Cognitive ASSESSMENT of CHILDREN with ASD

Background: Many children with autism spectrum disorders present with behavioral difficulties that necessitate the use of an abbreviated measure of intelligence.  The authors of the Stanford- Binet Intelligence Scales, Fifth Edition (SB5) (Roid, 2003) report a correlation of .81 between the Full Scale IQ (FSIQ) and Abbreviated IQ (ABIQ) for children ages 2 to 5 and a correlation of .87 for children ages 6 and above. Coolican, et. al, (2008) examined the performance of children with ASD on the SB5 and determined that ABIQ scores accounted for 89.9% of the variance in the FSIQ, and that ABIQ overestimated  FSIQ in 15 out of 17 cases. 

Objectives: To determine the relationship between the SB5 ABIQ and FSIQ in a sample of children with ASD; to determine this relationship when the sample is stratified on level of cognitive functioning; and to determine the rate of false positives/negatives for children with scores above and below 70.

Methods:   This study utilizes the Autism Treatment Network (ATN) Registry data.  519 children (male = 441; ages 2-17 years) who had a SB5 FSIQ (thus, generating an ABIQ) were included in the study. Correlations between the FSIQ and ABIQ and a regression analysis were performed to determine the proportion of variance in the FSIQ accounted for by the ABIQ.  Subjects were also split into High Functioning (IQ >70) vs. Low Functioning groups.  Correlation coefficients were run between ABIQ and FSIQ for these two groups.   FSIQ scores and ABIQ scores were analyzed to examine the rate of false positives (ABIQ <70 and FSIQ > 70) and false negatives (ABIQ > 70 and FSIQ < 70). 

Results: The correlation between the ABIQ and the FSIQ for the entire sample was r=.899 (p<.0001) and a regression using the FSIQ as the dependent variable showed a strong relationship (r=.910, p<.0001) for the entire group.  There was a weaker (but significant) correlation between ABIQ and FSIQ for the 212 subjects with IQ scores < 70 (r=.672) while the relationship between ABIQ and FSIQ for subjects with cognitive levels > 70 was stronger (r=.806, p<.0001).  There was 90% agreement for individuals (n=178) who were low functioning on the FSIQ and the ABIQ (10% rate of false positive) and 85% agreement rate for individuals (n=341) who were high functioning on the FSIQ and the ABIQ (15% false negatives). 

Conclusions: Results show a strong relationship between abbreviated and full cognitive measures of intelligence in a sample of children with ASD. Compared to Roid’s rate of false positives for the SB5 standardization sample (<1%), we see an increased rate of false positives and false negatives.  This suggests that while the ABIQ may both over and underestimate overall cognitive abilities, it is a valid estimate of intelligence.