Correspondence between statistically derived behavior problem syndromes and child psychiatric diagnoses in a community sample

The correspondence between Diagnostic and Statistical Manual (3rd ed.) (DSM-III) diagnoses and statistically derived syndromes was examined within a community sample of children and adolescents in Puerto Rico. Specifically, the extent to which behavior dimensions, derived from the Child Behavior Checklist and the Youth Self-Report, corresponded to psychiatric diagnoses, derived from parent and child versions of the Diagnostic Interview Schedule for Children, was examined. The alternative approaches for assessing psychopathology in children and adolescents were compared against external validators. The results indicated a meaningful convergence between DSM-III diagnoses and statistical syndromes; however, a one-to-one correspondence did not emerge. Little evidence was found for diagnostic thresholds. There was no evidence of the superiority of either the statistically derived syndromes or the DSM-III diagnoses. The incorporation of a measure of impairment improved the validity of both approaches. Adding parental reports to the selfreports of adolescents yielded little gain in the validity of either the statistical or diagnostic approach. The implications for the definition and assessment of child and adolescent disorders are discussed.This research was supported by grant MH 38821 from the National Institute of Mental Health.The authors wish to acknowledge Glorisa Canino, Ph.D., and Maritza Rubio-Stipec, M.A., of the University of Puerto Rico, co-investigators in the epidemiologic project and to Barbara Bettes, Ph.D. for her contribution to preliminary analyses for this manuscripl. Portions of this paper were presented at the 38th Annual Meeting of the American Academy of Child and Adolescent Psychiatry, San Francisco, October 1991.