In her commentary, Einarson1 discusses a study by Nakhai-Pour and colleagues that found an increased rate of spontaneous abortion with antidepressant use in pregnancy.2 Einarson found similarly increased risks in a 2009 study of women who contacted teratology information centres.3 Despite this replication, she casts doubt on Nakhai-Pour and colleagues’ results, stating that “this study cannot make any definitive conclusion as to whether antidepressants increase the risk of spontaneous abortion,” and that “if there is a true increased risk of spontaneous abortion caused by gestational use of antidepressants, it is very small.”1
Einarson fails to mention a single strength of the study, such as the large population-based sample, prospective recording of filled prescriptions, and controls for potential confounders such as socioeconomic status, comorbidities and indicators of depression severity. She also raises doubts about statistical significance, citing an expected 5% false-positive rate and over 100 analyses. We counted 15 analyses; of these, 13 crude and 7 adjusted ORs were significant (p < 0.05). Thus, even with a 5% type I error, the association between antidepressant use and increased risk of spontaneous abortion appears robust.
The relation between antidepressants and spontaneous abortion meets many of Bradford-Hill’s epidemiologic criteria for causation, such as biologic plausibility, dose-response, temporal sequence and replication.
We disagree with Einarson’s characterization of the absolute risk increase of 2.8% (unadjusted) and relative increase of 1.68 (95% confidence interval 1.38–2.06) as “very small” and unconfirmed. Moreover, there is evidence linking antidepressants to cardiac defects, persistent pulmonary hypertension and a neonatal syndrome.4
Women facing depression in pregnancy need accurate and unbiased information on treatment efficacy. For mild to moderate major depression, efficacy of antidepressants differs little from that of placebo.5,6 Nondrug treatments such as psychotherapy do not have the same potential for harm; they should be offered first line.