Efficacy of the Social Communication Questionnaire in a Community-Based Sample of Toddlers

Background:  Previous research has demonstrated that ASD can be diagnosed in children younger than 2 years of age (Zwaigenbaum et al., 2015), yet the median age of diagnosis is currently over 4 years (Christensen et al., 2016). Given the AAP’s recommendation for routine screening at 18-24 months (Johnson & Myers, 2007), most early screening measures target that age range (e.g., M-CHAT-R/F assesses children 16-30 months [Robins, Fein, & Barton, 2009]). There are limited screening measures for 2- and 3-year-old children, further contributing to the under identification of ASD before age 4. While the Social Communication Questionnaire (SCQ; Rutter, Bailey, & Lord, 2003) was intended for children 4+ years, its effectiveness has been examined in children under 4. When prioritizing the accuracy of identifying ASD (i.e., sensitivity≥80%), the rate of false positives was extremely high (i.e., specificity at 25% for a cut-off of 11) in children at-risk for ASD (Oosterling et al., 2010). The SCQ does not appear to perform optimally in young children, but further investigation is warranted.

Objectives:  (1) Examine the effectiveness of the SCQ in a community-based sample of toddlers; (2) Maximize its efficacy for 2-3 year olds

Methods:  Children were ascertained through screening in primary care using the Infant Toddler Checklist (Wetherby & Prizant, 2002) or Early Screening for Autism and Communication Disorders (Wetherby et al., 2015). Both negative and positive screens were included in the current sample (N = 384). Parents completed the SCQ–Lifetime Form when their child was 24-47 months; on items probing about ages 4-5, parents were asked to consider the last 12 months. Children received a concurrent diagnostic battery to determine a best-estimate diagnosis of ASD (n=181), DD (n=91), or TD (n=112).

Results: A receiver operating characteristic (ROC) curve analysis resulted in an area under the curve (AUC) of .694. At the recommended cut-off of 15, sensitivity for ASD was 40.9% and specificity was 85.7%. Prioritizing sensitivity in this sample (i.e., 81.2%) resulted in a cut-off of 7 and specificity of 42.9%. ROC curves were also conducted on individual items and 6 items had an AUC>.60, which assessed reciprocal conversations (item 2), unusual hand and body mannerisms (15-16), nodding head for ‘yes’ (24), offering comfort (31), and reciprocal imaginative play (39). These 6 items were summed, which resulted in an AUC of .757. A cut-off of 1 had a sensitivity of 88.9% and specificity of 39.1%.

Conclusions: Results indicated that a lower cut-off is needed for young children and demonstrated a shortened version of the SCQ may be more effective for children 2-3 years. A 6-item composite exhibited moderately better sensitivity and comparable specificity than all 40 items at a cut-off of 7, and was substantially better than the known cut-off of 15. Both versions of the SCQ demonstrated a high rate of false positives indicating further assessment of ASD is warranted, but suggested utility for identifying ASD. Confirmatory factor analyses and examination of measure performance by developmental level and age will provide more insight into the most useful SCQ items in toddlers.