Measuring the Broad Autism Phenotype in Parents of Children with Autism

Background: The broad autism phenotype (BAP) refers to a constellation of subclinical personality and language features in unaffected relatives of individuals with autism spectrum disorder (ASD), thought to index genetic liability to ASD, and therefore informative for understanding ASD etiology. Whereas gold standard diagnostic tools exist for evaluating ASD, methods for characterizing the BAP are less consistent, including direct assessment measures that have the benefit of objective independent ratings based on concrete behavioral examples, as well as questionnaires that rely on self- or informant-reports and are therefore more efficient but susceptible to rater bias. Indeed discrepancies between self- and informant-reports have been observed, pointing to a strong need to evaluate different measures of the BAP to determine the most accurate methods for characterizing this important construct.

Objectives: This study compared rates of the BAP using gold standard direct assessment measures, and questionnaire data in mothers and fathers of individuals with ASD. We examined interrelationships among measures, and within-family correlations of clinical-behavioral features across measures that could provide insights into those measures most sensitive to biologically meaningful clinical features.

Methods: One hundred eighty four parents of individuals with ASD were assessed for personality and language features of the BAP using three different methods: 1) the Modified Personality Assessment Scale Revised (MPAS-R; Tyrer, 1988), a direct assessment interview, 2) the Broad Autism Phenotype Questionnaire (BAPQ; Hurley et al., 2007), and 3) the Pragmatic Rating Scale (PRS; Landa, 1992), which was used to assess pragmatic language during semi-naturalistic conversations. BAP status was established based on established cut-offs (Hurley et al., 2007; Losh et al., 2008). Within family associations in clinical-behavioral symptoms were examined with child variables from the ADOS, ADI-R, and the Pragmatic Rating Scale-School Age (PRS-SA).

Results: The self and informant ratings of the direct assessment MPAS-R classified 48% and 49% of the sample as BAP+, respectively. The self and informant ratings of the BAPQ classified 15% and 19% of the sample as BAP+, respectively. The BAPQ captured only 25% (self) and 20% (informant) of those classified by the MPAS as BAP+. The PRS classified 22% of the sample as BAP+ on the pragmatic language trait only. The BAPQ classified only 23% (self) and 20% (informant) of these individuals as “positive” for the pragmatic language feature of the BAP. Parent-child correlations revealed negative associations between the BAP (as measured by direct assessment on the MPAS-R) and child ADI-R social and verbal communication scores (ps<.05). Similarly, the pragmatic language domain on the BAPQ was negatively associated with child’s overall ASD symptom severity (p<.03).

Conclusions: Findings suggest that questionnaires may markedly under-classify the BAP, relative to objective, direct assessment measures of personality and pragmatic language. Parent-child correlations were also most robustly observed with direct assessment measurement of the BAP. Theoretical and methodological considerations for future studies of quantitative traits in ASD and in family members will be discussed.