Our understanding of autism has seen a gradual shift from a categorically distinct clinical entity to a continuum of autistic traits that are distributed across the general population. According to this view, the autism spectrum condition (ASC) reflects an extreme of the normal variation of autistic traits. This shift has been helped by the introduction of the Autism-Spectrum Quotient (AQ), a widely used measure of autistic traits.
For the first time since the inception of the AQ in 2001, researchers from the Autism Research Centre (ARC) in the Department of Psychiatry have performed a large-scale systematic review of empirical AQ data over 13 years in 6934 individuals without a diagnosis of ASC. One of the key questions of interest concerns the average AQ score in the general population, or in people without a clinical diagnosis of autism. This research, published in Molecular Autism, represents the first step towards the establishment of a normative AQ score in the general population.
In this systematic review, the authors reported that the mean AQ score in the non-clinical population is 17, compared to 35 points in individuals with ASC. Despite the considerable gap of AQ scores between both groups, the authors also cautioned against using a rigid cut-off AQ score to define experimental groups for a few reasons. Firstly, the AQ was designed to be a descriptive measure of autistic traits, rather than a tool used to accurately diagnose ASC. Secondly, its diagnostic utility for ASC has not been rigorously evaluated as a psychometric measure. Finally, previous evidence has shown that the AQ lacks the sensitivity and specificity for population screening in the context of a formal diagnosis. Following that, the authors suggested that future studies could use the AQ to segment the population into bands of autistic phenotype (broad, medium and narrow), as that would be more in keeping with the descriptive nature of the AQ.
Interestingly, the study also revealed considerable fluctuations and variability in the reported mean AQ scores across the prevailing literature. This has important implications for research studies and raises fresh concerns about common assumptions of homogeneity in a ‘true’ representative sample, which might in turn lead to an attenuation of real group differences. To avoid risk of bias, the authors stressed the need for a careful reconsideration of methods that are currently used to define nonclinical and clinical groups.
We spoke to the lead author, Emily Ruzich, a PhD student from the ARC and the Cambridge Intellectual & Developmental Disabilities Research Group, about the motivation behind the study: “Because the AQ is now quite widely used, we felt it was important to stop and take stock of the measure and how it relates not only to people with an autism diagnosis but also to nonclinical volunteers. Over the course of the development and continued use of the questionnaire, we’ve gained a pretty good idea of how people with ASC score, but the question kept arising, even within our own group, about how a ‘true general population sample’ might score. We hope the reference norm will be useful for researchers collecting new data, as well as for those evaluating pre-existing database samples”.
Avenues for future research were also discussed. Presently, there are other widely used versions of the AQ, including but not limited to AQ-Adolescent and -Child, as well as cross-cultural versions that have been translated into multiple foreign languages, such as Chinese, Dutch, and French among others. Finally, the authors also suggested directions for future studies extending from the current work and highlighted the need for other versions to be empirically validated.
Emily Ruzich will be presenting this research at the International Society for Autism Research conference in Salt Lake City, Utah, USA.