Fetal Antipsychotic Exposure Cleared for Kids’ Neurodevelopmental Disorders?

The increased risk of neurodevelopmental disorders among children born to mothers taking antipsychotics is likely not due to in utero exposure to these drugs alone, a national birth cohort study found.

While an unadjusted model showed an elevated risk for offspring born to mothers who were prescribed an antipsychotic in the second half of pregnancy compared with unexposed offspring (pooled unadjusted HR 1.91, 95% CI 1.79-2.03), this risk was no longer meaningfully increased after adjusting for treatment indication, lifestyle behavior, proxies for severity of mental illness, and concomitant use of other psychotropic drugs (pooled adjusted HR [paHR] 1.08, 95% CI 1.01-1.17), reported Loreen Straub, MD, MS, of Brigham and Women’s Hospital in Boston, and colleagues.

“Although the observed 2-fold increase in risk of neurodevelopmental disorders is not causally related with in utero exposure to antipsychotic drugs, it does highlight the importance of closely monitoring the neurodevelopment of the offspring of women with mental illness to ensure that early intervention and support can be instituted when needed,” the group wrote in JAMA Internal Medicine.

“The benefits of antipsychotic treatment for pregnant women with severe mental illness are undisputed,” they added.

When broken down by individual antipsychotic type, there was only one agent — aripiprazole (Abilify, Aristada) — that appeared to drive this risk.

In the adjusted model, offspring born to mothers who filled a prescription for aripiprazole during the second half of their pregnancy had a 36% higher risk for any neurodevelopmental disorder (paHR 1.36, 95% CI 1.14-1.63), as well as for autism spectrum disorder (paHR 1.49, 95% CI 0.91-2.47), ADHD (paHR 1.36, 95% CI 0.98-1.89), speech or language disorders (paHR 1.61, 95% CI 1.28-2.02), and behavioral disorders (paHR 1.63, 95% CI 1.18-2.26).

Other antipsychotics, including haloperidol (Haldol Decanoate), risperidone (Risperdal), olanzapine (Zyprexa), and quetiapine (Seroquel), were assessed, but were not linked with neurodevelopmental disorders in children.

“The potential signal identified for aripiprazole requires replication in other data sources before causality can be assumed,” Straub’s group noted.

They also pointed out that this agent was FDA approved more recently than the other antipsychotics included in this analysis, “and therefore, may be prescribed preferentially to patients with more severe illness and/or to patients unresponsive to other antipsychotic treatment.”

“Although illness severity and treatment response cannot be directly assessed in the present study data, comparing the characteristics of patients treated with aripiprazole vs other antipsychotic drugs — including proxies for disease severity and number of different antipsychotic drugs taken before pregnancy — did not reveal clear differences, suggesting that this explanation is unlikely,” they wrote.

This analysis drew on data from publicly and privately insured women and their offspring in the nationwide Medicaid Analytic eXtract (2000 to 2014) and the IBM Health MarketScan Research Database (2003 to 2015). This included 2,034,883 and 9,551 children unexposed and exposed to antipsychotics on Medicaid, respectively, as well as 1,306,408 and 1,221 children unexposed and exposed in the private insurance group, respectively. Children were followed for up to 14 years.

“Exposed” children were defined as those whose mothers filled a prescription for any antipsychotic medication during the second half of pregnancy — beyond 18 gestational weeks, the period of synaptogenesis.

The most common treatment indications for both insurance groups were bipolar disorder, depression, and anxiety. Other prescription medications that were common among these women included antidepressants, anticonvulsants, opioids, benzodiazepines, and corticosteroids.

One limitation to the study was the researchers’ inability to assess for antipsychotic adherence, with prescription fills as the only form of measurement. As a result, the findings “could be biased toward the null,” Straub and team wrote.