Citalopram

Pregnancy

Citalopram-specific malformation evidence conflicts. The 2017 meta-analysis included eight studies and 1,507,896 participants; the pooled odds ratio for major malformations was 1.07 (95% CI 0.98-1.17), and the pooled cardiac estimate was 1.31 (0.88-1.93), falling to 1.03 (0.84-1.26) after excluding an outlier. The 2017 Quebec Pregnancy Cohort included 18,487 depressed or anxious pregnancies ending in live singleton births and reported a citalopram aOR of 1.36 (1.08-1.73) for major malformations, with 88 exposed cases; its organ-specific comparisons included musculoskeletal defects and craniosynostosis. The cohort's live-birth restriction and multiple comparisons, along with the meta-analysis's heterogeneous observational studies, limit interpretation. The Quebec paper disclosed a lead-author consulting role as an expert for plaintiffs in antidepressant litigation. MotherToBaby's miscarriage statement is not verified as citalopram-specific by its cited Andersen 2014 study. Andersen analyzed SSRIs as a class: among 22,884 first-trimester SSRI exposures in 1,279,840 pregnancies, the adjusted miscarriage HR was 1.27 (1.22-1.33); people who stopped SSRIs before pregnancy also had an elevated HR of 1.24 (1.18-1.30). Martin et al. 2025 analyzed 1,021,384 UK pregnancies and 73,540 first-trimester antidepressant prescriptions; the adjusted HR was 1.04 (1.02-1.06), with adjusted absolute estimates of 13.1% without and 13.6% with prescriptions. Among people prescribed before pregnancy and continuing into trimester one, the aHR was 1.00 (0.98-1.03). These are class-level estimates, not citalopram-specific results. The 2024 VigiBase analysis found a citalopram disproportionality signal for fetal death, but spontaneous reports have no exposure denominator and cannot provide incidence. Its abstract does not verify MotherToBaby's seven-case count. A 2026 cohort across the UK, Norway, and Sweden reported class-level stillbirth and preterm associations; the adjusted stillbirth estimates were 0.34% in unexposed and 0.40% in exposed pregnancies, attenuating to aOR 1.07 (0.94-1.21) among mothers with depression or anxiety. The authors omitted drug-specific stillbirth estimates for privacy. The same article qualitatively associated citalopram with small-for-gestational-age birth. The exact citalopram estimate from its supplement was not checked and is not stated. A 2026 Finnish cohort compared SSRI users with people diagnosed with depression who did not use antidepressants and reported lower risks of some preterm and low-birth-weight outcomes, but higher low Apgar, breathing-problem, and neonatal-care-unit outcomes. These are SSRI-class results and use a different comparator. For PPHN, the Celexa label states that risks exist with SSRI exposure, including Celexa, and the label also notes risks associated with untreated depression. Huybrechts et al. 2015 studied 3,789,330 Medicaid pregnancies. Crude PPHN rates were 31.5 per 10,000 SSRI-exposed births and 20.8 per 10,000 unexposed births; after restricting to women with depression and high-dimensional propensity adjustment, the SSRI OR was 1.10 (0.94-1.29). In the primary-PPHN analysis, the adjusted SSRI OR was 1.28 (1.01-1.64). This is class-level evidence, not a citalopram-specific estimate. EMA's March 2022 signal addressed pulmonary hypertension generally; in September 2022 PRAC said evidence was insufficient to establish a causal association for new or worsening pulmonary hypertension. It should not be described as a citalopram-specific PPHN finding. Cornet et al. 2024 studied 280,090 term infants, including 7,573 exposed to an SSRI after 20 weeks. Delayed neonatal adaptation occurred in 11.2% of exposed and 4.4% of unexposed infants; the adjusted OR was 2.14 (1.96-2.32). The citalopram monotherapy subgroup included 825 infants and had an aOR of 2.29 (1.91-2.76). Sivojelezova et al. 2005 prospectively compared 132 citalopram-exposed pregnancies with disease-matched and unexposed groups; 125 participants used citalopram in at least the first trimester. Near-term citalopram exposure was associated with special-care nursery admission (RR 4.2, 1.71-10.26), but the cohort was small. For maternal postpartum hemorrhage, the Celexa label says SSRI exposure, particularly in the month before delivery, was associated with a less-than-two-fold increase in observational studies. Palmsten et al. 2013 studied 106,000 Medicaid pregnancies with mood or anxiety disorder. Current serotonin reuptake inhibitor users had PPH in 4.0% versus 2.8% among depressed nonusers; the adjusted class estimate was RR 1.47 (1.33-1.62). The paper reports that all individual SSRIs available for analysis, including citalopram, were associated, but the 4.0% rate is class-level and residual confounding could not be ruled out.

