Expression of the Broad Autism Phenotype In Simplex Autism Families From the Simons Simplex Collection

Background: To investigate the biological pathways that contribute to autism, non-autistic relatives who manifest qualitatively similar traits of the disorder have been utilized in research. This broad autism phenotype (BAP) is principally measured in research through the use of a rating scale. Our group examined some of these measures previously on a smaller dataset. We concluded that classification differed depending on the measure used to assess the BAP; that the SRS:ARV and BAPQ may be assessing different constructs; and that some parents from simplex families demonstrate BAP traits. We extend our previous work by examining a larger dataset to determine if our previous findings hold.

Objectives: To examine the relationship among three of the most commonly used measures that assess the broader autism phenotype in parents of children who have an autism spectrum disorder diagnosis and explore how gender relates to scores on these measures.  

Methods: Data were collected via the Simons Simplex Collection (SSC) which is operated by the Simons Foundation Autism Research Initiative. The SSC uses rigorous and uniformly applied phenotyping procedures to study families in which one child (ages 4 to 17) has been diagnosed with an autism spectrum disorder but neither parent nor any sibling has been determined to be on the autism spectrum. Parents in the study completed the Social Responsiveness Scale: Adult Research Version (SRS:ARV) on their spouses, the Broad Autism Phenotype Questionnaire (BAPQ) self-report; and some received ratings from clinicians on the Family History Interview (FHI). These three measures provide quantification of traits related to the BAP.

Results: Our sample (N=3311) was comprised of 1652 mothers and 1647 fathers. Correlations among measures used to assess the BAP were significant (p<.05) but weak (r²<.20). Contingency analyses for BAPQ by SRS:ARV total scores were significant; but classification appears to overlap primarily at the extremes; that is, for participants who exceeded cutoff on the SRS:ARV, only 5% also exceeded cutoffs on the BAPQ; in contrast, for participants who exceeded cutoffs on the BAPQ, only 6% also exceeded cutoffs on the SRS:ARV. Differences by gender will also be discussed.

Conclusions: Differences suggest that the BAPQ, SRS:ARV, and FHI are not highly correlated even within this larger dataset and may be measuring different constructs. The SRS and BAPQ may concurrently predict presence or absence of the BAP, but only for individuals who are at the extremes, with a high level of variability at the moderate level of BAP expression. The results also suggest weak correlations between self, informant, and clinician ratings across measures. Findings suggest that the assessment of the BAP varies considerably depending upon the measure being used. Implications, including the need to combine multiple methods of measurement from multiple informants and to clarify underlying constructs being assessed will be discussed.