#367 - Tylenol, pregnancy, and autism: What recent studies show and how to interpret the data

Oct 6, 2025 Episode Page ↗
Overview

Dr. Peter Attia critically examines recent headlines linking prenatal acetaminophen (Tylenol) use to autism, providing a framework for evaluating complex conditions and research. He emphasizes using the Bradford Hill criteria to assess causality, balancing medication risks with maternal health, and understanding the true drivers of rising autism diagnoses.

At a Glance
12 Insights
1h 27m Duration
17 Topics
5 Concepts

Deep Dive Analysis

Introduction to Acetaminophen, Autism, and Critical Thinking

Groundwork: Rising Autism Rates and Scientific Thinking Challenges

FDA Pregnancy Drug Categories and Evaluation Framework

Understanding Claims About Acetaminophen and Autism

Increase in Autism Rates and the Multiple Comparisons Problem

Analyzing the Review Paper on Acetaminophen and Autism

Detailed Breakdown of the Xi Study Limitations

The Swedish Cohort Study on Prenatal Tylenol Exposure

Why Observational Studies Can't Prove Causality

Applying the Bradford Hill Criteria to Acetaminophen and Autism

Summary of Bradford Hill Criteria Application

Genetics and Heritability of Autism Risk

Explaining the Dramatic Rise in Autism Diagnoses

Other Risk Factors for Autism: Parental Age and Environment

Balancing Risks and Benefits of Medication Use During Pregnancy

Considerations for Taking Tylenol During Pregnancy

Final Thoughts on Critical Thinking and Risk Assessment

Multiple Comparisons Problem

This statistical issue arises when many variables are tested against an outcome, increasing the likelihood of finding statistically significant associations purely by chance. It means that if you look at enough variables, you are bound to find associations that appear significant but are not truly causal.

Bradford Hill Criteria

A set of nine principles laid out in the mid-1960s to help determine whether an observed epidemiological association is likely to be causal. These criteria include strength of association, consistency, temporality, biological plausibility, and others, providing a disciplined way to interpret correlations.

Confounding Variables

These are unobserved or unmeasured factors that influence both the exposure and the outcome, making it appear as though there is a direct causal link when there isn't. An example is the correlation between ice cream consumption and drowning, both confounded by warm weather.

Heritability

The proportion of variation in a trait within a population that is attributable to genetic factors. It is often assessed by comparing identical (monozygotic) and fraternal (dizygotic) twins to understand the genetic contribution to a trait like autism.

Diagnostic Aperture

Refers to the scope and criteria used for diagnosing a condition. A progressively expanding diagnostic aperture, as seen with autism spectrum disorder, means that more individuals fall under the diagnostic umbrella, contributing to an apparent increase in prevalence.

?
What specific claims are being made about acetaminophen (Tylenol) use during pregnancy and autism?

The basic scientific claim is that maternal use of acetaminophen during pregnancy is associated with an increased risk of autism in the exposed child, prompting calls for FDA warnings and label changes, though authoritative sources do not yet conclude a causal relationship.

?
Why are so many different factors being linked to autism, and why do these ideas persist?

The dramatic rise in autism rates drives a strong motivation to find triggers, leading to the 'multiple comparisons problem' where many variables are tested, increasing the chance of finding spurious associations. These ideas persist because it's nearly impossible to definitively disprove a link in epidemiology, unlike in controlled experiments.

?
What was the specific publication that triggered the recent concern about acetaminophen and autism?

The recent alarm was triggered by a systematic review of earlier research published in late August in BMC Environmental Health, which summarized existing studies on prenatal acetaminophen exposure and neurodevelopmental disorders like autism.

?
How reliable is the Xi study, which showed a strong association between acetaminophen and autism?

The Xi study, while showing a strong association, has significant limitations including a very small sample size, a potential concentration of autism cases due to participant retention, and unreliable exposure measurement based on a single umbilical cord blood sample that doesn't reflect overall pregnancy exposure.

?
What did the large Swedish cohort study find regarding prenatal acetaminophen exposure and autism?

The Swedish study, including nearly 2.5 million children, initially showed a small but statistically significant 5% increase in relative risk for autism with prenatal acetaminophen exposure. However, this association was entirely abolished when a sibling-controlled analysis was performed, suggesting confounding factors were at play.

?
Why is it difficult to determine causality from observational studies in epidemiology?

Observational studies struggle with the potential influence of confounding variables that are not immediately obvious or cannot all be identified. These unmeasured factors can make it seem like one variable causes another when a third, hidden variable is actually responsible for both.

?
Does maternal Tylenol use explain the five-fold increase in autism prevalence over the past generation?

No, it is almost certain that maternal Tylenol use is not the primary cause for the dramatic rise in autism. If it plays any causal role, it would be very small, especially compared to other factors like expanded diagnostic criteria and increased awareness.

?
What are the most significant risk factors for autism?

Genetics play the largest role, accounting for an estimated 80-90% of the inter-individual variability in autism risk. Other significant factors include advancing parental age (both maternal and paternal), maternal obesity, metabolic disease, preterm birth, and air pollution exposure.

?
How much of the increase in autism diagnoses can be attributed to changes in diagnostic criteria and awareness?

Analyses suggest that the expanded diagnostic criteria and increased awareness account for 40-60% of the increase in autism prevalence. Additionally, narrowing racial and socioeconomic disparities in diagnosis indicate more widespread screening.

?
What advice should pregnant women consider regarding acetaminophen use?

For minor discomfort, it may be best to skip acetaminophen. However, for debilitating pain or fever, judicious use of acetaminophen, under a physician's oversight, may be beneficial, as fever itself carries known risks to a developing fetus, and acetaminophen is generally the safest option for fever reduction during pregnancy.

