Besides Normal Children, Can the Children and Adolescent Versions of Autism-Spectrum Quotient (AQ) Also Differentiate ASD from ADHD Children? a Validation Study of AQ in Hong Kong

Background:  

The Children and Adolescent versions of Autism-Spectrum Quotient (AQ-Child and AQ-Adol) are respectively 50-item, parent-report questionnaires. They had been validated for assessing autism spectrum disorder (ASD), differentiating children with ASD from normal children. However, so far, few study reported on their ability to differentiate ASD from ADHD (attention-deficit hyperactivity disorder). Both neurodevelopmental disorders were common referrals to and presentation in child psychiatric clinics. It was important to test for practical reasons if AQ could aid clinicians in making differential diagnosis between ASD and ADHD.

Objectives:  

This study aimed at testing whether the AQ-Child and AQ-Adol were not only capable in differentiating ASD children/adolescents from normal control, but also from ADHD children/adolescents.

Methods:  

Two Chinese-speaking samples were recruited from child psychiatric clinics and mainstream schools to examine the psychometric properties of AQ-Child and AQ-Adol respectively in Hong Kong. They were translated into Chinese via the standard procedure of translation and back-translation. The first sample included three groups of children of 4 to 11 years for validating AQ-Child: (1) community control (N=1,196), (2) ASD children (N=124), and (3) ADHD children (N=82). A second sample included adolescents of 12 to 18 years for validating AQ-Adol: (1) community control (N=809), (2) ASD adolescents (N=78), and (3) ADHD adolescents (N=51). The AQ-Child and AQ-Adol were filled in by parents of the recruited samples.

Results:  

The psychometric results found for AQ-Child and AQ-Adol were largely similar. Both did not find age and gender effects in their total scores. Factor analysis was able to replicate three of AQ-Child’s five subscales, namely, Social Skills, Attention to Detail, and Communication, and four of AQ-Adol’s five subscales, adding Imagination. The remaining subscale, Attention Switching, failed to be replicated in both AQ versions. The internal consistency of the total scores of AQ-Child and AQ-Adol were respectively 0.80 and 0.87, while test-retest reliability 0.80 and 0.95. There were significant group differences in the total scores of both AQ-Child and AQ-Adol across the three groups of participants with large effect sizes (>1.3) between ASD vs Control and ASD vs ADHD, but only small-to-moderate effect sizes (≤0.5) between ADHD vs Control. The AUCs for AQ-Child between ASD vs Control and ASD vs ADHD were respectively 0.91 and 0.87, while for AQ-Adol, the corresponding figures were 0.91 and 0.81. The performance of the five subscale scores of both AQ versions in the above psychometric indices varied. Instead, the total scores appeared to be the consistent good performers in them.

Conclusions:  

The psychometric properties of AQ-Child and AQ-Adol in Hong Kong would be comfortably regarded as good to excellent. They were highly compatible to those of the original and other previous studies in AQ. Particularly, the added contribution of this study was to indicate that both AQ-Child and AQ-Adol were able to differentiate competently ASD from ADHD, underscoring the specificity of the AQ for ASD.