Abstract

Aim: To identify pre- and perinatal risk factors for autism.

Method: Case control study. We matched names of patients from North Dakota who met DSM criteria for autism, a pervasive developmental disorder, and autisticdisorder with their birth certificates. Five matched controls were selected for each case.

Results: Univariate analysis of the 78 cases and 390 controls identified seven risk factors. Logistic modeling to control for confounding produced a five variable model. The model parameters were χ² = 36.6 and p <0.001. The five variables in the model were decreased birth weight, low maternal education, later start of prenatal care, and having a previous termination of pregnancy. Increasing father's age was associated with increased risk of autism.

Conclusion: This methodology may provide an inexpensive method for clinics and public health providers to identify risk factors and to identify maternal characteristics of patients with mental illness and developmental disorders.