1. Cultivate Critical Thinking

Actively practice scientific and critical thought, recognizing that humans are not naturally wired for it, to consciously avoid falling into non-scientific thinking patterns.

2. Evaluate Associations Systematically

When confronted with an association between an exposure and a condition, first confirm its statistical existence, then determine the likelihood of causality, and finally assess the effect size to inform your behavior.

3. Apply Bradford Hill Criteria

Utilize the Bradford Hill Criteria as a disciplined framework to assess the likelihood that an observed epidemiological association is truly causal, considering factors like strength, consistency, and temporality.

4. Resist Single-Cause Explanations

For complex conditions like obesity or autism, resist the temptation to assume a single cause, as they typically involve multiple contributing factors.

5. Approach Evidence Objectively

Examine scientific evidence, especially on emotionally or politically charged topics, carefully and objectively, avoiding political debate to ensure unbiased understanding.

6. Update Beliefs with New Data

Embrace the scientific mindset of constantly updating your conclusions and understanding probabilities as new evidence emerges, viewing changes in perspective as a strength.

7. Be Wary of Confounding Variables

Recognize that observational data can be influenced by unseen confounding variables; randomized controlled trials are generally the only way to establish causality with high certainty.

8. Prioritize Maternal & Fetal Health

Generally, pregnant women should stop most medications and supplements, but balance this against the mother’s health needs, as untreated conditions can also pose risks to the unborn child.

9. Use Acetaminophen for Fever

If pregnant and experiencing a fever, use acetaminophen to reduce it, as fever itself carries known risks to the developing fetus.

10. Judicious Acetaminophen for Pain

For minor aches and pains during pregnancy, err on the side of caution and consider skipping acetaminophen; for significant, debilitating pain, use it judiciously under a physician’s oversight.

11. Understand Autism Risk Factors

Recognize that genetics are the predominant factor in autism risk (80-90% heritability), and do not attribute the rise in autism rates to maternal Tylenol use, as its potential role is very small or unmeasurable.

12. Review Show Notes for Detail

Access the detailed show notes, which are freely available for this episode, to follow the logical analysis and practice critical thinking on complex topics.

We are not wired for critical and scientific thought. This is something I've written about, and we're going to actually link to a piece I wrote over 10 years ago that I think synthesizes that point really well.

Peter Attia

There are no proofs in biology. It's not like mathematics. You don't get to write QED at the end of your work here. What we're really dealing with here is probabilities.

Peter Attia

A thousand sow's ears makes not a pearl necklace.

Peter Attia

The probability that the association between acetaminophen use by a mother and the development of autism of her child is a very low probability event to be causal.

Peter Attia

If you just want soundbites, you're never going to learn. Honestly, if you just want soundbites, this isn't the podcast for you. But if you actually want to be able to learn to think for yourself, then that's what we're here to do.

Peter Attia

Structured Approach for Evaluating Exposure-Condition Associations

Peter Attia
  1. Confirm that there is indeed a statistical association between exposure X and condition Y.
  2. Determine the likelihood that the statistical association is causal, using tools like sensitivity analyses and falsification tests (not trying to 'prove' causality, but assess probability).
  3. If the association is believed to be more likely causal than not, understand the effect size to determine its practical importance.
6.7 cases per thousand children
Autism prevalence in 2000 According to the CDC
32.2 cases per thousand children
Autism prevalence in 2020 According to the CDC, a five-fold increase from 2000
0.05 (5%)
Significance level (p-value) for a single hypothesis test Used in the coin flip psychic test example
98.5%
Correlation between per capita margarine consumption and divorce rate in Maine An example of a spurious correlation
97.1%
Correlation between number of physicists in California and Michael Schumacher's F1 ranking (2003-2012) An example of a spurious correlation
2 out of 6
Percentage of studies in review paper that showed no significant association between acetaminophen and autism Contrary to the review paper's claim of consistent positive association
72%
Correlation between self-reported acetaminophen use and cord blood levels Reported in a separate study, indicating room for error in cord blood measurements
11%
Autism rates in the Xi study participants Compared to general population rates of 3% currently and 0.7% at enrollment, suggesting biased participant inclusion
1.05 (5% increase)
Relative risk increase for autism with prenatal acetaminophen exposure (Swedish study, general cohort) Statistically significant with a 95% CI of 1.02 to 1.08
0.09%
Absolute risk increase for autism with prenatal acetaminophen exposure (Swedish study) Increase at 10 years, less than one-tenth of one percent
7.5%
Percentage of mothers using acetaminophen in the Swedish study Lower than rates in the US (up to 50%) and some other countries
220,000
Number of children in the Japanese cohort study Almost 40% exposed to acetaminophen during gestation; found a 6% uptick in autism rates in general cohort, but not statistically significant
1.5
Threshold for 'weak' effect size in pharmacoepidemiology Effect sizes below this are generally considered weak due to pervasiveness of bias
80-90%
Heritability of autism risk Accounts for inter-individual variability in autism risk
40-60%
Contribution of expanded diagnostic criteria to autism prevalence increase Based on direct analyses
5-15%
Contribution of advancing parental age to autism prevalence increase Seen mostly in the U.S. and other high-income countries
38%
Increase in PM 2.5 air pollution globally Driven largely by industrialization of China and India, a potential environmental risk factor for autism
25-200%
Increased risk of birth defects with maternal fever during pregnancy For conditions like cleft palate or neural tube defects
25-40%
Increased risk of autism with maternal infection exposure Reported in several studies
Up to 200%
Increased risk of autism with maternal fever exposure Across most analyses