Prenatal Exposure to Delta 9-Tetrahydrocannabinol and Child Neurodevelopment: A Critical Windows Analysis.

Moore B.; Crume T.; Wilkening G.; Mueller N.; Dabelea D.; Sempio C.; Klawitter J.; Christians U. · The Journal of pediatrics · 2026

**Background & Methods** This prospective cohort study enrolled 460 mother-child pairs from the Healthy Start study and measured Δ9-THC and metabolites in maternal serum at approximately 17 weeks gestation (n=460) and 27 weeks gestation (n=434), as well as cord blood at delivery (n=238). Neurobehavioral outcomes were assessed at age 5 years using the Child Behavior Checklist (CBCL) for behavioral assessment and the NIH Toolbox for cognitive evaluation. Generalized estimating equation models were used to examine exposure-by-timepoint interactions to identify critical windows of prenatal cannabis exposure. **Key Findings** • Δ9-THC exposure prevalence decreased across pregnancy (8% at 17 weeks, 6% at 27 weeks, 5% at delivery), with exposure timing significantly modifying associations with neurodevelopmental outcomes. • Late gestational exposure (at delivery) demonstrated the strongest associations with adverse behavioral outcomes, including higher T-scores for attention problems (β=5.2, 95% CI: 2.3-8.0), ADHD symptoms (β=2.4, 95% CI: 0.9-3.8), and pervasive developmental problems (β=3.9, 95% CI: 0.8-7.1) at age 5 years compared to unexposed children. • Most neurobehavioral outcomes showed timing-dependent associations with prenatal Δ9-THC exposure, suggesting that third trimester exposure carries particular neurodevelopmental risk. **Dosage & Administration** Not reported. **Safety & Adverse Effects** The study identified behavioral and cognitive adverse effects associated with prenatal Δ9-THC exposure, particularly when exposure occurred at delivery. Documented effects include increased attention problems, ADHD symptoms, and pervasive developmental problems in early childhood. No other maternal or neonatal safety parameters were reported. **Evidence Quality** This prospective cohort study represents moderate-quality evidence with several limitations. Loss to follow-up occurred between enrollment and delivery (n=238 had cord blood measured of 460 enrolled). The study was observational without randomization, limiting causal inference. Confounding by maternal characteristics, concurrent substance use, and socioeconomic factors was not fully addressed. The relatively small number of exposed children at delivery (5% prevalence) may limit statistical power for subgroup analyses. However, the longitudinal design with biomarker-confirmed exposure and standardized neurodevelopmental assessments at age 5 provides stronger evidence than retrospective or self-reported exposure studies. The findings suggest neurodevelopmental vulnerability during late gestation but require replication in larger cohorts with comprehensive confounding adjustment.

More recent cannabis research