The spectrum effect: Convergence of clinical and neuropsychological characteristics in adults referred for autism assessment

Abstract

Clinical and neuropsychological profiles of adults referred for autism assessment are not thoroughly understood, and information derived from studies comparing autistic adults to neurotypical controls might not be accurate to infer on individuals in a real-world, clinical setting. 263 adults (aged 18–65 years, 70 % males) referred to our clinic with a suspected diagnosis of autism were clinically explored and administered neuropsychological investigations. The suspected diagnosis was either confirmed or rejected. We conducted multivariate and post-hoc univariate analyses of covariance (MANCOVA; ANCOVA), controlling for possible confounders, to evaluate differences between autistic and non-autistic participants and further associations. Effect sizes (partial η²) were calculated for significant results. There were no significant differences in age means and sex ratios. Both groups showed high Autism Questionnaire (AQ) scores, high schizoid, avoidant, and obsessive-compulsive scores, and mild depressive symptoms, without significant difference between groups. The effect sizes of differences ranged from negligible to small for measures of nonsocial cognition, but were large for a measure of social cognition, the emotion recognition 40 (ER40). Autistic and non-autistic participants present converging features across multiple measures. In this diagnostic setting, psychiatric and neuropsychological measures are helpful in identifying individual difficulties and strengths. However, most of them, including the AQ, are poor indicators of autism. Our results mostly aligned with previous research and showed that information derived from comparisons to neurotypical controls cannot be directly transferred to a real-world setting. Detected impairments in emotion recognition were fairly specific to autism, expanding on previous findings.