Implications of Using the Social Responsiveness Scale in First-Time Diagnostic Assessment

Background: The Social Responsiveness Scale (SRS) is a 65-item, Likert scale assessment designed to measure ASD-specific symptoms. While this scale is commonly used in diagnostic assessment, several studies have shown that the SRS may be correlated with traits other than ASD. For example, some studies have indicated that the SRS may be associated with general behavioral problems, cognitive functioning level, sex, and age.

Objectives: The purpose of this study was to determine whether there was an association between SRS scores and ASD diagnosis in a clinical pediatric population, when controlling for potential cofounding variables, including non-ASD behavioral problems and cognitive impairment.

Methods: This study included patients (ages 2-17) who presented for first-time diagnostic assessment. As part of their diagnostic assessment, parents of children completed the SRS in addition to other non-ASD clinical measures (e.g., CBCL). All children in our sample were diagnosed with one or more developmental disorders at the conclusion of their visit by a neurodevelopmental pediatrician or pediatric psychologist. Analysis of SRS total T-scores was conducted on the full population of patients for which SRS results were available. Subsequently, a regression analysis was performed on a subset of this population, in order to assess the degree of association between SRS total T-scores and ASD diagnosis, when controlling for behavioral and demographic factors. The variables controlled for in this analysis included age, sex, cognitive level (IQ), and behavioral problems (measured by the Child Behavior Checklist Externalizing Behavior section, or CBCL-E).

Results: Within the full group with SRS results available (n = 289), mean SRS scores were higher for participants diagnosed with ASD (t = 3.242, p = .001). However, both the mean T-scores of the group diagnosed with ASD and the group diagnosed only with non-ASD developmental disorders fell within the range of “moderate impairment”. When a regression model was fit to the subset of participants with IQ and CBCL-E scores available (n = 61), only behavioral problems (CBCL-E T-score) had a statistically significant relationship with the SRS total T-score (coefficient = .776, p < .001). ASD diagnosis had a small positive association with SRS score (coefficient = 3.580), but was not statistically significant. Subsequent analysis of the correlation between the SRS and CBCL-E found that the CBCL-E explained about 43% of the variance in the SRS scores.

Conclusions: The results of this study indicate that SRS scores are highly correlated with non-ASD behavioral problems, as measured by the CBCL-E, in a clinical pediatric population. In addition, while SRS scores were higher for participants diagnosed with ASD, both the ASD and non-ASD groups had scores that were elevated to clinically significant levels. Overall, this study supports the interpretation of the SRS in the context of a broader behavioral phenotype at first-time diagnostic visits.