A retrospective observational study by Tuan and colleagues used electronic health record data from the TriNetX Research Network to examine pregnancy outcomes among women with prior stimulant prescriptions. The database included patients who delivered between 2010 and 2022. All women in the analysis had received at least two stimulant prescriptions before pregnancy, allowing comparisons between those who continued stimulants, those who switched to atomoxetine, and those who discontinued treatment.
The final analytic sample comprised 32,152 pregnancies: 8,394 (26.1%) continued stimulants during pregnancy, 405 (1.3%) switched to atomoxetine, and 23,353 (72.6%) received neither medication during pregnancy. Exposure definitions required at least two prescriptions during pregnancy for the relevant medication; women with only one prescription were excluded. To account for duration and adherence, the investigators calculated the proportion of days covered (PDC) during pregnancy: mean PDC was 38.7% in the stimulant group and 22.5% in the atomoxetine group.
Outcomes assessed via electronic health records included maternal prenatal complications (gestational diabetes, gestational hypertension, pre-eclampsia/eclampsia), placental complications (placental abruption, placenta previa), fetal growth indicators (fetal growth restriction, large-for-gestational-age), labor and delivery complications (cesarean section, instrumental delivery, shoulder dystocia, induction of labor), and birth-related outcomes (preterm delivery, spontaneous abortion/miscarriage).
The authors used propensity-score matching to reduce baseline differences between comparison groups and logistic regression to estimate adjusted odds ratios (aORs) for individual outcomes.
The study reported multiple associations that varied by comparison group. The publication summarized that “The stimulant group was 1.2–3.1 times more likely to develop prenatal maternal, placental, and birth-related complications than the atomoxetine and control groups,” but a closer look at individual outcomes provides a more nuanced picture. Key findings are summarized in the sections below.
This comparison evaluated women who continued stimulant treatment during pregnancy against women who had a history of stimulant treatment before pregnancy but received neither stimulants nor atomoxetine in pregnancy. Using this design reduces some confounding compared with comparing stimulant users to never-users, but residual differences between women who continue versus discontinue medication may remain.
Increased risks associated with stimulant exposure (stimulant group compared with discontinuers) included:
Decreased odds observed with stimulant exposure included:
The magnitude of increased risks with stimulants in this comparison was generally modest; for example, a roughly 36% higher odds of pre-eclampsia/eclampsia and a roughly 34% higher odds of preterm delivery were reported when comparing continuers to discontinuers.
Comparing women who continued stimulants with those who switched to atomoxetine produced larger effect estimates for several outcomes. Reported increased odds with stimulant exposure included:
Decreased odds with stimulants vs. atomoxetine included fetal growth restriction, cesarean section, instrumental delivery, and induction of labor:
These larger differences should be interpreted cautiously because the atomoxetine group was small and had markedly different baseline characteristics from the stimulant and control groups.
When comparing atomoxetine exposure with discontinuers, several associations were reported:
Increased odds with atomoxetine: fetal growth restriction (aOR 2.29, 95% CI 2.00–2.62), large for gestational age (aOR 1.94, 95% CI 1.71–2.21), instrumental delivery (aOR 15.0, 95% CI 11.7–19.3), and induction of labor (aOR 9.49, 95% CI 7.90–11.4).
Decreased odds with atomoxetine: shoulder dystocia (aOR 0.63, 95% CI 0.47–0.85).
The investigators found no significant differences between atomoxetine and controls for other outcomes evaluated. However, the atomoxetine group had substantially higher prevalence of psychiatric and substance-related comorbidities, along with greater smoking and alcohol use, which complicates interpretation.
The most consistent associations across analyses were between stimulant exposure and increased odds of gestational hypertension, pre-eclampsia/eclampsia, preterm delivery, and miscarriage. The reported effect sizes were modest when stimulants were compared with women who discontinued treatment. A plausible biological mechanism exists: stimulants can raise blood pressure and cause vasoconstriction, which could potentially affect uteroplacental blood flow.
Nevertheless, this was an observational study, and associations do not prove causation. Women who continue stimulant treatment during pregnancy may differ from those who discontinue in ways not fully captured by electronic health record data, including illness severity, health behaviors, or access to care. These residual differences may confound observed associations.
Comparisons involving atomoxetine are especially difficult to interpret. The atomoxetine cohort was small (n=405) and had markedly different baseline clinical and substance-use characteristics; larger odds ratios observed in stimulant-versus-atomoxetine comparisons may reflect these baseline differences rather than true differences between medications.
An unexpected finding was that stimulant exposure was associated with lower odds of fetal growth restriction compared with either atomoxetine exposure or controls. The authors observed low rates for some outcomes and caution in interpreting these signals.
Clinicians should interpret these results as evidence of associations, not causal effects. The study provides useful, large-scale observational data indicating modestly higher odds of hypertensive disorders, preterm birth, and miscarriage with stimulant continuation in pregnancy, but it cannot definitively attribute these outcomes to stimulant medications.
Important limitations include potential residual confounding, limited PDC indicating incomplete or intermittent exposure, and a small, clinically distinct atomoxetine group that limits head-to-head comparisons. Decisions about ADHD treatment during pregnancy should be individualized, balancing the potential risks of medication exposure against the risks of inadequately treated ADHD for the pregnant person and pregnancy course.
The authors and the summary emphasize cautious interpretation and shared decision-making when considering ADHD medication management in pregnancy.