Variation in the Prevalence of the Pervasive Developmental Disorders by Diagnostic Criteria

Background: The criteria used by professionals to diagnose autism and related disorders are defined by the Diagnostic and Statistical Manual (DSM) of the American Psychiatric Association (APA).  Autism was initially defined in the DSM as a form of childhood schizophrenia.  In 1980 (DSM-III), autism was listed under the category of “Pervasive Developmental Disorder” which included the more stringently-defined “infantile autism” and the more broad “childhood onset pervasive developmental disorder.” The PDD criteria were revised again for the DSM-IIIR in 1987, and again, with the DSM-IV in 1994 with a text revision (TR) in 2000.  The update in 1994 included the addition of Asperger’s disorder. Over the same time period, prevalence estimates changed from about 1/2000 to about 1/100 today. There has been significant debate regarding the extent to which DSM criteria changes have affected ASD prevalence estimates.

Objectives: To evaluate how the identification of children with an ASD varies by DSM version used.

Methods:  The Autism and Developmental Disabilities Monitoring (ADDM) Network has identified the   prevalence and described the population characteristics of children with ASDs in multiple areas of the United States.  ADDM Network clinician reviewers use a coding guide based on the DSM-IVTR criteria for autistic disorder and PDD (including Asperger’s).  The coding is applied to developmental and behavioral information abstracted from existing records of developmental evaluations for children who may or may not have been diagnosed with an ASD by a community professional. For this analysis, DSM-III, IIIR, and IVTR criteria for autism and PDD were applied to the same developmental and behavioral information abstracted from existing records.

Results: A total of 5,007 eight-year-old children were suspected of having an ASD in the year 2006 from areas of the US with a population of 280,423 eight-year-old children. Of these, 2,054 met DSM-IVTR criteria for autism and 2,641 for overall ASD/PDD.  When using DSM-IIIR criteria, 1,892 met criteria for autism and 2,223 for PDD. For DSM-III, 1,921 met criteria for autism and 2,232 for overall PDD.  Autism prevalence estimates per 1,000 children (95% confidence intervals) for DSM-IVTR, DSM-IIIR, and DSM-III are as follows: 7.3 (7.0-7.6); 6.7 (6.4-7.1); and 6.9 (6.5-7.2), with overall PDD prevalence estimates: 9.4 (9.1-9.8); 7.9 (7.6-8.3); and 8.0 (7.6-8.3).  There was no difference in overall PDD prevalence between DSM-III and IIIR criteria (p=.45), but PDD prevalence was 17.50% lower (p<.001) using DSM-III criteria and 18.98% lower (p<.0001) using DSM-IIIR criteria compared to DSM-IVTR. 

Conclusions: Given that ASDs are behaviorally-defined disorders, the diagnostic criteria impact the number of children identified with the condition(s) as seen when the different criteria are applied to the same population of children. These results indicate that a portion of the increase in identified ASD prevalence over time may be attributed to differences in community identification based on the criteria used to characterize the symptom constellation that makes up the PDDs.