(CNN) International health agencies hit back against Trump’s claims about Tylenol and autism .
Ruling HHS by hunch, guess, and caprice, Robert Francis Kennedy Jr. has been immensely destructive to the public health establishment. But like the broken clock that shows the correct time twice a day, he may be about to have his moment. Kennedy bases his determination on a paper that does not claim to prove the point. But proof in the biological sciences is a murky concept. When does correlation between a drug and morbidity transition to the stronger statement of causality?
Quoting (BMC) Evaluation of the evidence on acetaminophen use and neurodevelopmental disorders using the Navigation Guide methodology,
Numerous well-designed studies have indicated that pregnant mothers exposed to acetaminophen have children diagnosed with neurodevelopmental disorders (NDDs), including autism spectrum disorder (ASD) and attention-deficit/hyperactivity disorder (ADHD), at higher rates than children of pregnant mothers who were not exposed to acetaminophen.
This is a metastudy, which uses Bayesian statistical analysis to combine the results of multiple clinical studies. This methodology can reduce the effect of confounding variables. As a simple though unlikely example, pain could be a confounding variable, the actual cause of ASD, rather than the pain reliever.
A study by Gustavson et al. attempted to remove confounding variables, comparing the rate of ADHD (but not ASD) among sibling with differing prenatal exposure. Quoting,
This study carried out extensive bias analyses and showed that the entire effect disappeared in the sibling-controlled analyses.
The statistical evidence of the constituent clinical studies is weighted depending on the likelihood the results are not from chance; this is called statistical power. Gustavson’s study had a small number of patients, so it was given small weight. Studies with larger numbers of patients, or less likelihood of confounding relationships, were given more weight.
All of this is so far about probabilities. When does probability become fact? In every field that uses Bayesian statistics, the form of the answer is exactly the same. Every study, no matter how careful or how large, can produce a result that arises purely from chance. That chance is provided by the statistics of the study.
If that chance is smaller than a certain number, we say “proven.” If it is larger, we say “unproven.” But how is the cutoff decided? It is a choice, not a given. In physics, the number is 0.00006% . In the biological sciences, it is more like 3%. Practicality rules; we are unlikely to do better than 3% in medicine; the confounding variables cannot be further unwound.
There is an obvious problem with 3%. If you have enough random ideas, and test all of them, a certain number will satisfy rigorous studies, and come to be accepted as (false) fact. The scientific method as defined by John Dewey, early in the 20th Century, provides further safeguards against nonsense. It begins with observation and follows with a hypothesis, a possible explanation, followed by a theory which can be tested and proven wrong. If you can’t develop a hypothesis, you stop, while continuing to observe. If it can’t be proven wrong, you’re not doing science. This prevents the random proof of nonsense.
Unfortunately, it is easy to generate a hypothesis relevant to acetaminophen and ASD. Like NSAIDs, this drug inhibits inflammation. Unlike NSAIDs, it also acts directly on the brain, raising the threshold for a neuron to fire. Quoting (Wikipedia) Paracetamol,
In 2018, Suemaru et al. found that, in mice, paracetamol exerts an anticonvulsant effect by activation of the TRPV1 receptors[121] and a decrease in neuronal excitability by hyperpolarization of neurons.
In the prenatal brain, this could interfere with Hebbian learning. See Hebb’s Rule Applies During Early Brain Development, With Subcellular Precision.
ASD children are born with an excess of neurons. A hypothesis:
The increased neuron firing threshold, caused by acetaminophen, inhibits pruning associated with prenatal Hebbian learning, leading to the excess of neurons associated with ASD.
Robert Kennedy’s war on the scientific establishment is abhorrent. But he may yet have his day, reminding us that knowledge does not guarantee correctness, and bureaucracy can blunt intellect.