Clustering of co-occurring conditions in autism spectrum disorder during early childhood: A retrospective analysis of medical claims data.

Insurance claims codes may not prove a child truly had or did not have a condition. ASD diagnoses were not checked with gold-standard behavioral assessments, so some diagnoses may be inaccurate. Developmental-delay codes may have been used as placeholders for ASD rather than as separate conditions. The condition categories included common problems such as ear infections and allergies, which may inflate the proportion with at least one co-occurring condition. Children were followed from health-plan enrollment rather than exact birth date, because exact birth dates were not available. The sample was privately insured, so it may not represent all U.S. children. Some conditions, including certain psychiatric disorders, often appear later than early childhood, so they may be undercounted. The clusters were not validated in another sample, so they may not generalize. Using broad condition categories, not individual diagnoses, limits conclusions about specific conditions. The way the researchers grouped the data can affect the clusters found.

The authors reported no conflicts of interest.

Do not treat these clusters as clinically validated autism subtypes or as evidence that therapy should change. They come from insurance claims codes, not assessments, and the study did not validate them in another sample.