Autism Spectrum Disorder in the DSM-5: Diagnostic Sensitivity and Specificity in Early Childhood.
DSM-5 autism rules catch most toddlers but also sweep in too many; tweak cut-offs and use ADOS-2 delay rules to fix it.
01Research in Context
What this study did
The team asked: do the new DSM-5 rules for autism catch the same toddlers who got the older PDD label?
They looked at kids already diagnosed with PDD. Then they checked who still met DSM-5 ASD rules.
They counted how many were caught (sensitivity) and how many were wrongly added (specificity).
What they found
DSM-5 caught most kids (sensitivity .84).
But it also over-tagged many kids (specificity only .54).
Children with fewer extra problems were the ones most often missed.
How this fits with other research
Matson et al. (2013) found higher specificity (.74) in toddlers. Their stricter cut-off (2 of 3 social items and 1 of 4 repetitive items) shows you can tighten the rules and still keep high sensitivity.
Wong et al. (2016) saw 91 % of preschoolers keep the label. The small drop matches the low specificity here.
Liu et al. (2026) gives a fix: raise the ADOS-2 CSS cut-off to 6 when global delay is present. This lowers false positives without losing real cases.
Why it matters
You now know DSM-5 can both miss and over-call ASD in very young kids. Use Matson et al. (2013) cut-offs when you score. Add Liu et al. (2026) ADOS-2 rules for kids with delay. These two moves give you cleaner decisions and fewer wasted hours on wrong cases.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Score your next toddler evaluation with the 2-of-3 social and 1-of-4 repetitive cut-offs from L et al. (2013).
02At a glance
03Original abstract
Changes to the DSM-5 Autism Spectrum Disorder (ASD) criteria raised concerns among parents and practitioners that the criteria may exclude some children with Pervasive Developmental Disorder (PDD). Few studies have examined DSM-5 sensitivity and specificity in children less than 5 years of age. This study evaluated 185 children aged 20-55 months with DSM-IV PDD or developmental delay. Autism Diagnostic Interview-Revised (ADI-R) and Autism Diagnostic Observation Schedule (ADOS) data was assigned to DSM-5 subdomains. Children displaying the required symptomatology were classified with DSM-5 ASD. DSM-IV clinical diagnoses were compared to DSM-5 classifications. Using combined ADI-R/ADOS information, sensitivity was .84 and specificity was .54. Comorbid behaviour and emotional problems were significantly lower in children with PDD that did not meet DSM-5 criteria.
Journal of autism and developmental disorders, 2016 · doi:10.1007/s10803-016-2734-4