Autism Diagnostic Interview-Revised and the Childhood Autism Rating Scale: convergence and discrepancy in diagnosing autism.
ADI-R and CARS agree only two-thirds of the time, so use both and expect CARS to flag more kids.
01Research in Context
What this study did
The team compared two gold-standard autism tests. They gave the same kids both the ADI-R parent interview and the CARS clinician scale. Then they checked how often the two tools agreed on an autism diagnosis.
They wanted to know if you could swap one test for the other or if you need both.
What they found
The tools matched only two-thirds of the time. CARS called more kids autistic than ADI-R did. When the team loosened the ADI-R rules, agreement got a little better.
How this fits with other research
Kim et al. (2012) fixed the gap. They built new ADI-R rules for toddlers that catch more cases. Their update supersedes the 2003 worry that ADI-R misses kids.
Lancioni et al. (2006) explains why ADI-R under-counts toddlers. Little kids have not yet shown enough repetitive behaviors for the old ADI-R cutoff. Using CARS plus ADOS-G in that age group catches them.
Hirota et al. (2018) systematic review backs the two-tool habit. They found that no single screen works well across ages, so stacking tools is still best practice.
Why it matters
Do not rely on just ADI-R or just CARS. Use both and expect CARS to flag more kids. If you test toddlers, ask your team for the 2012 toddler ADI-R rules or add ADOS-G. A second tool keeps you from missing—or over-labeling—kids on the spectrum.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Add the free CARS form to every ADI-R intake and compare scores before you decide.
02At a glance
03Original abstract
The agreement between the Autism Diagnostic Interview-Revised (ADI-R) and the Childhood Autism Rating Scale (CARS) was investigated in the diagnostic assessment of 54 children aged 22-114 months referred for possible autism. The observed agreement between the two systems was 66.7% (Cohen's kappa = .40) when the ADI-R definition for autism was applied (i.e., scores reaching cutoff in three domains on the ADI-R), but increased considerably with less stringent criteria; that is, scores reaching cutoffs in two domains and in one domain on the ADI-R. As predicted, the CARS identified more cases of autism than the ADI-R. Children classified as autistic according to both instruments had significantly lower IQ/DQ and more severe autistic symptomatology than those classified with the CARS only.
Journal of autism and developmental disorders, 2003 · doi:10.1023/a:1024410702242