Despite our appreciation for the work of Calderon-Margalit et al carried out recently, we wonder whether the drawn conclusions are legitimate and the estimation of odds ratios (OR) are realistic, in consideration of its methodological limitations.

The authors aimed to explore benzodiazepine and selective serotonin reuptake inhibitor effect on pregnancy outcomes. Primary outcome of the prospective cohort study was the odds of preterm delivery among women exposed to psychotropic medications during pregnancy, in comparison to women who were not exposed.

The authors reasonably carried out a multivariate analysis adjusting results for potential confounders and/or effect modifiers such as maternal age, race, years of maternal education, marital status, smoking during pregnancy, preeclampsia, and singleton/multiple pregnancy.

Their results showed that maternal use of benzodiazepine in pregnancy was associated with an adjusted OR for preterm delivery as large as 6.79 (95% confidence interval [CI], 4.01–11.5; P < .001). The odds of preterm delivery among exposed vs unexposed varied by gestational age at first drug administration. Adjusted OR before or during first trimester was 5.15 (95% CI, 1.34–19.8; P < .017), in the second trimester was 7.39 (95% CI, 2.35–23.2; P < .001), and in the third trimester was 10.1 (95% CI, 4.84–21.1; P < .001). The overall maternal use of selective serotonin reuptake inhibitor was associated with a small increase in the risk of preterm delivery, but this result was not significant (adjusted OR, 1.34; 95% CI, 0.76–2.36).

Our concerns about these results mainly regard possible confounding factors or effect modifiers neglected in the multivariate analysis. In fact, preterm delivery is thought to be associated with infection and inflammation. Despite previous clinical trials failing to demonstrate the effectiveness of antibiotic therapy at reducing preterm delivery incidence, strong evidence suggested that infections could contribute to 25% of preterm birth. Based on the latest findings, it seems unreliable to estimate a risk of preterm delivery in a reference population, without collecting information on genital tract infections and vaginal flora variation. Furthermore, genetic factors have been demonstrated to predispose to preterm delivery. Women with a previous preterm delivery are 5.6 times more exposed to recurrence of preterm than women without a history.

Although neither infections nor genetic factors could be considered causal factors for preterm delivery, we truly believe those factors should be always evaluated when preterm delivery is considered as primary outcome.