Early Diagnostic Severity Is Predictive of Not Maintaining ASD Diagnosis

Background: Although early diagnosis of Autism Spectrum Disorders (ASD) is generally stable over time, some toddlers diagnosed with ASD no longer meet criteria when they are older. Studies have sought to identify features at initial diagnostic assessment that distinguish those who maintain a diagnosis from those who do not; however, findings have been varied. Some indicate that early communication abilities tend to predict positive outcomes better than symptom severity.

Objectives: To identify abilities and symptoms at 2-year-old evaluation that predict which children no longer meet ASD criteria by age 4.

Methods: 73 children were evaluated at age 2 (T1, M=26.24 months, SD=4.42) and re-evaluated near age 4 (T2, M=47.79 months, SD=6.22) after screening positive on the Modified Checklist for Autism in Toddlers (M-CHAT; Robins et al., 1999) or being flagged for possible ASD. Evaluations included ADOS, ADI-R, CARS, Vineland Adaptive Behavior Scales-II (VABS), and Mullen Scales of Early Learning (MSEL). Children received a diagnosis of ASD, nonASD developmental delay, or no diagnosis. At T1, 49 received an ASD diagnosis and 24 did not. At T2, 12 (24%) of the children diagnosed with ASD at T1 no longer met criteria for an ASD (ASD-NON), and 37 maintained an ASD diagnosis (ASD-ASD); 20 maintained non-ASD status (NON-NON). Four children who did not receive an ASD diagnosis at T1 were diagnosed at T2; they were excluded here due to small sample size.

Results: A repeated-measures ANOVA revealed that groups differed in ADOS comparison scores, with a significant group x time interaction, F(1,2)=28.29, p<.001. In children who received an ASD diagnosis at T1, logistic regression revealed that those with lower ADOS comparison scores at T1 were more likely to no longer meet criteria at T2, OR=.641, p=.048. ASD-ASD and ASD-NON groups differed at T1 on DSM-IV total symptom count, F(1,47)=6.832, p=.01, social impairment symptoms, F(1,47)=6.62, p=.01, and RRB symptoms, F(1,46)=4.54, p=.04. Lower symptom counts predicted no longer meeting criteria at T2, OR_total=.48, p=.02; OR_social=.37, p=.02; OR_RRB=.36, p=.047. Surprisingly, VABS daily living skills were predictive of not maintaining a diagnosis; as scores increased, odds of no longer meeting criteria decreased, OR=9.31, p=.045. MSEL developmental quotient, VABS adaptive behavior composite, and DSM-IV communication symptom count were not predictive of diagnostic instability (ps>.05).

Conclusions: In contrast to some of the existing literature, metrics of diagnostic severity of ASD symptoms at 2-year-old evaluation, including the ADOS comparison score and DSM-IV symptom count, were predictive of not maintaining an ASD diagnosis. Language and cognitive abilities, however, were comparable between groups who did and did not maintain ASD diagnosis. These findings imply that the lack of diagnostic stability in the ASD-NON group is due to true differences in ASD symptom severity, and is not simply explained by differences in cognitive and language skills at initial evaluation. Future research should examine whether other factors, such as intervention type and intensity, contribute to predictions of diagnostic stability.