Trazodone

Pregnancy

Human pregnancy evidence is limited and mostly observational. In the 2023 ENTIS cohort, 221 trazodone-exposed pregnancies were compared with 869 SSRI-exposed pregnancies. For first-trimester major congenital anomalies, the study reported 1 of 169 trazodone exposures (0.6%) versus 19 of 730 SSRI exposures (2.6%); the adjusted odds ratio was 0.2 with a wide 95% confidence interval of 0.03 to 1.77. Pregnancy-loss incidence was 25% versus 18%, but the cohort did not find a statistically significant increase after its competing-risk analysis. The trazodone group also had more tobacco use and previous miscarriages, and 43% were lost to follow-up. An earlier controlled cohort combined 58 trazodone exposures with 89 nefazodone exposures, so it could not isolate trazodone-specific outcomes. A 2009 antidepressant cohort included only 17 trazodone exposures in its major-malformation analysis and found no individual antidepressant associated with a specific malformation. Miscarriage findings differ by population and comparison group. A 2009 Motherisk cohort found 122 miscarriages among 937 antidepressant-exposed pregnancies (13.0%) versus 75 (8.0%) in a comparison group, but the abstract did not provide a trazodone-specific estimate. A 2024 meta-analysis of 29 studies and 5,671,135 people found a higher pooled estimate versus the general unexposed population, which decreased when studies used a comparison group with maternal depression; the authors said there was little evidence of an association after accounting for depression. A 2025 UK cohort of 1,021,384 pregnancies found a modest class-level adjusted hazard ratio of 1.04; among people prescribed antidepressants before pregnancy and during the first trimester, the adjusted hazard ratio was 1.00. A 2025 review of 14 human pregnancy studies found no consistent evidence of higher congenital-malformation, stillbirth, or low-birth-weight rates, but noted possible miscarriage and therapeutic-abortion associations in some studies. These class-level findings do not give a trazodone-specific absolute risk. For other pregnancy outcomes, the 2023 ENTIS cohort reported preterm birth in 22 of 164 trazodone pregnancies (13.5%) and 57 of 691 SSRI pregnancies (8.3%, P=.05); birth weight and small-for-gestational-age rates did not differ. The authors noted more tobacco use and maternal conditions in the trazodone group. A 2010 study of 1,243 antidepressant exposures reported a class-level increase in preterm delivery, but no significant difference in small-for-gestational-age rates or mean birth weight; it did not isolate trazodone. In a 2013 Medicaid cohort, the trazodone-specific preeclampsia estimate was 14 cases among 339 users, adjusted relative risk 0.63 (95% CI 0.38 to 1.05), not a statistically significant increase. A separate 2010 SGA analysis reported a signal in a mixed 'other antidepressant' group, mainly venlafaxine, not a trazodone-specific result. A 2026 FAERS analysis identified trazodone among 12 medicines flagged as possible gestational-diabetes signals after reviewing 1,137 potentially drug-related reports. The study is based on spontaneous reports and cannot establish incidence or causation. A 2026 linked-record cohort reported an autism association for exposure to at least one of 15 sterol-biosynthesis-inhibiting medicines, a group that included trazodone; its pooled estimate does not describe trazodone alone. A 2026 commentary argued that illness severity and other confounding could explain the findings. These are emerging, observational signals, not evidence of a causal trazodone effect. The FDA label also reports fetal resorption and other effects in rats at 7.3 to 11 times the adult maximum recommended dose and increased congenital anomalies in rabbits at 7.3 to 22 times that dose; these animal findings do not establish human outcomes.

Breastfeeding

Evidence during breastfeeding is limited. In the 1986 milk study, six women received one 50 mg trazodone dose; the authors estimated that a fully breastfed infant would receive less than 0.005 mg/kg, compared with 0.77 mg/kg for the mother. The study did not measure the active metabolite. LactMed's 2022 review found no published infant blood-level data and described only a few infant case reports, several with other medicines. A 2021 case report measured trazodone and its active metabolite in maternal serum, cord blood, and milk; the mother also took etizolam, and the infant had no reported drug-related effects at 1, 3, and 6 months. The DailyMed label reports transfer into milk, no data on effects on milk production, and limited postmarketing reports without an identified association with adverse effects in breastfed children. NHS SPS says trazodone may be used with caution and infant monitoring, but SSRIs or tricyclic antidepressants are preferred. It recommends watching for drowsiness, irritability, stomach symptoms, poor feeding, and weight gain.

What was tested

The pregnancy evidence includes prospective observational cohorts, health-record studies, a systematic review, and spontaneous-report analyses. No pregnancy trial was large enough to rule out small differences in major birth defects or miscarriage. The largest trazodone-specific comparative cohort includes 221 trazodone exposures and 869 SSRI comparators; older studies either included few trazodone exposures or combined trazodone with nefazodone. Newborn findings come from follow-up reports rather than detailed chart review. Breast-milk evidence includes one six-person milk-level study and a few infant case reports. The 2026 gestational-diabetes study used FDA adverse-event reports, while the autism study used linked health records and grouped trazodone with 14 other medicines.