Background: Recently great attention has been paid to the still unmet clinical needs of most adults with autism spectrum disorder (ASD) who live in the community, an increasing number of whom visit psychiatric clinics to seek accurate diagnosis and treatment of concurrent psychiatric symptoms. However, different from the case of children diagnosed with ASD in childhood, it is difficult in adults to identify the ASD symptoms underlying psychopathology and to differentiate ASD from other psychiatric disorders in general psychiatric practice. This study aimed to verify the utility of the Social Responsiveness Scale-Adult version (SRS-A), a quantitative measure for identifying ASD symptoms, in non-clinical and clinical adult populations in Japan.
Methods: The total sample aged 19 to 59 years consisted of a non-clinical population (n =592) and clinical population with and without ASD (n =142). We examined score distributions of the Japanese version of the scale, and the effects of gender, age, and rater on the distribution. We analyzed factor structure and internal consistency in the non-clinical normative sample, and analyzed convergent, divergent, and discriminative validities in the clinical sample. We applied receiver operator characteristic (ROC) analysis to determine optimal cutoff scores discriminating the ASD clinical population from the non-ASD clinical population.
Results: The score distributed continuously, which replicated findings in children. For non-clinical adults, except in men aged 19 to 24 years, we found no or few gender, age, or rater effects. Both single- and two-factor models were supported for adults. Total SRS-A scores demonstrated high internal consistency and capably discriminated adults with ASD from those with non-ASD psychiatric disorders such as major depressive disorder, schizophrenia, and bipolar disorder with an overlap across diagnoses. Moderate to high correlations of the SRS-A with other-rated ASD measures indicated sufficient convergent validity. Based on the ROC analysis, we recommend cutoff points by gender for use in clinical settings.
Conclusion: This study provides additional supportive evidence that the Japanese version SRS-A can reliably and validly measure ASD symptoms in non-clinical and clinical adult populations, and thus can serve as a useful tool for ASD research as well as for secondary screening in Japanese adults.