Comparing siblings removed the paracetamol-autism and ADHD signal entirely

Design

Population-based cohort with sibling-matched and negative-control analyses

Population

708,020 mother-child pairs in Hong Kong; sibling cohorts of 124,333 (autism) and 97,285 (ADHD)

Primary outcome

Diagnosis of autism spectrum disorder or ADHD in offspring

Effect

Autism aHR 1.00 (95% CI 0.91–1.11); ADHD aHR 1.01 (0.93–1.08)

This population-based cohort used 708,020 mother-child pairs in Hong Kong from 2001 to 2023, with paracetamol exposure taken from dispensing records (about 43% exposed). To remove family-level confounding, the main analysis compared siblings who differed in prenatal exposure: 124,333 children for autism and 97,285 for ADHD.

In sibling comparisons, prenatal paracetamol was not associated with autism (aHR 1.00, 95% CI 0.91 to 1.11) or ADHD (aHR 1.01, 0.93 to 1.08), regardless of timing, cumulative dose or pattern of use. Conventional analyses did show positive associations, but so did a negative control: paracetamol taken before pregnancy was also linked to autism (HR 1.12) and ADHD (HR 1.24). A drug taken before conception cannot cause these outcomes, so the conventional signal is best explained by family factors.

This adds to large Scandinavian sibling analyses that reached the same conclusion. Undertreated fever and pain in pregnancy carry their own risks, and the alternatives, especially NSAIDs later in pregnancy, are worse.

• Paracetamol remains the first-line analgesic and antipyretic in pregnancy.
• Use the lowest effective dose for the shortest time needed, as for any drug in pregnancy.
• Tell worried patients that sibling comparisons show no link with autism or ADHD.
• Avoid switching to NSAIDs, particularly after 20 weeks, because of fetal kidney and ductus risks.
• Untreated maternal fever is itself a risk; do not withhold treatment.

Why it matters

It gives important reassurance against a public fear that has been pushing pregnant women away from the analgesic that guidelines recommend first.

Don't overread it

This is observational; sibling designs remove shared confounding but cannot exclude factors that differ between pregnancies.

The statistics, in plain English

A hazard ratio of 1.00 with a 95% CI of 0.91 to 1.11 means no difference, and rules out any large increase in risk. A sibling comparison controls for everything siblings share, such as genes and home environment. The pre-pregnancy 'negative control' showing a similar association is strong evidence the original signal came from confounding.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.