How Certain Are Clinicians in Determining Outcome Diagnoses in 2 Year-Olds?

Background:

Clinicians have become increasingly more reliable with identifying autism spectrum disorder (ASD) as young as age two (Mandell, et al., 2005). Community providers are recognizing symptoms earlier, resulting in earlier referrals to specialists (Charman et al., 2002), and use of the Autism Diagnostic Observation Schedule (ADOS) enhances diagnostic validity (Gotham, et al., 2009). The impact of clinician’s certainty in determining ASD diagnoses in toddlers, however, is less clear.

Objectives:

This study examines certainty ratings made by experienced, licensed and research-reliable psychologists when determining diagnoses of 24-month-old toddlers participating in autism research studies that involved diagnostic outcome evaluations.

Methods:

Toddlers participated in studies at the Marcus Autism Center. The clinically-referred sample was collected through a study on early detection consisting of 181 toddlers: Mean age=24.1m; 75% male; 134 with ASD; 47 with non-ASD developmental delays (DD). The infant sample was collected through a longitudinal prospective study on infants siblings of children at High Risk (HR) or Low Risk (LR) for ASD. This sample consisted of 107 toddlers at their 24-month diagnostic evaluation: Mean age=24.5m; 64% male; 11 with ASD; 13 with broader phenotype (BAP); 7 with DD; and 75 with typical outcome (TD). Measures included Mullen Scales of Early Learning (Mullen); ADOS-2; Communication and Symbolic Language Scales (CSBS), and Vineland Adaptive Behavior Scales, 2^nd Edition (Vineland-II).

Results:

Clinically-Referred Sample: Independent T-Tests revealed a significant difference in clinician certainty ratings (CR) between toddlers diagnosed with ASD vs. DD (Mean CR = 86.8% vs. 77.2%, respectively; t(1,179)=-3.13, p<.01). A trend for negative correlations were found between CR and ADOS-2 social affect (-2.7; p=.07) and RRBs (-.26; p=.09) scores for the DD toddlers. However, strong positive correlations were found between CR and ADOS-2 SA and RRB scores (.49 and .33, respectively; p<.001).

Infant Longitudinal Sample: ANOVA results revealed significant differences in CR percentages for diagnoses in the infant sample (F(3,105)=10.1; p<.001). Clinicians were most confident in determining TD outcome (CR mean=88.13%) compared to ASD (79.1%), DD (77.1%), and BAP (71.5%). Clinicians were also more confident in diagnoses merely based on risk status (LR mean=86.4%; HR mean=80.8%; t(1,106)=2.22; p<.05). In contrast to the clinically-referred toddlers, no significant correlations were found between CR and ADOS-2 scores for the DD-outcome infants; however, strong negative correlations were found in the TD infants (r=-.46 for ADOS-2 SA and -.53 for RRB, p<.001). In the ASD-outcome infants, a positive correlation was found between ADOS-2 SA and CR (r=.76; p<.01).

Gender, race, and maternal education levels had no impact on CR for either sample.

Conclusions:

Findings highlight that, for both samples, clinicians are more confident in diagnosing ASD versus non-ASD developmental delays. For HR toddlers, clinicians are less confident in diagnosing BAP. When blind to risk status, clinicians are less confident when determining diagnosis in HR toddlers. Increased symptomatology enhances certainty for ASD toddlers, whereas the opposite is true for TD toddlers. These results underscore the importance of clinician training in accurate and reliable assessment of ASD, particularly for young children and those who are at high-risk for ASD.