A large sibling-matched study challenges years of concern by showing that the apparent link between prenatal paracetamol use and autism or ADHD disappears after accounting for shared family-related factors, providing reassurance for its clinically indicated use during pregnancy.
A large population-based cohort study published in the journal JAMA Internal Medicine found that prenatal acetaminophen (paracetamol) use does not appear to increase autism or ADHD risk after accounting for familial confounding factors.
Familial factors complicate prenatal paracetamol research
Acetaminophen (paracetamol) is the most commonly recommended treatment for pain and fever during pregnancy and is considered the first-line option by obstetric societies and regulatory agencies worldwide. Because the drug crosses the placenta, however, researchers have long sought to determine whether it could influence fetal neurodevelopment.
Over the past decade, several observational studies have reported that prenatal paracetamol use may be associated with an increased risk of autism spectrum disorder (ASD) or attention-deficit/hyperactivity disorder (ADHD) in children. These findings have raised concerns among expectant parents, despite clinical guidance from organizations such as the Medicines and Healthcare Products Regulatory Agency (UK) and the World Health Organization (WHO), which continue to regard its use during pregnancy as reasonable and appropriate when clinically indicated.
One of the biggest challenges in interpreting the evidence is that the conditions prompting paracetamol use, such as fever, infection, inflammation, or pain, may themselves be associated with neurodevelopmental outcomes. This issue, known as confounding by indication, makes it difficult to determine whether any observed increase in risk is due to the medication or the underlying illness. In addition, ASD and ADHD have strong genetic and familial components, and studies that have accounted for these shared family-related factors have generally found that the apparent associations disappear.
The question is particularly important because treatment alternatives are limited. Non-steroidal anti-inflammatory drugs (NSAIDs), for example, are avoided during parts of pregnancy because they can increase the risk of complications including teratogenicity, oligohydramnios, and fetal kidney damage.
Previous systematic reviews have concluded that much of the existing evidence is of low quality and remains vulnerable to these sources of bias. The current study therefore sought to provide more robust evidence using a large population-based cohort from Hong Kong and a sibling-matched study design that better accounts for shared familial factors.
Researchers also examined whether any associations differed according to the timing, duration, or cumulative dose of paracetamol exposure. Exposure was classified as sporadic (first trimester only), intermittent (any two trimesters), or persistent (all three trimesters), while cumulative dosage was categorized as low, medium, or high based on the study population distribution.
Sibling comparison tested prenatal exposure
Here, the researchers examined whether taking paracetamol during pregnancy increases the risk of autism spectrum disorder (ASD) or attention-deficit/hyperactivity disorder (ADHD) in children. The sibling analysis included two cohorts: 124,333 children for ASD analyses and 97,285 for ADHD analyses.
They also carried out a conventional cohort analysis comparing offspring of women exposed versus unexposed during pregnancy, and a negative control analysis comparing risk in offspring of women exposed before or after pregnancy.
In each cohort, children from the same family with different paracetamol exposure histories were followed up for at least two and five years, respectively, to identify ASD and ADHD. Clinical diagnoses or prescription of ADHD-specific medication were used to identify these conditions.
Electronic health records were used to track prescriptions for paracetamol in pregnancy. The association of such diagnoses with paracetamol exposure during intrauterine life was stratified by timing, pattern of usage, and cumulative dosage.
Sibling analysis found no increased neurodevelopmental risk
The sibling-matched analysis found no evidence that prenatal paracetamol exposure increased the risk of either ASD or ADHD. Children who were exposed during pregnancy were no more likely to receive a diagnosis of either condition than their unexposed siblings, suggesting that shared familial factors, rather than the medication itself, explain the associations reported in earlier observational studies.
The findings remained consistent across a range of additional analyses. No increased risk was observed among children whose mothers used paracetamol for sustained periods (seven or more consecutive days), over a cumulative duration of seven or more days, or when the comparison was restricted to offspring of women who had not been prescribed other analgesic medications during pregnancy.
Conventional analyses pointed to residual familial confounding
Conversely, when paracetamol exposure was examined against ASD/ADHD diagnoses in a conventional cohort, there was an increase in risk in the exposed cohort. Similarly, the risk of ASD and ADHD was increased in offspring born to women exposed before or after pregnancy but not during pregnancy, compared to those not exposed in the corresponding periods.
The observation of positive associations in conventional exposure analyses suggests that unmeasured familial factors are responsible for residual confounding but are largely addressed in the sibling comparison.
The same applies to the associations with pre- or post-pregnancy paracetamol use, when fetal exposure could not have occurred, but not with use during pregnancy.
These findings corroborate earlier null findings from other sibling-matched studies in different populations (Nordic and Japanese), and support current clinical guidelines on the safety of paracetamol use in pregnancy under medical supervision.
Over the counter use and sibling design remain limitations
This population-based cohort study used prospectively recorded electronic health records and complementary analyses to account for unmeasured familial confounding. The sibling-matched cohorts adjusted for genetic and environmental confounding operating within families. The agreement with sibling cohorts across other populations supports the validity of this evidence.
The long follow-up period, up to 23 years, is another strength, as it enhances the reliability of the risk estimates. Objective and granular exposure data enabled detailed analyses of timing and usage patterns and dose-response relationships.
The study also had several limitations. The sibling-matched design allows some unmeasured confounding to persist or even be amplified. The sibling design assumes that paracetamol use during one pregnancy does not influence outcomes in subsequent pregnancies (“carryover effects”). Although the authors found no evidence of this, it cannot be entirely ruled out. The sibling-matched design might also limit generalizability.
Over-the-counter (OTC) use of paracetamol could cause misclassification of exposure, as could uncertainty about use of medication when prescribed. Prescription use could indicate a more difficult situation compared to OTC use. However, the researchers consider this an unlikely source of bias based on their further analyses, although they note that the applicability of these findings to OTC use remains uncertain, and because the pattern of OTC paracetamol use by any given mother probably remains constant across pregnancies.
Evidence supports current pregnancy prescribing recommendations
The researchers concluded that when clinically indicated, paracetamol use in pregnancy is unlikely to be associated with a higher risk of either ASD or ADHD in the offspring and is not a major risk factor for these conditions. The apparent associations seen in many earlier studies were likely due to familial confounding rather than the medication itself.
This suggests the need to use triangulation and sibling-matched cohorts to address familial confounding in such studies. The authors also note that future research should investigate other neurodevelopmental outcomes, such as intellectual disability.
