Utility and Feasibility of Using the Autism Symptom Interview: A Replication Study.

Background: The Autism Symptom Interview (ASI) school-age was developed as a brief phone parent interview about current ASD symptoms to facilitate recruitment and characterization of large-scale research samples. Its first validation study showed that the ASI distinguishes verbal children with ASD from those with other neurodevelopmental disorders with high sensitivity (0.82) and low specificity (0.62). The latter increased to 0.92 when combined with the Autism Diagnostic Observation Schedule-Second Edition (ADOS-2) (Bishop et al., 2017).

Objectives: We aimed to examine 1) the ASI feasibility in the context of an ongoing neuroimaging study, 2) its ability to distinguish those with ASD from those without, and 3) the individual variables that contribute to ASI elevated scores.

Methods: One clinician administered the ASI during an in-person interview, followed by a thorough developmental and psychiatric history. This was paralleled by a child assessment including an ADOS-2 and IQ testing conducted by an independent evaluator. The parent interviewer had knowledge of previous diagnoses/concerns, the child evaluator was “blind” to prior history/concerns. Following assessment/scoring, these independent evaluators discussed their observations and clinical impressions to reach a diagnostic consensus based on DSM-5 criteria. We examined data from 76 verbal children between the ages of 5.5 to 11.9 years (M=8.4±1.7) enrolled in an ongoing study. ASI sensitivity and specificity were assessed, both alone and in conjunction with the ADOS-2. Parent ratings on core ASD and associated symptoms were used to compare groups defined based on diagnostic consensus as well as on those based on the ASI total diagnostic cutoff score.

Results: Per diagnostic consensus, of the 76 children, n=36 were classified as ASD and n=40 as non-ASD, while n=42 met diagnostic criteria on the ASI (ASI+). The ASI distinguished children with ASD from non-ASD with relatively high sensitivity (0.78) and moderate specificity (0.67). When used with the ADOS-2, specificity increased to 0.85. As expected, on average the ASD group had higher ASI total scores than the non-ASD group, as well as elevated ADOS-2 scores and other parent ratings of ASD core symptoms. The diagnostic groups did not differ in parent ratings of associated psychopathology indexed by the Child Behavior Checklist (CBCL). Notably, dividing the children in those with and without elevated ASI scores, regardless of diagnosis (ASI+ and ASI-, respectively), indicated elevated internalizing, externalizing and total CBCL T scores in ASI+ vs ASI- with no differences in ADOS-2 and other ASD core symptom ratings.

Conclusions: Our results suggest the feasibility of administering the ASI in-person in the context of a research study. Our findings are consistent with prior results of ASI moderately high sensitivity and increased specificity when combined with the ADOS-2. Differences in ratings of comorbid psychopathology between groups defined by ASI cutoff, but not in those defined by ASD diagnosis, indicate that comorbid psychopathology may confound results from parent interviews and should be accounted for in the diagnostic process.