Agreement Between ADI-R and ADOS in a Twin Sample

Background:

Clinical best practice in the diagnosis of ASD calls for use of standardized measures of both parent report and clinical observation of interaction. The ADOS and ADI-R are currently considered gold-standard measures in this capacity. However, disagreement between these instruments is not uncommon. Understanding the likely sources of disagreement between parental report and clinical observation is important for both clinicians and researchers.

Objectives:

This study explores possible sources of disagreement between parental report (from ADI-R) and clinical observation (ADOS), by contrasting data from MZ and DZ twins.

Methods:

Data is being collected as part of the Social Relationships Study (SRS), a twin study of the full autism spectrum embedded within the Twins Early Development Study (TEDS). The SRS study includes MZ and DZ twin pairs in which one or both twins are suspected of having an ASD (Autism, atypical autism or PDD).

To date, 20 MZ individuals(including 9 completed pairs) and 33 DZ individuals(including 8 completed pairs) aged twelve to fourteen, have been assessed using the ADI-R and ADOS. Different examiners completed the ADOS and ADI-R with each twin in order to minimize rater bias. In order to maintain quality and high levels of inter-rater agreement, examiners maintain reliability and coding disagreements are resolved by consensus.

Data were examined for diagnostic agreement by zygosity using both the Autism cut-off and an ASD threshold on the ADI and ADOS. ADI-R scores are divided into 4 domains. The threshold on this instrument for ASD was determined to be a score above published thresholds on three domains with a score that was +/-2 on the remaining domain.

Results:

Discrepancy between parent report and examiner observation was more marked for MZ than for DZ twins when using the Autism threshold on the ADOS and ADI.

The level of agreement between the ADOS and ADI when using the ASD cut-off for MZ twins was 80% and the rate for DZ twins was 85%. The level of agreement between the ADOS and the ADI when using the Autism cut off for MZ twins was 30% and the rate for DZ twins was 76%.

Conclusions:

This pattern of preliminary data suggests there are contrast effects in parent reporting that are more pronounced for MZ twins than DZ twins when using the Autism cut-off on the ADOS and ADI. It is much less likely that a parent will report concerns about an MZ twin when their co-twin has a diagnosis of an ASD even when autism –specific behaviors are observed by an examiner. If these findings are confirmed, they suggest possible rater bias and contrast effects. Such biases may have implications for twin and family studies of ASD.