Over the past couple of months, headlines have warned expectant parents that something as ordinary as a pain reliever or an antidepressant taken during pregnancy could “cause autism.”
The stories have focused on acetaminophen (also known as paracetamol or the brand name Tylenol) and selective serotonin reuptake inhibitors (SSRIs), such as Prozac (fluoxetine) and Zoloft (sertraline).
But here is what the headlines leave out: acetaminophen is commonly used during pregnancy to manage fever, pain or stress, all of which can themselves affect fetal development. Similarly, SSRIs are prescribed for depression or anxiety, conditions that also influence pregnancy outcomes. In many cases, it may well be the illness, not the treatment, that shapes child development.
Both classes of medications have been studied extensively for decades. Yet despite what the headlines suggest, the evidence that acetaminophen or SSRIs cause autism is weak, inconsistent and easily misinterpreted.
With a background in genetics and clinical teratology — the scientific study of birth defects — my research examines how maternal exposures in pregnancy interact with genetic and environmental factors to influence child development. From this perspective, I want to explain why the research on acetaminophen and SSRIs is often misunderstood, and why reducing complex science to alarming headlines does more harm than good.
With the recent U.S. Food and Drug Administration (FDA) panel on SSRIs in pregnancy, and the public claims made by United States Secretary of Health and Human Services Robert F. Kennedy Jr. regarding acetaminophen and autism, there is a need for evidence-based information. While my focus will be on autism, the same issues apply to media coverage linking pregnancy exposures to attention deficit/hyperactivity disorder (ADHD).
Association is not causation
Much of the research behind these headlines is observational. Such studies can spot associations but cannot prove cause and effect. The associations they identify are usually small to modest, and other factors are often responsible.
Confounding is a good example. Pregnant women may take acetaminophen because they have a fever, but fever itself has been linked to higher risks of neurodevelopmental outcomes such as neural tube defects. Similarly, someone prescribed SSRIs may be experiencing depression or anxiety, which on their own are associated with differences in pregnancy outcomes and child development. Here, the medication may appear to be the cause, when in reality it is the condition being treated.
Another problem is misclassification. Most studies rely on mothers recalling how often they used acetaminophen during pregnancy. Memory is imperfect: some under-report, others over-report, and details about dose or timing are often missing.
With SSRIs, misclassification can arise when prescriptions are used as a proxy for exposure. A woman may fill a prescription in early pregnancy but stop taking the medication later, while records still count her as continuously exposed. Both scenarios distort results.
Even the outcomes themselves are not always measured consistently. Diagnoses like autism spectrum disorder vary across countries and over time. Some studies use parental questionnaires instead of medical diagnoses, which can be subjective. Two children with the same traits might be classified differently depending on who reports them.
When researchers adjust for these kinds of factors, the apparent risks often shrink or even disappear.
Research shortcomings and media spin
Beyond these challenges, research in this area has other limitations: timing and dose are often recorded crudely; use of other medications taken at the same time is not systematically assessed; results are inconsistently replicated; and while biobank and biomarker studies — which analyze measurable biological signals, such as blood levels of a substance, to indicate exposure — hold promise, they are uncommon and usually capture exposure only once.
Furthermore, studies that report positive associations are more likely to be published, and once they are, news outlets are far more likely to amplify findings that sound alarming than ones that reassure. “Everyday painkiller linked to autism” makes a clickable headline; while a more balanced one that might read something like “Evidence inconsistent, no strong effect found” does not.
This cycle amplifies fear, leaving parents confused and anxious.
The real dangers of untreated conditions
It also matters what happens when pain, fever, depression or anxiety go untreated.
High fever in pregnancy is known to increase the risk of neural tube defects and other complications. Untreated maternal depression and anxiety can lead to poor prenatal care, substance use, preeclampsia, premature birth, impaired bonding and even suicide — one of the leading causes of maternal death.
In these cases, acetaminophen and SSRIs are not just helpful. They can be lifesaving.
Understanding autism
Autism is not caused by a single medication or choice. It is a complex neurodevelopmental difference with a strong genetic basis. Heritability estimates are around 70–80 per cent, meaning much of the variation in risk is tied to parental traits and shared family environments.
Autism also clearly runs in families: siblings of autistic individuals are 10 to 20 times more likely to be diagnosed, and many parents or relatives show autistic traits even without formal diagnoses. This familial pattern reinforces that genetics and shared environment play a major role.
Sibling studies add weight by comparing siblings where one was exposed to a medication in pregnancy, and the other was not.
If the medication were truly causing autism, clear differences would appear. But often they shrink or disappear, pointing instead to shared genetics and environment.
Of course, environmental factors can still play a role. But to suggest that a common medication like acetaminophen “causes” autism oversimplifies the picture and risks stigmatizing families, while fuelling guilt among mothers who already face intense scrutiny during pregnancy.
Communicating risk responsibly
One of the greatest challenges is not the research itself, but how its results are communicated. Studies often report risks using relative measures. For example, a study might report that acetaminophen use is associated with a 30 per cent increase in autism risk. That sounds alarming. But in absolute terms, the difference is much smaller.
Autism affects about three in every 100 children. Even taking the highest reported increase in studies — a 30 per cent relative rise — that number only goes up to about four in 100. In other words, instead of 97 children without autism, you’d have 96. So while the increase is real, the absolute change in risk remains small.
Therefore, balanced communication matters. When parents hear only the alarming side, some may stop taking needed medications abruptly, which can be dangerous. Others may endure untreated illness out of fear. Clinicians and researchers should emphasize absolute risks, acknowledge limits, and aim to inform, not frighten.
Informed, not alarmed
The lesson isn’t that acetaminophen or SSRIs are risk-free. No medication is. But decades of research show that, when clinically indicated, they are generally safe in pregnancy. The risks of untreated illness are often greater.
Autism is a condition caused by many factors, including genetics, not something to blame on common medications — or mothers.
Expectant parents deserve clear, compassionate, evidence-based information, not fear-driven headlines. Association is not causation, absolute risks are small, and informed choice should never be replaced by alarm.
Sura Alwan is the founder and executive director of PEAR-Net Society, a Canadian nonprofit that advocates for maternal fetal health and safety of medications and other environmental exposures during pregnancy.
This article was originally published on The Conversation. Read the original article.
