Differential Diagnosis of ASD Subtypes and ASD v. Nonspectrum Disorders in a Clinic-Referred Sample: Application of Two Diagnostic Approaches

Background: Diagnostic distinction among autism spectrum disorders (ASD) has been a topic of considerable debate, with some clinicians and researchers questioning the validity of distinctions across subtypes. While clinicians, both experienced and inexperienced, generally make reliable diagnostic distinctions between cases of Autism versus non-ASD, interrater reliability decreases significantly for Autism v. nonautistic ASD (Klin et al., 2000), particularly for inexperienced clinicians. Application of DSM-IV criteria improved interrater reliability for inexperienced clinicians. Inconsistent and varying diagnostic schemes used by different researchers further complicate the debate (Volkmar & Klin, 2000), and there are criticisms regarding the shortcomings of current DSM-IV/ICD-10 definitions. Researchers have questioned the validity of giving Autistic Disorder priority over Asperger Syndrome (AS) in making diagnoses. Researchers at Yale developed a new set of criteria for AS based on specific behavioral features, without precedence given to Autistic Disorder. Using a sample selected to have a very high probability of having higher functioning autism (HFA) or AS, they found assignment of AS, HFA, and PDD-NOS varied greatly across three diagnostic systems (DSM-IV, presence of language delay, and the New System). Application of the New System resulted in more valid diagnostic distinctions between HFA, AS, and PDD-NOS (Klin et al., 2005). Interrater agreement for the New System was not calculated, as diagnoses were determined through consensus.
Differentiating milder PDD-NOS from non-ASD disorders also proves difficult. In a study of DSM-IV/ICD-10 criteria, only a limited number of items differentiated PDD-NOS from non-ASD disorders (Buitelaar et al., 1999). Anecdotally, many clinicians with expertise in ASD report that they see many children who have been misdiagnosed ASD by other, less-experienced clinicians. In the ASD clinic from which the current sample is taken, close to 50% of children with previous ASD diagnoses were determined not to have ASD by the evaluation team.
Objectives: To examine interrater agreement using three diagnostic systems for diagnosis of HFA, AS, PDD-NOS, and nonspectrum in a clinic-referred sample.
Methods: Clinical records from 25 children referred to an ASD clinic for diagnostic evaluation were reviewed independently by two psychologists experienced in ASD, blind to diagnosis. Inclusion criteria were VIQ and NVIQ ≥ 70 and the following measures administered during evaluation: ADOS, SCQ, BASC-2, and a checklist of developmental history including age of first concern, first words, and first phrases.
Results: Results will be reported on interrater agreement on case assignment using strict DSM-IV criteria, and then again applying Yale’s New System for AS. Data will be presented on Percent of Observed Agreement (PO) on diagnosis. Ratings of clinician level of confidence in diagnosis also will be presented and compared across diagnostic systems.
Conclusions: Consistency in defining and applying diagnostic systems for ASD and ASD subtypes is critical for replication and cross-site collaboration. This study provides evidence for validation of standard criteria for differential diagnosis of ASD, within ASD subgroups and between ASD and non-ASD disorders. Whereas past research selected samples based on high-probability of HFA or AS, this study provides support for application of standard criteria to a less straightforward, clinic-referred sample.