Continuing your ADHD stimulant through pregnancy — the postpartum data doesn't show extra risk once you control for confounders

Bottom line

A retrospective cohort of 3,676 people with ADHD who were adherent to a psychostimulant before pregnancy found that continuing the medication during pregnancy was not associated with worse postpartum mental health once the analysis controlled for maternal age, gestational age, comorbidities, and pre-existing mental health diagnoses. The raw, unadjusted numbers looked worse for continuers — that’s the number that circulates as a warning. The adjusted numbers, which account for why some people continue and others don’t, show no difference.

The research

Xu et al. 2026 used the Merative MarketScan Commercial Claims Database (2011–2021) to identify 3,676 people aged 13–50 with an ADHD diagnosis and stable psychostimulant use before a singleton pregnancy delivered at 20+ weeks. Of these, 1,521 (41.4%) had no stimulant refills during pregnancy (discontinued), 1,899 (51.7%) had a proportion of days covered below 80%, and 256 (7.0%) maintained 80%+ coverage throughout. The outcomes were postpartum mental health-related emergency or inpatient admissions and new postpartum mood or anxiety diagnoses, both measured in the year after delivery.

Unadjusted, continuers had a higher rate of new postpartum mood or anxiety diagnoses than people who discontinued (19.3% vs 12.8%; rate ratio 1.46, 95% CI 1.17–1.83) — a real, statistically significant gap. After adjusting for maternal age, gestational age at delivery, maternal comorbidities, and pre-existing mental health diagnoses, that gap disappeared: adjusted rate ratio 1.01 (95% CI 0.59–1.71) for postpartum admissions, 1.30 (95% CI 0.97–1.74) for new mood/anxiety diagnoses — neither reaching significance. The strongest predictors of postpartum admission were pre-existing mental health diagnoses (rate ratio 2.60) and medical comorbidities (rate ratio 1.98) — not medication status.

Read the direction of that shift carefully: this is a case where confounding-by-indication ran toward making the medication look worse, and the adjustment reversed it. People who stayed on their stimulant through pregnancy were not a randomly selected group — adjusting for who they were removed the apparent effect. This is a retrospective claims-based cohort, not a randomised trial; residual confounding from unmeasured factors (ADHD symptom severity itself, for instance) can’t be fully ruled out.

What this means

If you’re pregnant or planning to be and weighing whether to stay on your stimulant, this study doesn’t answer the question for you — that’s still a conversation with your prescriber and OB, weighing your own history against the drug-specific safety literature. What it does do is take one specific fear off the table as a population-level finding: continuing your stimulant, on its own, was not the thing driving worse postpartum mental health outcomes in this cohort. Pre-existing mental health history and medical comorbidity were doing that work.

Postpartum mental health is one outcome. A separate 2026 study of 32,152 US pregnancies in women prescribed stimulants before pregnancy looked at complications during the pregnancy itself. After propensity matching, the 8,394 who kept taking stimulants were 1.2 to 3.1 times as likely to have prenatal maternal, placental and birth-related complications as women on atomoxetine or neither drug. The atomoxetine group had its own raised risks, for fetal growth and labour problems (Tuan, Sarkar, Griffin & Babinski 2026, Psychiatric Research and Clinical Practice). It is electronic health record data, and the abstract does not report severity or which complications drove the range.

The broadest summary so far is a 2026 meta-analysis of 25 studies on stimulant or atomoxetine exposure in pregnancy (Komargodski, Tillawe, Berlin & Matok 2026, AJOG MFM). Major birth defects looked 25% more common before adjustment (risk ratio 1.25) and were not raised after it (adjusted 1.04, 95% CI 0.97–1.12). Two signals survived adjustment: heart defects with methylphenidate specifically (adjusted 1.33, 95% CI 1.01–1.74) and a low Apgar score at five minutes (adjusted 1.21). Miscarriage, preterm birth, low birth weight, neonatal intensive care, and autism or ADHD in the child were not associated with exposure after adjustment. That narrows the question to two specific signals to raise with an OB.

Read next on this site:

• Women and ADHD — the foundational article on what’s different for women across the lifespan, including reproductive-stage transitions.
• ADHD medication — general medication-decision framework this briefing sits alongside.
• Menstrual cycle and ADHD medication — the estrogen-dopamine mechanism relevant across reproductive transitions.
• PMDD and ADHD — another reproductive-hormone intersection with ADHD medication decisions.