The Autism Diagnostic Interview-Questionnaire (ADI-Q)

Background: The Social Communication Questionnaire (SCQ-L; Rutter, Bailey, & Lord, 2003) is a parent-rated autism screener that has been widely used in research. However, the SCQ-L is based on past behavior and like many ASD screeners, the utility of the instrument has been found to be affected by age and developmental level (Eaves et al., 2006; Chandler et al., 2007; Charman et al., 2007; Wiggins et al., 2007). Moreover, recent analyses of NDAR data has found the SCQ’s sensitivity and specificity scores to be lower than originally reported; at the recommended cutoff score of 15, a sensitivity value of .64 (CI₉₅ : 0.57; 0.69) and a specificity value of .72 (CI₉₅ : 0.62; 0.81) were reported by Barnard-Brak and colleagues (2015). To address the limitations of the SCQ, a new caregiver-reported screening questionnaire was developed.

Objectives: The goals of this study are to develop and test an ASD research screening instrument that can be used to rapidly screen verbal children and adolescents, ages 5 to 18 years.

Methods: Previously collected ADI-R scores from two large samples of children and adolescents with ASD and non-ASD diagnoses, available through the University of Michigan Autism and Communication Disorders Center (UMACC) and Kaiser Permanente Northern California (KPNC), informed initial item development. To test the new instrument, named the Autism Diagnostic Interview-Questionnaire (ADI-Q), a new sample of 977 males and 287 females with ASD and non-ASD diagnoses was collected at UMACC, KPNC, and Cincinnati Children’s Hospital Medical Center (CCHMC). Data were split into an initial test sample and a final validation sample. ADI-Q items were rank ordered by odds ratios and item distributions were examined to inform final item selection. Sensitivity and specificity values were analyzed using ROC analyses.

Results: The resulting instrument, the Autism Diagnostic Interview-Questionnaire (ADI-Q), consists of a 20-item measure. AUC values and sensitivity and specificity values were comparable to those of the SCQ-Lifetime. Results from the current analyses supported the use of unique algorithms and cutoffs for each sex.

Conclusions: The study presents a new ASD screening instrument, based on current behavior, that can have applicability to the participant recruitment efforts of a wide variety of research endeavors. Study results are discussed in light of the sex differences in the behavioral phenotype of ASD.