Cross-setting replication of the associations between maternal health and autism

Autism spectrum disorder (ASD) is a neurodevelopmental condition with early-life origins. Maternal health conditions during pregnancy have been linked to autism risk, but most studies focus on single populations, limiting generalizability. In this study we examined whether associations previously reported in a Danish registry-based study hold in a U.S. cohort.

Methods

We analyzed electronic health records of children born between 2010 and 2017 at Kaiser Permanente Northern California (KPNC) and their mothers. Maternal diagnoses were classified as chronic or non-chronic and their potential associations with ASD diagnosis in the child were assessed using Cox proportional hazards models, adjusting for key sociodemographic, healthcare utilization, and comorbid maternal diagnoses. Methods were aligned with the Danish study to ensure comparability.

Results

Among 224,353 children in the KPNC cohort, 5,448 (2.4%) were diagnosed with autism. Of the 42 maternal diagnoses significantly associated with autism in Denmark, 38 were evaluable in KPNC, and 18 remained statistically significant after full adjustment. The majority of associations had point estimates consistent with the Danish study, particularly psychiatric and cardiometabolic conditions (e.g., bipolar disorder: HR = 1.62, 95% CI = 1.30–2.02; gestational diabetes: HR = 1.22, 95% CI = 1.12–1.32).

Conclusions

Despite demographic and healthcare differences, and some differences in effect magnitudes of maternal health-autism associations, 35 of the 38 associations found in the Danish study replicated, qualitatively (direction of the effect), in the U.S. cohort, suggesting robust cross-setting relevance. Further research is needed to explore underlying mechanisms and potential effect modifiers.