Antenatal Acetaminophen Use and Attention-Deficit/Hyperactivity Disorder

Abstract

Attention-deficit/hyperactivity disorder (ADHD) is themost common childhood-onset neurodevelopmental disorder. Although it is highly heritable, ADHD has a complex etiology, andnoninherited factors also contribute.1 Early environmental exposures that may cause neurobiological dysfunction and confer risk for ADHD (eg, lead and other toxins) have been gaining increased attention of late.1 In this regard, an interesting new study2 in this issue of the journal has found preliminary evidence that prenatal exposure to a drug considered safe in pregnancy (acetaminophen, or paracetamol) may be associated with ADHD in childhood. The premise of the research question was that acetaminophen may act as a hormone disrupter and thus alter fetal brain development. This large prospective study,2 using the Danish National Birth Cohort, found that, compared with an unexposed reference group, children who had in utero exposure to acetaminophen were more likely to receive a subsequent diagnosis of hyperkinetic disorder, to use ADHD medication, and to have ADHD-like behaviors. Stronger effects were seen for those exposed to acetaminophen during more than 1 trimester and for those exposed for a greater number of weeks. Results did not differ according to report of fever or report of infectious or inflammatory conditions antenatally. This study2 has some notable methodological strengths. Assessment was prospective (ie, analgesic use was reported prior to the development of the outcome of interest), minimizing thepotential for recall bias. Liewet al2 usedavery large cohort (>60 000 forhyperkineticdisorderdiagnosis andmedication outcomes, and >40 000 for ADHD-like behavior outcomes),providingthemselveswith thepotential todetect small effects. Database-recorded ADHD diagnoses and prescriptions forADHDmedications are likely to be anaccurate reflection of its prevalence, and findings were robust to the inclusion of a wide range of covariates. However, although the findings are potentially important, caution shouldbeexercised in ascribing causation to statistical associations between prenatal risk factors and adverse outcomes.3 Indeed, causation cannot be inferred from the present observed associations, and Liew et al2 are right to pointout thata replicationof their study isneeded.Other types of research designs that go beyond observational epidemiologyarenecessary.Results shouldbe interpretedcautiously for the following reasons. Acetaminophen is the most common drug used during pregnancy.4 Indeed, 56%of theparticipants in this cohortwere exposed to thedrug;however, factors that influencewhotakes the drug require consideration. Liew et al2 highlight that residual confounding may have been a possibility (eg, by genetic factors, unmeasured maternal psychopathology, exposure to other medications, or indication for drug use). With regard to the indication for drug use, although it is a strength that fever, infections, and inflammatory conditions were accounted for in the study,2 thesemight not be the only reasons why pregnant womenmight take acetaminophen. Althoughanexposure-responserelationship is reported for all 3 outcomes, with increasing frequency of acetaminophen use (indexed by number of weeks of exposure) reported as showing stronger associations, the interpretation of this relationship is not straightforward. Firstly, there are several outcomes, bothwithin the risk ratio figures for ADHD-like behaviors (see their Table 3) and within the hazard ratio figures for hyperkinetic diagnosis and prescription of ADHDmedication (see their Table 4), whose 95% CIs cross 1, despite the ratios alone suggesting increased risk of adverse outcome (although results are more consistent overall when stratified according to indication for acetaminophen use [see their eTable 2] and pregnancy trimester [see their eTable 3]). Secondly, the gestational age at the time of acetaminophen use had to be assigned to the date of interview for over a quarter of mothers who could not recall their precise week of acetaminophen use. Thirdly, nearly a third of the eligible mothers were excluded for missing 1 or more telephone interviews, which means that the sample may not be representative because ADHD is heritable and so dropout may be differential according to the ADHD status of parents. Fourthly, and of critical importance for interpretation, it is not clear whether the number of weeks of exposure is an adequate reflection of the overall burden of exposure because information on the number of tablets taken was not available. Therefore, while stronger effects are reported for those exposed for a greater number of weeks, it is not clear what dosage was associated with the observed effects—whether these associations are with one-off use or with regular dosing. Related article page 313 Opinion