Potential Effect of DSM-5 Diagnostic Criteria on ASD Prevalence Estimates

Background:  The fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) will present revised diagnostic criteria for autism spectrum disorders (ASD).  Previous modifications of the ASD diagnostic criteria are thought to contribute to temporal changes in ASD prevalence. Future changes to ASD criteria have important implications for research, public health monitoring, clinical evaluation, and service delivery.

Objectives:  To assess the effect of the DSM-5 criteria on previous ASD prevalence estimates based on DSM-IV-TR criteria. A second goal was to compare children meeting the ASD case definition under both DSM-5 and DSM-IV-TR coding schemes with children meeting the ASD case definition under the DSM-IV-TR scheme only. 

Methods:  The Autism and Developmental Disabilities Monitoring (ADDM) Network estimated ASD prevalence among a population of 308,038 8-year-olds in 11 US communities in 2006, using information abstracted from health and/or educational records. Trained clinicians systematically reviewed the abstracted information and determined case status using a coding scheme based on DSM-IV-TR criteria for autistic disorder and pervasive developmental disorder-not otherwise specified (PDD-NOS). Of the 2,757 children that met the ADDM ASD case definition, 75.4% had a clinical diagnosis and/or a special education classification of autism/ASD noted in their records.  We operationalized several different coding schemes for the DSM-5 ASD criteria, and applied each scheme to the children meeting ADDM ASD case status. We compared the characteristics of those who met DSM-5 criteria to those who did not.

Results:  Using the best-fitting and most “inclusive” DSM-5 coding scheme, 21.5% (593 of 2,757) of children classified by ADDM as having ASD did not meet DSM-5 criteria for ASD. Children meeting ADDM criteria for autistic disorder were more likely to meet DSM-5 criteria than children meeting ADDM criteria for PDD-NOS  (89.6% vs 40.0%). Children classified by ADDM as having ASD who had a previous clinical diagnosis or special education classification of autism were more likely to meet DSM-5 criteria than those without a previous diagnosis or classification (83.0% vs 64.5%). The male:female ratio was slightly higher among children meeting than not meeting DSM-5 criteria (5.0:1 vs 4.0:1). Children classified by ADDM as having ASD with IQ≤70 were more likely to meet DSM-5 criteria than those with IQ>70 or unknown IQ (85.0% vs 78.6% vs 71.3%).

Conclusions:  The changes to the ASD DSM criteria have the potential to affect  many findings observed in epidemiological studies, to alter how clinical diagnoses are made, and have policy implications for using an ASD diagnosis as a criterion for services.  Although the DSM-5 ASD criteria do not differentiate between “subtypes” of ASD (i.e., Asperger syndrome, autistic disorder), the DSM-5 ASD criteria appear more similar to the current criteria for autistic disorder than PDD-NOS. Further work is needed to operationalize specific DSM-5 criteria for surveillance purposes and to ensure comparability across time points in surveillance systems.