Stability of the ADI-R
ADI-R algorithm status at least twice during this period; another 8 changed algorithm status once,
indicating that ADI-R classifications of autism may not remain stable over time.
Understanding the developmental trajectory of autism spectrum symptomatology over time also
provides further insight into the stability of ASD diagnoses. Comparisons of ADI-R current scores with
retrospective ratings for scores at 4-5 years (the age used for rating diagnostic algorithm items)
generally show steady improvements in symptomatology with age (Boelte & Poustka, 2000; Fecteau,
Mottron, Berthiaume & Burack, 2003; Piven, Harper, Palmer & Arndt, 1996; Seltzer et al., 2003;
Shattuck et al., 2007). There are clearly limitations to retrospective studies of this kind, but the
prospective study by Charman et al. (2005) also showed significant improvements in domain scores
from 2 to 7 years, although patterns of change differed across domains and age groups (see also Cox et
al., 1999). However, the variance in ADI-R scores also increased with age and although many children
did improve, others showed an increase in symptom severity. Starr, Szatmari, Bryson and
Zwaigenbaum (2003) analysed change over 2 years in 68 high functioning children originally assessed
on the ADI-R at a mean age of 5.8 years. They found significant improvements in each domain, with
only a few items indicating a worsening of behaviour.
The current prospective study employed a sample of 35 children with an independent diagnosis of
autism/ASD prior to participation in the study. Participants were originally assessed on the ADI-R at a
mean age of 3.5 years and then again at a mean age of 10.5 years. Standardised assessments of
cognitive and language level were available at both time points and participants had a broad range of
intellectual ability (T1 IQ <20 to >130). The study focused on the ADI-R with respect to the following
issues: a. stability of the ADI-R at three levels: meeting algorithm criteria, domain and item level