Community Screening at 12 Months with the First Year Inventory: Stability of Diagnostic Clinical Impressions over Time

Background: Research on emergence of autism (ASD) symptoms and stability of diagnosis in very young children has focused primarily on children at high-genetic-risk for ASD (Ozonoff, et al., 2015). The current study follows up a community sample of children identified as at-risk for ASD on the First Year Inventory (FYI) at age 12 months. Previous research found that approximately 31% of children who screened positive on the FYI at 12 months met diagnostic criteria for ASD at age 3 years (Turner-Brown et al., 2013). However, no research has examined whether a diagnosis would be reliable or stable under 18 months in community screened samples. A subsample of children returned for follow-up diagnostic evaluations between 3 and 5 years.

Objectives:  Our aim was to examine stability of clinical impressions of the at-risk children at 13 months (mean age = 13.2 months, s.d. = 2.5) with diagnosis established at 4 years (mean age = 47.4 months, s.d. = 7.4), and to examine characteristics at 13 months that distinguished children who were diagnosed with ASD from those that were not. We also aimed to examine clinician confidence about clinical impressions at 13 months.

Methods: Participants in this study represent a subsample of 87 children who screened positive on the FYI and participated in a 6-month RCT of Adapted Responsive Teaching intervention; this subsample (n = 32) participated in a follow-up study with independent raters at age 4 years. At 13 months, participants completed comprehensive evaluations with skilled clinical examiners that included assessment of developmental skills (Mullen Scales of Early Learning) and ASD symptoms (Autism Observation Scale for Infants). Assessors rated their clinical impression of the child (e.g., ASD/not ASD) and their confidence in this rating (Not at all confident (1) to Very Confident (5)). Diagnoses were not made at 13 months, though concerns were shared with families. At follow-up, a comprehensive diagnostic evaluation was completed.

Results: At 4-year outcome, 11 children (34%) were diagnosed with ASD, and 21 (66%) did not meet diagnostic criteria for ASD. Sensitivity (.73), specificity (.81), and positive predictive value (.67) of 13 month clinical impressions with diagnostic outcomes are presented in Table 1. Examiners expressed low confidence in their clinical impressions at 13 months (M=1.9, s.d. = .94) regardless of whether their impression was ASD or not. There were no significant differences in confidence for children whose impression was stable over time compared to those that changed (e.g., false positive/negative), p = .5. Furthermore, there were no significant differences between outcome groups in 13 month AOSI scores, p = .7. On the Mullen at 13 months, children with an outcome diagnosis of ASD had lower expressive language than those who did not develop ASD, t(30) = 2.7, p= .01.

Conclusions: Stability of clinical impressions was lower from 13 months to 4 years than prior research with older children (e.g., 18 -35 months). Clinician confidence was low at 13 months regardless of whether the impression was stable over time. Implications for very early ASD screening will be discussed.