Reply to Correspondence “Beyond Aggregate Risk: Drug- and Organ-Specific Nuances in Counseling on Antidepressant Teratogenicity” (Tsai et al. 2025)

We thank the authors for their positive feedback on our recent publication “Synthesizing cohort study results to promote knowledge transfer of safety data regarding gestational antidepressant exposure and offspring congenital anomalies: A test of concept” (Moses-Kolko et al. 2025). We appreciate the commendation for our test-of-concept fact box that we used to illustrate a strategy to facilitate prescriber-patient communication regarding potential benefits and risks of use of antidepressants during pregnancy. The idea of a layered fact box is intriguing. Because many patient encounters are virtual, technology could indeed be harnessed to create and display digital image layers during a risk discussion (Tsai et al. 2025).

The authors raise the important point that we mention in our discussion; that presenting pooled relative risks for overall major congenital anomalies and cardiac anomalies can obscure specific risk data regarding individual drugs as well as anomalies in specific organ systems and subsystems. As the manuscript title implies, we presented the example of gestational antidepressant exposure and the risk of major congenital malformations overall and cardiac malformations as a test of concept. We agree that it would be meaningful to extend the example we provided in the manuscript to organ-specific malformations as well as specific drugs. Ideally, there would be ample studies which minimize confounding and misclassification bias available to generate pooled risk estimates on specific drug exposure-organ malformation outcomes for use in a fact box. As we mention in the manuscript, a current challenge for summarizing data for specific drugs, for new drugs, or for rare anomalies is the paucity of high-quality evidence (Covington et al. 2004).

The authors cite several studies they believe provide clinically actionable signals. One study is misquoted. The Huybrechts study (Huybrechts et al. 2014) did not report an organ-specific and agent-specific increase in risk. They reported an adjusted RR of 0.94 (95% CI 0.73–1.21) for the association between paroxetine and cardiac anomalies and an adjusted RR of 0.73 (95% CI 0.49–1.09) for the association with ventricular septal defects and 1.07 (95% CI 0.59–1.93) for the association with right ventricular outflow tract obstruction, specifically.

An important direction of pharmacoepidemiology research is the harmonization across studies of exposure and outcome definitions as well as methods to address confounding and bias (Richardson et al. 2025) to make replication and meta-analysis possible in future research. We eagerly anticipate such research advances that will generate greater confidence in drug and organ-specific signals that can be incorporated into clinical risk discussions as well as into clinically useful fact boxes.

K.F.H. reports being an investigator on grants to her institution from UCB, GSK, and Takeda, unrelated to this work. L.S. reports being an investigator on a grant to her institution from GSK, unrelated to this work. E.L.M.-K. and K.F. have no commercial or financial conflicts of interest.