Sensitivity and Specificity of the Social Communication Questionnaire in a Diverse Clinical Sample

Background: Screening questionnaires precede a full evaluation to determine the likelihood of an Autism Spectrum Disorder (ASD) diagnosis. They are inexpensive, relatively easy to administer and are used by a variety of professionals across general (e.g., pediatrician office) and clinically referred (e.g., early intervention agency, diagnostic clinics) settings. When used in pediatric populations, developmental screening questionnaires are most often completed by parents and are intended to gather information about emerging developmental abnormalities that may prompt further assessment by a specialist. Ideally, the results of these questionnaires can inform clinical triage and evaluation processes. However, a review of several ASD-specific screening questionnaires suggested that they often show low sensitivity and are in need of higher levels of scientific scrutiny (Norris & Lecavalier, 2010). The Social Communication Questionnaire is a 40-item screening measure completed by parents for identifying symptomatology associated with ASD (SCQ; Rutter et al., 2003). Despite its popularity, there is limited research that examines how the SCQ functions in diverse clinical populations (Baio, 2018).

Objectives: This study provides a unique opportunity to understand better the sensitivity and specificity of the SCQ within a large, diverse clinical sample in New Mexico with Hispanic and Non-Hispanic populations.

Methods: We compiled a database using information from clinical evaluations completed at the University of New Mexico Center for Development and Disability. The Autism Spectrum Evaluation Clinic completes comprehensive diagnostic evaluations of children who are referred for concerns related to ASD. Our preliminary sample included children between ages 2-18. We conducted sensitivity and specificity analyses across age and ethnic groups using the SCQ published cut score of 15. Fisher exact tests and logistic regression with Firth’s penalized likelihood were used to assess heterogeneity in sensitivity and specificity. Non-parametric tests were used to compare continuous variables.

Results: Our preliminary sample included 218 children with 126 (58%) receiving a diagnosis of ASD, 167 (77%) being males and 55% self-identified as Hispanic. ASD was diagnosed for 56% of Hispanic and 60% of Non-Hispanic children. Sensitivity was 0.86 (95% CI = 0.73, 0.95) for the No Language group, 0.50 (0.34, 0.66) for age 2-4 years, and 0.76 (0.61, 0.87) for age 5-18 years. Specificity was 0.40 (0.16, 0.68) for the No Language group, 0.57 (0.34, 0.78) for age 2-4 years, and 0.28 (0.17, 0.41) for age 5-18 years. Overall, sensitivity was not different for Hispanic and Non-Hispanic (P = 0.162) groups, but specificity was lower for age 5-18 Hispanic than Non-Hispanic groups age 5-18 years (P=0.023).

Conclusions: Results of this study indicate that sensitivity and specificity of both Hispanic and Non-Hispanic groups vary greatly across age groups. Further analyses are needed to determine additional factors (e.g., cognitive ability, lower cutoff score) that impact sensitivity and specificity in this clinical sample that can be used to maximize the functionality of the SCQ to inform better clinical referral and triage decisions in diverse populations.