Maternal Obesity/Diabetes, Plasma Branched-Chain Amino Acids, and Autism Spectrum Disorder Risk in Urban Low-Income Children: Evidence of Sex Difference.

Maternal branched-chain amino acid levels were measured only once, 24 to 72 hours after delivery, in a non-fasted state, so they may not reflect the prenatal steady state. The study did not measure the children's branched-chain amino acid status at diagnosis, family history of autism, or full medical histories of both parents. No systematic screen for autism was conducted, and diagnoses came from electronic medical records during a change from DSM-IV to DSM-5 and ICD-9 to ICD-10, which may make diagnoses inconsistent. The sample was a predominantly urban, low-income, minority Boston population, so the findings may not generalise to other groups. Only 89 children had autism, which limits sex-specific and dose-response analyses. Obesity and diabetes were combined into one exposure, so their separate contributions cannot be separated. Branched-chain amino acid levels were split at the median, not at a clinically meaningful cutoff. Some stratified analyses were not adjusted for covariates because the models did not converge. As an observational cohort, it cannot show that maternal obesity, diabetes, or branched-chain amino acids caused autism.

The supplied text does not include a conflict of interest declaration or sponsor role statement. The journal header lists N.I.H., non-U.S. government, and U.S. Public Health Service research support.

Do not read maternal branched-chain amino acids as a cause of autism. The amino acids alone showed no association, and they were measured only once after delivery in a non-fasted state.