The US Department of Health and Human Services yesterday announced that pregnant women should avoid acetaminophen — known in Britain as paracetamol, and the active ingredient in Tylenol — because of concerns it may increase the risk of autism in unborn children. But millions of mothers use paracetamol to manage pain and fevers during their pregnancies because it’s the only drug available to them to safely do so. Or at least, that’s what they’ve been told for the past few decades, but apparently no longer.
The HHS’ rationale for the new Tylenol warning is based on a systematic review led by Harvard Dean of Public Health Andrea Baccarelli. Dr. Baccarelli’s review contends that there is a “strong, consistent association between prenatal acetaminophen exposure and ADHD/ASD/other NDDs”. The review emphasised that this result was exceptionally robust, that it stood strong in studies controlling for bundles of different confounders, with negative control exposure periods, and with propensity score matching. The only problem is that none of these methods are reliable and the authors actually ignored some good evidence that contradicts their conclusion.
The main failing point of the review is that its authors knew about good evidence in the form of sibling control studies — studies where scientists compare a sibling who is exposed to Tylenol in utero to a sibling who wasn’t exposed to it — and they ignored those results. But those studies were the strongest; it just happens to be the case that those studies didn’t find anything alarming. To me, that seems like the correct explanation for why the authors were so gung-ho to attack them on spurious grounds.
In the largest, highest-quality example, researchers used almost 2.5 million births from Sweden between 1995 and 2019, used great measures of Tylenol exposure in a country where most Tylenol use is by prescription, and they found much the same as the other non-causal studies: mothers who used more Tylenol during pregnancy have children with autism and other neurodevelopmental disorders more often. But they went further than replicating that purely correlational association. In fact, they showed that, when comparing differentially exposed siblings, the one exposed to Tylenol was no more likely to have any disorder than the one who was not exposed. In other words, the association isn’t causal at all.
The review was plagued by non-sequiturs. It cited weak evidence as if it were strong. The authors failed to quantitatively synthesise the findings of the studies they referenced. Worse, they botched the table entries and even cited studies in ways directly contradicted by the studies themselves. The review was lazy, but it ostensibly supports what some people really want to believe: that Tylenol causes autism.
As a result of this bastardised faux-science masquerading as real information, millions of pregnant women will be scared into avoiding Tylenol or any brand of paracetamol, even when it might be the drug they need. Mothers who need to break a fever or manage pain have no alternatives: opiates aren’t acceptable, and Ibuprofen isn’t safe before foetuses have undergone about two trimesters of growth. If it were true that Tylenol is causing autism, then there might be some debate to be had about the risk-reward calculus for letting pregnant women use it, but there isn’t. The science is abundantly clear: Tylenol does not cause autism.
When our public health authorities get scientific questions wrong, suffering follows. Who will cry out for the mothers?
Source link
