To the Editor In the study by Boukhris et al,1 the numbers of exposed women were small, and the first and second/third trimester samples overlapped considerably (n=25), with only 6 new second/third trimester exposures. The sharp increase in hazard ratios (0.84 to 1.87) between the 2 exposure windows is not biologically persuasive. Hazard ratios decreased (16% to 44%) between crude and adjusted models, highlighting the magnitude of confounding. However, confounders such as maternal autoimmune disorders, parity, and fertility treatments were not examined, nor was paternal age (a robust risk factor independent of maternal age), a strong limitation considering the older age of exposed mothers.
Fombonne E. Prenatal Antidepressant Use and Risk of Autism Spectrum Disorders in the Children. JAMA Pediatr. 2016;170(7):711-712. doi:10.1001/jamapediatrics.2016.0745