Reliability and Validity of the Turkish Version of the Social Communication Scale.

Abstract

The aim of this study was to evaluate the reliability and validity of the Turkish version of the Social Communication Scale (SCQ). Throughout 2010-2011, we evaluated children who were between 18-60 months of age in the outpatient clinic who were diagnosed as having autism (n=49) or pervasive developmental disorder not otherwise specified (n=18) according to the DMS-IV criteria. The control group consisted of 51children with developmental delay (25 mental retardation, 26 speech delay) and 71 children with typical development in the same age group. We used a one way analysis of variance and post-hoc Tukey HSD test to compare the SCQ scores of the groups. For the reliability and validity analysis the Cronbach alpha, item-total score correlations and test-retest correlations were used. Principal components analysis and varimax rotation were used to find the factor solutions. Receiver Operator Characteristic (ROC) curves were utilised to detect cut-off scores, sensitivity, specificity, and negative and positive predictive values. The analysis yielded 3 factors named Language, Reciprocal Social Interaction and Stereotypic Behaviors/Restricted Areas of Interest. The Cronbach alpha value of the total score was 0.88. The test-retest correlation was high (r=0.90, p<0.01). The SCQ Total, Language, Reciprocal Social Interaction and Stereotypic Behaviors/Restricted Areas of Interest scores were significantly different among the groups. The SCQ Total, Reciprocal Social Interaction and Stereotypic Behaviors/Restricted Areas of Interest scores were highest in the autism spectrum disorders group. The SCQ Language and Stereotypic Behaviors/Restricted Areas of Interest scores were not significantly different between the developmental delay and the typical development groups. ROC analysis indicated that the SCQ Total and Reciprocal Social Interaction scores differentiated very well between the autism spectrum and control group (area under the curve 0.89 and 0.91) while the Stereotypic Behaviors/Restricted Areas of Interest score differentiated less (0.72). Using a cut-off score of 14.5 for the SCQ Total score, sensivity was 0.84, specificity was 0.81, positive predictive value was 0.82 and negative predictive value was 0.83. Using a cut-off score of 7.5 for the SCQ Reciprocal Social Interaction score, sensivity, specifity, positive predictive value and negative predictive value were 0.91, 0.82, 0.83, 0.90, respectively. The results indicated that in children younger than 60 months of age the SCQ Total score can differentiate subjects with autism spectrum disorders from children with developmental delay and typically developing controls; however, the Reciprocal Social Interactions score was more sensitive and specific. On the other hand, the negative and positive predictive values indicated that there was a fair percentage of false negatives and positives. Our results showed that, particularly in younger children, the Reciprocal Social Interaction items were more specific and were more helpful for differentiation of the children with autism spectrum disorders. The availability of the SCQ as a shorter and more practical form of the Autism Diagnostic Interview, revised in Turkish, is important as both a research and clinical tool.