Bioclinical profiles of autism and other developmental disorders using a multivariate statistical approach

To study the relationships between clinical and biological data that are necessary for physiopathological analysis in the field of developmental disorders, we developed a quantified grouping system, based on four developmental assessment parameters. Parallel with this clinical research, we developed electrophysiological procedures adapted to the pathology of autism. In this paper, we report the utilization of an original multivariate descriptive statistical approach (correspondence analysis followed by cluster analysis) that allowed us to identify different bioclinical profiles using these clinical and electrophysiological data conjointly. These profiles are believed to be evidence for different underlying cerebral dysfunctions. This procedure proved effective in identifying two main bioclinical dimensions in a population of 145 developmentally disordered children. These dimensions reflect the association of intellectual impairment and centroparietal electrophysiological reactivity on the one hand, and autistic behavior and temporal electrophysiological reactivity on the other. This study, performed on a large population of children using objective methods of data analysis, provides validation of numerous multidisciplinary studies of autism and other developmental disorders carried out on small samples of children.