The Validation of the 3DI-Sva As a Diagnostic Tool for Autistic Spectrum Disorder

Background: The diagnosis of autistic spectrum disorder (ASD) requires the integration of information from multiple sources regarding social communication, language and repetitive stereotyped behaviours. Many adults with an ASD are not appropriately diagnosed in childhood. Efficient and accurate tools that can be utilised with adults to detect current clinical features in addition to those present in development are required to facilitate accurate diagnoses in adults.

The Dimensional, Developmental and Diagnostic Interview (3DI) is a parental interview designed to assess children’s behaviour and provide an accurate tool for scoring and interpretation. It has high test-retest and inter-rater reliability and accurately distinguishes between children with ASD and those with non-ASD disorders.

The Dimensional, Developmental and Diagnostic Interview: short form for adults (3DI-sva) has been developed to provide quick and accurate diagnostic information regarding adults.  It consists of 73 items covering three subscales (Social, Communication and Repetitive Stereotyped Behaviours), including four items on early development and is administered to the parents of adults being assessed for an ASD.  

Objectives: This study has two aims.

1. To validate the 3DI-sva as a tool for accurately discriminating between typically developing adults and those with an ASD.
2. To quantify appropriate minimum cut points for each subscale that would define an adult as being likely to have an ASD. 

Methods: In total, forty-six adults and one of their parents/carers were recruited. This consisted of thirty-three typically developing adults (17 males and 16 females) and thirteen adults with an established diagnosis of ASD.

Four interviewers were trained in the appropriate administration and coding of the 3DI-sva and they interviewed parents/carers of typically developing subjects by telephone. Each rater transcribed four of their interviews for subsequent inter-rater reliability analysis. Parents/carers for the ASD group were interviewed by telephone by one researcher only. 

Results: The mean ages in years (and standard deviations) of the groups were as follows; ASD 23.12 (3.92), typically developing males 23.24 (3.73) and typically developing females 21.81 (2.64).

Telephone interviews took forty-five minutes to complete.

Interviews for typically developing samples had inter-rater reliability scores over 0.98 as calculated by Pearson’s Correlation Coefficient.

Cronbach’s alpha was calculated to measure internal consistency of the three subscales and was above 0.7 for all domains across both groups with the exception of the communication domain in the ASD group (0.683).  All the original interview items were therefore retained.

ASD subjects scored significantly higher than typically developing adults across all three domains as calculated using a one-way multivariate analysis of variance (MANOVA).

Receiver operating characteristic curves were utilised to calculate cut off points to discriminate between typically developing and ASD adults. The 3DI-sva demonstrated an ability to discriminate with a sensitivity of 1 and a specificity of 0.87. 

Conclusions: The 3DI-sva is quick to use, has good inter-rater reliability, provides dimensional ratings of symptom severity and effectively discriminates between typically developing and ASD adults across all three subscales associated with a diagnosis of ASD.

It therefore has important utility as a diagnostic tool for adults.