Prescription Benzodiazepine Use During Pregnancy and Risk of Attention-Deficit/Hyperactivity Disorder in Offspring

Abstract

Introduction Benzodiazepine (BZD) prescribing rates during pregnancy have risen over the last two decades. There is little research into the potential relationship between in utero exposure to BZD and offspring risk of attention deficit/hyperactivity disorder (ADHD), although there is some evidence of negative neurodevelopmental outcomes. Objectives and Approach We used comprehensive linked administrative data to investigate the association between maternal use of prescription BZDs during pregnancy and ADHD in offspring. We included mother-newborn dyads in Manitoba born from 1996-2012, with follow-up through 2017. BZD exposure was defined as 2+ prescriptions between conception and delivery. We matched exposed children to unexposed children to account for differences in characteristics between women who used BZDs during pregnancy versus non-users. Several sensitivity analyses addressed the potential for residual confounding, including a negative control group and a group made entirely of recent users of BZDs. Cox Proportional Hazard Regression Models were used to estimate the risk of ADHD among offspring. Results Among 495 children with at least two BZD exposures throughout pregnancy, 25.4% (n=452) had a diagnosis of ADHD compared with 18.0% (n=68) of children not exposed (adjusted HR 1.91, 95% CI 1.35-2.69). However, the association was unchanged in the negative control group analyses (aHR 1.76, 1.09-2.86), and not significant in the recent users of BZD (aHR 0.91, 0.63-1.30). Mother’s history of ADHD and teen births were also associated with ADHD in offspring. Conclusions In a large population-level analysis, in utero exposure to prescription BZDs during pregnancy appeared to increase risk of ADHD. However, sensitivity analyses suggest this relationship was likely due to residual confounding. The power to link data for the whole population and across generations enabled powerful sensitivity analyses that alter the initial inference.