PATIENTS AND METHODS: We conducted a retrospective review of the medical records of all pregnant women presenting at Mayo Clinic's site in Rochester, MN, from January 1, 1993, to July 15, 2005, and identified 25,214 deliveries. A total of 808 mothers were treated with SSRIs at some point during their pregnancy. We reviewed the medical records of the newborns exposed to SSRIs during pregnancy to analyze their outcomes, specifically for congenital heart disease and persistent pulmonary hypertension of the newborn.

RESULTS: Of the study patients, 808 (3.2%) took an SSRI at some point during the antenatal period. Of the 25,214 deliveries, 208 newborns (0.8%) were diagnosed as having congenital heart disease. Of the 808 women exposed to SSRI during pregnancy, 3 (0.4%) had congenital heart disease compared with 205 (0.8%) of the 24,406 women not exposed to an SSRI (P=.23). Of the total number of deliveries, 16 newborns were diagnosed as having persistent pulmonary hypertension of the newborn, none of whom had exposure to SSRIs (P>.99).

Overall, data regarding the use of SSRIs during pregnancy have been reassuring. Most studies^3-9 have reported relatively consistent findings on the adverse effects of SSRI use during pregnancy, which include neonatal withdrawal syndrome, low birth weight, and preterm birth. Many studies^10-12 have indicated no major fetal malformations higher than the baseline population risk of 1% to 3% with use of SSRIs during pregnancy. One study¹³ has indicated increased risks for specific defects, whereas another study¹⁴ noted an association between overall SSRI use during pregnancy and 3 distinct types of birth defects.

Congenital heart disease (CHD) is recognized in approximately 0.4% to 1% of all live births,^15-17 and several studies have suggested that use of an SSRI, particularly paroxetine, may be associated with an increased risk of CHD. In particular, ventricular septal defects (VSDs) have been associated with use of paroxetine during the first trimester,^18-22 although a recent meta-analysis²³ that reviewed both previously published and unpublished data did not confirm this association. One report²⁰ showed a dose response in neonates exposed to paroxetine during the first trimester; infants exposed to paroxetine at dosages greater than 25 mg/d had an increased risk of major malformation or major cardiac malformation, whereas those exposed to lower dosages had no increased risk above baseline. Finally, a recent report²⁴ showed an increased risk of persistent pulmonary hypertension of the newborn (PPHN) associated with late-pregnancy exposure to SSRIs.

Because recent reports have shown an increased risk of VSD and PPHN in neonates exposed to SSRIs in utero, we sought to determine whether use of SSRIs during the prenatal period was associated with CHD and PPHN.

Only 3 of the 208 newborns who were diagnosed as having CHD had exposure to SSRIs in utero. The prevalence of CHD was similar in neonates born to mothers who used SSRIs at any point during their pregnancy (0.4%) compared with mothers who did not (0.8%; P=.23) (Table 2). Of the 25,214 deliveries, 16 newborns were diagnosed as having PPHN; however, none had known exposure to SSRIs (Table 2).

Several studies^18-22 have shown an increased risk of VSDs in neonates exposed to paroxetine during the first trimester. This finding led the US Food and Drug Administration to advise women to avoid paroxetine during pregnancy if possible and to change pregnancy labeling from a category C to D. Paroxetine is the only SSRI whose pregnancy category labeling has changed. However, a recent meta-analysis²³ that reviewed both previously published and unpublished data noted no association between paroxetine and cardiovascular defects.

In our study, 134 of the SSRI-treated mothers took paroxetine. None of the paroxetine-exposed newborns were diagnosed as having VSDs. In reviewing the 119 cases of paroxetine use during pregnancy for which all data were available, the average dose was 31.55 mg, with most exposures (76.47%) occurring at conception and continuing throughout pregnancy.

We found that SSRI use during the prenatal period is relatively low, occuring in 3.2% of our study population. We noted that doses of prescribed SSRIs are significantly lower if initiated during the second and third trimesters of pregnancy compared with those prescribed at the time of conception. This finding may demonstrate that some physicians may be hesitant to prescribe higher doses of SSRIs to pregnant women. It also raises the question of whether the doses prescribed later in pregnancy are clinically useful.

In light of this and other research, important clinical questions include whom to treat and how to formulate an informed consent discussion. An informed consent discussion should include the risks of depression during pregnancy, nonpharmacological treatment options for depression, possible risks of SSRI exposure to the fetus, potential adverse effects to the mother, and benefits of SSRI use for treatment of depression. The patient's ultimate decision and the reason for her decision should also be discussed. Treatment is a personal decision between the patient and physician, with the balance weighing toward the patient's decision. Furthermore, physicians should have a similar discussion with all women of reproductive age when considering treatment with an SSRI because most of our patients were taking SSRIs at the time of conception. Accumulating data demonstrate the safety of SSRI use during pregnancy in terms of congenital defects; SSRIs should be considered for use during pregnancy when the benefit of treatment outweighs the estimated risk posed by medication exposure.

Our study has several strengths. We reviewed data throughout several years at a single medical center. In addition, the obstetrics database used is prospectively maintained. Although this is a tertiary medical facility, most of the obstetrics population is community based. We reviewed all pregnancies during this period, which certainly decreased our selection bias and allowed us to look at prescription practices. Because drug exposure information in our data was not based on a patient's memory, there is no possibility of recall bias.

Limitations of our study should not be overlooked. We did not review demographic or clinical information, including use of other prescription or nonprescription drugs, tobacco use, or alcohol use, that may affect fetal outcomes. Determining adherence to prescribed SSRI therapy during pregnancy is problematic. A patient who had SSRI use documented or prescribed during pregnancy but who did not ingest the SSRI would be misclassified as an exposure. Our data were taken from physician prescriptions rather than pharmacy records; the former may document an increased risk of nonadherence. Furthermore, we did not review the actual amount of time the fetus was exposed to an SSRI.

The ability to diagnose these disorders is also a concern; although PPHN would be diagnosed soon after birth, a small VSD may be undetected. Therefore, we may have underestimated the total number of heart defects. However, our overall rate of 0.8% is similar to previous CHD estimates of 0.4% to 1.0%,^15-17 suggesting that we did not miss large numbers of outcomes.