Are IUDs Giving People Breast Cancer?

Why the data isn't nearly as scary as the headlines suggest.

The alt text of this image on Unsplash is “green and red anchor-shaped toy” which I modified to be a bit more useful. Photo by Reproductive Health Supplies Coalition on Unsplash

If there’s one thing that the media loves, it’s a story about how standard healthcare for women is secretly dangerous and bad. This is a trope I’ve been writing about for nearly a decade now. One of my first blog posts back in 2016 was on a misleading story about the contraceptive pill causing depression. I’ve written several times about the false belief that abortion causes all sorts of long-term health issues, which comes up every time a right-wing group wants to ban healthcare for women.

This time, the story is that hormonal intrauterine devices - IUDs - are causing breast cancer. There have been a wave of headlines implying that anyone using an IUD faces a worrying increased risk of developing the nasty condition in the future.

But the data is not nearly so strong. In fact, the evidence suggests that there isn’t much to be worried about at all.

Cancerous Claims

The study itself was a recent short research letter published in the Journal of the American Medical Association. This is a type of scientific article with a very limited word count and few references - in this case, the letter was just 758 words in total. The authors described a study based on the Danish National Registry looking at hormonal IUD use and breast cancer risk.

The Scandinavians are annoyingly good when it comes to large linked datasets (yes, I know this is Sweden, they also have great data that I’m jealous of). Photo by Dmitrii Eliuseev on Unsplash

The Danish National Registry is a very cool database that includes every single medical intervention and outcome provided in Denmark. It also includes a wide range of data on non-medical things like demographic information and earnings. This is the sort of data that every epidemiologist who doesn’t live in Scandinavia dreams of having, and it makes for pretty interesting scientific research.

For this study, the authors looked at women who used the levonorgestrel IUD and a matched cohort of women who didn’t, and compared their risk of breast cancer in the following 15 years. After adjusting for a range of confounders, the researchers found that the IUD was associated with a hazard ratio of 1.4, which means a 40% increased risk of breast cancer.

There are a number of weaknesses to this study. The first obvious one is that the absolute risk was very small. The authors looked at women with an average age of 38 years. Over the following 15 years, there were on average 21 breast cancers per 10,000 women per year in the group who didn’t use the IUDs, compared to a rate of 28 per 10,000 per year in the IUD group. While this is a relatively low-risk group - breast cancer risk is strongly linked to age, and 53 is still very young in cancer terms - it’s still a tiny absolute increase in risk.

The study was also, as you might imagine, extremely minimal. There’s only so much you can do with 758 words. The authors didn’t do any causal modelling, they didn’t spend time on sensitivity analyses to rule out things like immortal time bias or competing risks, and they only controlled for a small range of confounders.

A confounder is a variable that can cause both with the intervention and the outcome - in this case IUDs and breast cancer diagnoses. One example of confounders that the authors didn’t look into was income. Women with higher income may be more likely to access IUDs, and also more likely to have a diagnosis of breast cancer due to better access to diagnostic testing.

There was also no evidence that longer use of an IUD was associated with a greater increased risk of breast cancer. This is the opposite of what we’d expect if IUDs were the causal factor in these cancers. If having an IUD made you more likely to get cancer, then having an IUD for 10 years would be worse than for just 5. The researchers instead found that the risk was the same for 5, 10, and 15 years of IUD use.

This sort of increase in risk is one of the ways that we figured out that smoking causes lung cancer. There is a strong and linear increase for smoking and lung cancer whereby every additional year of smoking increases your risk of cancer by a predictable amount. Photo by GRAS GRÜN on Unsplash

Statistical Noise

If you want a great overview of the practical implications of this IUD study, I recommend

Dr. Jen Gunter

’s piece:

Scaring Women About Hormones Makes Copy

Dr. Gunter argues persuasively that the study isn’t that strong, and that pregnancy is also risky so these findings don’t really change the picture about what you should do for your own healthcare.

I’m not a clinician, so I’d recommend Dr. Gunter’s perspective on that. But I think there’s another really interesting point to make about this particular study. The paper shows that there’s a small increased risk of one type of cancer - breast - associated with IUD use. But does this even make sense as a research question?

I’d like to contrast this small paper, which made global headlines, with a much larger and more robust study from Sweden that was published online a few month’s earlier. This paper used a similarly robust dataset, but had a much larger cohort of women. They also focused on all gynaecological cancers, not just breast cancer. They looked into competing risks, discussed causality, and generally did a pretty decent job of looking into the potential long-term cancer risks that hormonal IUDs can have.

In this study, women who used IUDs also had an increased risk of breast cancer. However, this risk was offset by a reduced risk of endometrial, ovarian, and cervical cancers. The authors of the study also argued that the increased risk of breast cancer may have simply been due to confounding factors rather than a causal connection.

There’s another explanation here. These findings are all just reflective of statistical/epidemiological noise. It’s possible that IUDs have complex and interacting impacts on your future risk of cancer, but it’s also possible that we are simply identifying correlations in different datasets that have no real relationship with human health. It’s very hard to know whether this explanation is the most likely one, but at the very least it’s an alternative explanation that gives a very different perspective on the results of this study.

The point here is that the new study is neither particularly convincing nor particularly robust. The evidence thus far is a complex mix which shows that hormonal IUDs may have long-term risks, but they may also have long-term benefits. For one thing, they’re very good at preventing pregnancy, which can be a more dangerous condition than breast cancer these days.

As someone who has neither a uterus nor a medical degree, I have no advice on what to do for your birth control. I will say that the latest data does not seem to be a good reason to ditch your IUD, especially given how effective they are at preventing pregnancy.