20958
Clinical Implications of the ADI As a Measure of Development in Children with ASD

Thursday, May 12, 2016: 5:30 PM-7:00 PM
Hall A (Baltimore Convention Center)
N. Navot, S. J. J. Webb and R. Bernier, University of Washington, Seattle, WA
Background:  

Predicting future Autism Spectrum Disorder symptom level and developmental trajectory, is a primary goal for facilitating service development and supporting families. However, the heterogeneity within ASD and resulting prognostic opacity creates difficulties for clinicians seeking to provide quality information. Published longitudinal studies suggest several developmental trajectories of improving/worsening in symptom level. The Autism Diagnostic Interview was designed to provide a framework for collecting both the developmental history and current functioning within a diagnostic framework. Although not designed to examine symptom change over time, analyses with the ADI collected at one time point in large samples may illuminate changes in symptom presentation.

Objectives:  

 We describe autism symptom trajectories derived from individual item ADI-R data among 1629 verbal Simons Simplex Collection (SSC) children from age 6 to 16 years. Focusing on differences between current and 4-5 year functioning, we used cluster analytic approaches to grouping symptoms according to their change over time and compared the different clusters among participants with different levels of language. 

Methods:

Data were obtained from the SSC. Only verbal participants that received ADOS Module 3 and Module 2 were included. Data were first analyzed for baseline age differences. For each item, current value as function of age at time of interview was plotted.  Second, we created a difference score (DS) at the item level. For each item, the past score (PS) was subtracted from the current score resulting in a difference score(e.g., PS = 3; current = 1; DS = 2). To identify clusters of items that show similar patterns in DS or PS, we ran a double-way clustering on the DS and PS matrix: this allows clustering of the probands based on the ADI item values, as well as clustering of items based on proband values. We use K-means with K=2 for clustering. The different clusters of symptom level were compared between module 2 and Module 3 participants. 

Results:  

Our analyses show that several patterns of change over time can be detected from a single time point administration of the ADI-R. Most prominent, there is significant improvement in symptoms from 4-5 to current functioning. However, despite individual change, there is little difference in the degree of improvement by current age (72-192 months); older children did not demonstrate “more change” or were “more improved” than younger children in this age range. Clustering analysis reveals subgroups among ADI items, characterized by a different level of severity of symptoms and level of change over time. Those subgroups were different among Module 2 and Module 3 participants. Thus, pattern of severity of symptoms can provide important information about eventual language outcomes (i.e. whether the child will be appropriate for a Module 2 or 3 at age 6 to 16 years).

Conclusions:  

Our analyses suggest that the ADI, when used with verbal children after the age of 6 years, may also provide the parent with prognostic information about the pathways that could be expected for particular symptoms and the probability of improvement of certain symptoms at various stages of development.