Problems With Maternal Antidepressant Treatment and Neonatal Outcomes Study

The article by Oberlander et al¹ attempts to address potential perinatal adverse effects of selective serotonin reuptake inhibitor (SSRI) use during pregnancy while accounting for depression in the mother. However, we believe that some of the conclusions they made are not supported by the data presented.

In the conclusion in the abstract, 2 statements they made were not supported by the data. They conclude that (1) “ . . . exposure was associated with an increased risk of low birth weight. . . . ” The definition of low birth weight is lower than 2500 g² and in neither the introduction (ie, statement of purpose) nor the “Methods” section, did the authors indicate that low birth weight was an outcome that they investigated. They mention only average weight of newborns and the proportion falling lower than the 10th percentile (ie, small for gestational age). In fact, the rates of infants with weights less than the 10th percentile were 8.5%, 8.1%, and 7.4% in the 3 groups, which reveals the opposite of what the authors concluded. In the normal population, we would expect that 10% would fall at or lower than the 10th percentile (that is its definition). However, they found much lower rates, all of which are statistically significant (all P values <.05) when tested using the binomial test. Consequently, the proportions of babies who were small for gestational age were significantly lower in all 3 groups than what would be expected according to population norms. (2) “ . . . maternal illness severity was accounted for,” which was also incorrect, as they themselves stated in the “Comment” section. The complex description of propensity scores lacked validity and consequently were rather meaningless. Therefore, both of these statements are misleading, especially if an individual only reads the abstract.

Another major concern is that the authors have relied on statistical significance rather than clinical relevance. When sample sizes exceed more than 400, one begins to see statistical significance for even trivial differences between groups because statistical significance is a function of sample size. In the present case, the authors found 2 very trivial differences in birth weights of 32 g and 24 g to be “significant,” when these small differences have no clinical relevance. The statistical significance between groups is strictly due to the extremely large sample size, enough to make any difference “significant.” To claim that these minuscule differences may be of concern is unwarranted. If anything, the opposite may be true, that these results are reassuring in that the use of SSRIs in pregnancy does not affect the birth weight of the offspring.

In this study, Oberlander et al conducted a large number of tests but made no adjustment for multiple testing, without acknowledging that their results could all be random error. They also attempted to identify depressed untreated pregnant women but provide no solid evidence that they actually succeeded in doing so. Drawing conclusions based on such analyses would be considered tenuous at best and women and their health care professionals should not be alarmed by the results of this study. Failure to treat depression during pregnancy can have significant negative ramifications for both mother and child, as well as being the strongest predictor of postpartum depression.³ Consequently, the decision to treat depression in pregnancy with an SSRI should be made to ensure the best outcome for both the mother and baby.

Correspondence: Ms Einarson, The Motherisk Program, Division of Clinical Pharmacology/Toxicology, The Hospital for Sick Children, 555 University Ave, Toronto, ON M5G 1X8, Canada (einarson@sickkids.ca).

Financial Disclosure: None reported.