Ascertainment of Quantitative Autistic Traits in a National Survey Involving 22,529 Japanese Schoolchildren

Background:  The evidence to date favors the concept of the autistic spectrum, from clinical level to subthreshold traits, along which behavioral and cognitive impairments vary continuously. Furthermore, continuous distribution of ASD traits in the general child population was found by large-scale behavioral studies employing quantitative measures such as the Social Responsiveness Scale (SRS).

Objectives:  This study aimed to clarify the population distribution and factor structure of quantitative traits related to autistic syndromes in a large national population sample in order to examine the psychometric properties of such measurements and their implications for cross-cultural research. A recent Korean study suggested the highest ever reported prevalence for categorically defined autism spectrum disorders (ASD), and notably, symptom counts were found to be continuously distributed in that population, as has been observed in epidemiologic studies in the U.S., U.K., and Germany.

Methods:  The Japanese version of the SRS was completed by parents on their 6- to 15-year old children (n=22,529) via a nationwide survey to examine quantitative autistic traits in the largest population-based sample to have specifically assessed for such traits.

Results:  The SRS scores showed a skewed normal distribution in the Japanese population, high internal consistency, and no significant relation to IQ within the normal range of IQ scores. Principal components factor analysis supported the existence of a primary factor contributing to a substantial proportion of variance across all three DSM-IV criterion domains for autism.

Conclusions:  Our study provides strong confirmation of the dimensional nature of autistic symptomatology in the general population. The findings underscore the recognition that paradigms for categorical case assignment may rest on arbitrary cutoffs imposed on a continuous distribution, which can result in substantial variation in prevalence estimation when the measurements used in case assignment are not standardized for a given population.