Diagnostic Validity of the ADOS and ADI-R in Children Aged <48 Months: A European Perspective

Background:   The diagnostic validity of the ADOS (revised algorithms) and the ADI-R (research algorithms) for ASD in young children with developmental difficulties has been studied in research settings in the USA (e.g. Kim & Lord, 2012). However, less is known about the validity of the gold standard in early childhood populations in clinical settings and in other countries.

Objectives:   To examine the diagnostic validity of the ADOS (revised algorithms) and ADI-R (research algorithms) in a clinical child psychiatric out-patient unit in Stockholm, Sweden.

Methods:   All children with unclear developmental problems (aged <48 months) assessed between 2007 and 2012 at the Neuropsychiatric Unit South East, Stockholm County (www.childdevelopment.se) were included in the study (N>200; ~25%girls; ~60% of Swedish origin, 30% of non-European origin). Aside from the ADOS and ADI-R the assessments included the Vineland-II, Merrill-Palmer-R/Mullen Early Learning Scales and observations in the children’s preschool. Diagnostic validity of the ADOS and ADI-R was investigated against a best estimate clinical consensus diagnosis made by an experienced clinical team (child psychiatrist, psychologist, social worker).

Results: About 70% of the referred children were diagnosed with ASD; the remainder primarily with ADHD and intellectual disability. The ADOS (sensitivity: 96%; specificity: 64%) had a higher diagnostic accuracy than the ADI-R (sensitivity: 58%; specificity: 78%).

Conclusions:   Both ADOS and ADI-R provide clinically valuable information in the assessment of young children with a suspected ASD. Overall, the ADOS showed a higher diagnostic validity than the ADI-R, although the interview yielded a somewhat higher specificity than the ADOS in a European clinical outpatient unit.