Although ASD treatment-outcome research is increasing at a rapid pace, there remains a need for psychometrically sound measurement tools that are sensitive to change in these core symptom areas. The Autism Impact Measure (AIM) was developed to assess change in clinically relevant features of core ASD symptoms. The AIM assesses frequency of symptom occurrence using a 2-week recall period, enabling measurement of incremental change over short periods of time. In addition, the AIM assesses the functional impact of ASD symptoms, allowing for identification of symptoms that are associated with the greatest functional impairment. The measure includes 41 parent-rated items, each requiring two corresponding 5-point ratings (frequency and impact). This two-pronged approach has the potential to inform both treatment outcome research and clinical practice.
The purpose of the current study was to examine the psychometric properties of the AIM using a large, well-defined, and nationally representative sample of children with ASD.
The AIM was administered to 390 children (ages 2-16) with ASD across 8 sites participating in the Autism Treatment Network. All participants met DSM-IV criteria for ASD, and met or exceeded clinical cut-off for diagnosis on the Autism Diagnostic Observation Scale (ADOS). For a subset of participants (n=70), the AIM was re-administered within 3 weeks following initial administration to establish test-retest reliability. For another subset of participants (n=65), ratings were obtained from two caregivers to establish inter-rater reliability for the measure across reporters. Construct validity was investigated by examining the strength of association between AIM scores and scores on other scales that purport to measure similar constructs (i.e., Vineland-II, Social Communication Questionnaire [SCQ]).
Initial item reliability estimates were very strong, with Cronbach’s alpha of .96 for the entire scale, .90 for the Frequency scale, and .95 for the Impact scale. Test-retest reliability was strong for both the Frequency scale (r=.84) and Impact scale (r=.82). Inter-rater reliability was moderately strong for both the Frequency scale (r=.66) and the Impact scale (r=.62). A subsample of participants (n=66) completing both the SCQ and AIM indicate strong evidence for construct validity, with scores on the SCQ significantly correlated with both Frequency (r = .60, p < 0.001) and Impact (r = .53, p < 0.001) scores. Both the Frequency and Impact scales were significantly correlated with the Vineland Adaptive Behavior Composite score (r = -.37, p < 0.001 and r = -.23, p < 0.001).
These results provide evidence that the AIM shows promising psychometric properties, with excellent internal consistency, test-retest reliability, inter-rater reliability, and relations to measures of related constructs. Next steps include continued examination of structural and construct validity, development of interpretable and valid subscale and summary scores, and examination of the AIM’s sensitivity to change in core symptoms following treatment.
We acknowledge the members of the Autism Treatment Network (ATN) for use of the data and the families who participated in the Registry. The ATN is funded by Autism Speaks and a cooperative agreement (UA3 MC 11054) from HRSA to the Massachusetts General Hospital.
See more of: Core Symptoms
See more of: Symptoms, Diagnosis & Phenotype