18690
Classifying Autism Spectrum Disorders By ADI-R: Separate Subtypes or Severity Gradient?

Friday, May 15, 2015: 11:30 AM-1:30 PM
Imperial Ballroom (Grand America Hotel)
H. Cholemkery1 and C. M. Freitag2, (1)Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, JW Goethe University, Frankfurt am Main, Germany, (2)Department of Child and Adolescent Psychiatry, Psychosomatics and Psychotherapy, JW Goethe University, Frankfurt a. M., Germany
Background:  The recently published DSM-5 comprises DSM-IV TR diagnoses Autism, Asperger´s disorder, and Atypical Autism under the umbrella concept of Autism Spectrum Disorder suggesting a continuum of symptoms with different grades of severity. The novel diagnosis social communication disorder(SCD) encompasses children comprising persistent social communication impairments without restrictive, repetitive behaviors.

Objectives: By using a cluster analytic approach, and building on results of previous studies (Spiker et al. 2002; Verté et al., 2011), this study aimed at describing empirically derived subgroups based on the Autism Diagnostic Interview-revised (ADI-R) algorithm items. We explored if these subgroups are more similar to the DSM-IV TR/ICD-10 or DSM-5 approach.

Methods: Sample size consisted of n=466 individuals with Autism (n=194), Asperger´s syndrome (n=158), Atypical Autism (n=114), aged 3 - 21 (M=10.43, SD=4.15), IQ≥35 (M=94.70, SD=20.60). Clinical diagnoses were based on ICD-10 criteria using a combination of ADI-R, Autism Diagnostic Observation Schedule (ADOS), Social Communication Questionnaire, and clinical judgement.

Statistical analysis: A single-linkage procedure identified 2 outliers. The ADI-R algorithm subscales were analysed by hierarchical cluster analysis.  Fusions were made by Ward´s method, the Euclidean distance was used as proximity measure. The number of clusters was determined by dendrogram. Adjustment was done by the K-means procedure. MANOVAs with cluster affiliation and algorithm scores were calculated for controlling the cluster solution. Quality was ensured by discriminant analyses. The cluster groups were compared for differences in age, gender, IQ, ICD-10 diagnoses, Vineland-II, ADOS-severity scores, and Child Behavior Checklist by (M)ANOVA.

Results: The hierarchical clustering method resulted in a 3-cluster solution. N=153 individuals were included in cluster 1, n=124 in cluster 2, and n=182 in cluster 3. A MANOVA showed significant differences between the three clusters for the ADI-R subscale scores (F(26)=54.65, p>.000, η2=.62). Impairments in reciprocal social interaction, and communication just like abnormality of development were mostly pronounced in cluster 2, followed by 3, and 1. For the domain of stereotyped pattern only significant post-hoc effects between cluster 2 versus 1, and 2 versus 3 were observed. Cluster 1 and 3 did not differ. Discriminant analysis confirmed that 96.3% of the cases were correctly allocated. The clusters did not differ with regard to age (F(2)=.241, p=.786), IQ (F(2)=.338, p=.713), gender (χ2(2)=5.028, p=.081), or ICD-10 diagnosis (χ2(6)=5.837, p=.442). No significant differences were found for CBCL (F(22)=.992, p=.473), Vineland-II (F(10)=1.260, p=.328), and ADOS severity scores (F(2)=.472, p=.624).

Conclusions:   The ADI-R algorithm scores for reciprocal social interaction, communication, and abnormality of development were distributed across the clusters with differing severity, supporting the DSM-5 concept of ASD. Only the domain of stereotyped behaviour did not differ between cluster 1, and 3. This might be due to the small number of items in this domain. The ASD phenotype as described by the 3 clusters seemed to be independent of demographic variables and cognitive abilities. Interestingly, there was no cluster group characterized by social-communication problems only, making it unlikely that children with ASD may be “misdiagnosed” having a social communication disorder. Findings support the current DSM-5 concept of ASD and SCD as different disorders.