was seen to be beneficial. But once you tested it in randomized trials, where the decision to give hormonal therapy was entirely random, there was no benefit because the two groups were now exactly the same.

We can’t say for certain whether people with eczema are at increased risk for broken bones or not.

We can’t say for certain whether people with eczema are at increased risk for broken bones or not. Given that people with eczema (and asthma) are more likely to get steroids than other people, and since steroids increase the risk of osteoporosis, and since osteoporosis in turn increases the risk of bone fractures, the idea is not completely implausible. But the question remains, is it eczema itself that increases the risk of bone fractures, or is it the steroids, or is it some unrelated factor we can’t measure that has nothing to do with eczema.

Clearly steroids can be very helpful for patients, but because of their side effects we should obviously only use them when necessary and for as short a time as possible. As for any other worry about eczema and fracture risk, since you can’t wake up one day and decide not to have eczema you can’t do much with this type of information. You can certainly check to make sure you don’t have osteoporosis but we generally do that with everyone (especially women) after a certain age anyway. 

It’s important not to over interpret the results of observational studies like this one. While they are very useful in medical research, they can sometimes fall victim to residual confounding. Since obviously can’t do a randomized trial and randomly give someone eczema just to see what will happen to them (this is both physically impossible and grossly unethical even if it were), we probably won’t have any truly definitive answer on this issue. All we can do is continue to treat people as best we can and hopefully limit the amount of steroids we prescribe to patients, to those who absolutely need them. There’s also no real value in worrying about something you cannot change. After all, the trick to life is knowing how to worry selectively. Worry about the things you can change and don’t worry about the things you can’t.

<em>Christopher Labos, M.D.</em>
Christopher Labos, M.D.