Breastfeeding

MotherToBaby and LactMed report low transfer of citalopram into milk and low infant serum levels in most published cases, with isolated reports of drowsiness, fussiness, irritability, restlessness, feeding difficulty, or weight change. LactMed advises monitoring for excess sleepiness, restlessness, irritability, poor feeding, and poor weight gain, especially in younger infants. Heikkinen et al. 2002 enrolled 11 mothers taking citalopram and their infants, with 10 matched controls. Infant plasma concentrations were very low or undetectable despite citalopram being measurable in milk, and infant development was reported as normal through one year. Lee et al. 2004 compared 31 breastfeeding mothers taking citalopram, 12 depressed untreated mothers, and 31 healthy controls; infant adverse-event counts were 3/31, 0/12, and 1/31, with no statistically significant difference. Den Besten-Bertholee et al. 2024 included 37 mother-infant pairs, 13 exposed to citalopram; infant citalopram plasma levels were undetectable in 9 of the 13 citalopram-exposed infants. These studies provide direct exposure data but have small citalopram samples. Heinonen et al. 2025 followed 97 children whose mothers used SSRIs during pregnancy; 22 were exposed to SSRIs through breastfeeding, but only 2 were exposed to citalopram. Adjusted IQ scores did not differ significantly between breastfed children with and without postnatal SSRI exposure. The study is not a citalopram-specific long-term estimate. The 2026 NeoWISE cohort examined feeding mode among newborns exposed to SSRIs in utero, not citalopram transfer through milk, so it is not used to infer a citalopram-specific breastfeeding effect. LactMed says citalopram, when needed, is not by itself a reason to discontinue breastfeeding, while NHS SPS advises caution, lists sertraline or paroxetine as preferred, and recommends infant monitoring. NHS public guidance says citalopram can be used while breastfeeding and passes into milk in small amounts. This is a difference in guidance emphasis; the primary citalopram studies are small and do not resolve rare or long-term outcomes.

What was tested

Pregnancy evidence includes citalopram-specific case-control and cohort studies, a citalopram malformation meta-analysis, SSRI-class miscarriage and PPHN cohorts, spontaneous-report analyses, and recent population cohorts. The Quebec malformation cohort used depressed or anxious live-birth controls; the 2025 UK miscarriage study compared first-trimester antidepressant prescriptions with no prescriptions; the 2026 Finland cohort compared SSRI users with diagnosed-depression nonusers and discontinuers; the 2026 UK/Norway/Sweden cohort used prescriptions or dispensations and sibling/paternal negative controls. Cornet et al. studied term infants and defined delayed neonatal adaptation using Apgar score, resuscitation, or neonatal intensive-care admission for respiratory support. Palmsten et al. used Medicaid dispensing records around delivery and postpartum-hemorrhage diagnoses. Breastfeeding evidence includes milk and plasma concentration studies, small prospective infant cohorts, case reports, and one small study of later cognitive testing. Heikkinen, Lee, and Den Besten-Bertholee provide direct citalopram exposure information; Heinonen's later-child analysis is predominantly other SSRIs. These designs cannot establish rare harms or isolate medication effects from all differences between treated and untreated families.