People are having fewer babies. There’s no argument about that: this drop in the fertility rate is happening in lower-, middle-, and high-income countries, and has been since around the 1970s.
The big question is why. To what extent is the decline in fertility due to societal changes, economic circumstances, and cultural shifts that make it less desirable or practical to have kids? As countries become richer and more women enter education, there’s a clear drop in fertility, so this is clearly a big part of it.
But what about a drop in our biological capacity to have kids? Has something changed in the environment that makes us—to use the scientific term for that biological capacity—less fecund?
A new study seems to hint at an answer. It focused on the class of chemicals known as PFAS, short for per- and polyfluoroalkyl substances. These are used in all sorts of consumer products like non-stick pans, water-resistant clothes, and food packaging. These chemicals are found in most people’s bodies, and because it takes so long for them to break down, they’ve been given the scary-sounding name “forever chemicals”.
PFAS, among other chemicals, have been highlighted as a possible cause of lower fecundity and fertility, possibly because they can disrupt hormones, or can potentially cause some medical conditions that make it harder to get pregnant (none of the evidence is conclusive here, though).
The Guardian—which has published dozens of articles about these “dangerous”, “potentially toxic” chemicals in the last few years—called the new research “the first known study on the effects of PFAS on female fertility”. That’s a little odd, since earlier this year a review study put together thirteen previous such studies, drawing the conclusion that “increased levels of PFAS exposure are associated with reduced fertility in women”.
As with all such reviews, that conclusion is dependent on the quality and consistency of the underlying thirteen studies. Unfortunately, the studies we have are all over the place: even though on average they point to an association of PFAS with lower fertility, the estimate of the size of the association varies a great deal from study to study. That’s probably because the studies were looking at women in a variety of different countries, at different ages, at different points in their pregnancy, and even measuring blood PFAS in different ways. The authors of the review wrote that these problems “limit the credibility of the evidence”.
More evidence is therefore welcome. The new study looked at a sample of 382 women in Singapore, all of whom were trying to conceive. They used a blood test to check the level of various kinds of PFAS in the women’s blood, and followed them up in a year to see who had, and who hadn’t, successfully become pregnant. After this point, they found that 332 were pregnant and 50 weren’t. Were there any differences in the blood PFA levels of these two groups?
Here’s where it starts to get tricky, before we even get to the blood samples. The groups didn’t just differ in their pregnancy status: there were very substantial differences in education between the groups. In the pregnant women, 71% had a college degree; only 44% of the non-pregnant women did (interestingly, this is the opposite direction from what you’d expect from the aforementioned global trends, but remember this is just a small sample within one country, so it’s not representative).
Such a huge difference in education level means the groups weren’t perfectly comparable: they likely led quite different lives on average, and this could’ve affected their fertility in all kinds of ways. In a randomised study, the “randomised” part would mean that these factors wouldn’t differ between the groups, but this was (understandably, given it was studying a toxic exposure) an observational study, so no randomisation was done.
It’s true that the researchers “adjusted” for education when they ran their analysis – that is, they included education in their statistical model to try and ask “if everyone had the same level of education, would these results be any different?”. But statisticians have pointed out that adjustment is much harder than a lot of scientists think. One has to wonder: if the groups differed this much in education, how much might they have differed in other potentially-important factors that the researchers didn’t measure?
Then we get to the results. The researchers measured three different outcomes: whether or not the woman got pregnant; the time it took for her to get pregnant (longer times indicating lower fertility); and whether she’d had a live birth. They analysed these against the blood levels of seven different types of PFAS, as well as a “mixture” of different PFAS. In total that’s 24 different analyses, and there are even more in the paper’s appendix. Because the authors didn’t pre-register a plan for their study, we have no idea if they ran even more analyses but didn’t publish them.
They found that two or three of these many analyses were statistically significant, indicating that PFAS blood levels were correlated with more difficulty getting pregnant. But as you do more and more analyses, the chances of finding a false-positive result get higher. The results aren’t particularly strong, and it’s very conceivable (pardon the pun, given the topic here) that they just came about because of normal statistical fluctuations, rather than by anything “real” going on in the data.
This is the kind of study that, had it been approached with more rigour—ensuring the groups were comparable demographically; being more conservative about what’s considered a “significant” result; planning out in exact detail the analyses to be run—it could’ve advanced our knowledge on the effects of PFAS. As it is, though, it tells us very little.
Nevertheless, the authors press-released their study, along with a statement that governments should ban the use of PFAS in “everyday products”.
None of this is to say that PFAS definitely aren’t dangerous at some level, or that proponents of the “PFAS are dangerous” idea aren’t right to be worried in general. Nor does it say anything about the other claims proponents put forward about the health effects of PFAS beyond fertility.
But if this is the quality of the evidence being put forward, governments can be forgiven, at least for now, for not taking those proponents too seriously.