Screening Children Between 18 and 24 Months Using the Systematic Observation of Red Flags (SORF) for Autism Spectrum Disorders: a Follow-up Study

Background: The American Academy of Pediatrics recommends screening for autism spectrum disorders (ASD) in all children at 18 and 24 months of age. However, there is not yet a well-validated autism-specific screening tool for this age range. Initial research using the Systematic Observation of Red Flags (SORF) for ASD resulted in the identification of 9 red flags that distinguished children with ASD in the second year of life. Further research is needed to confirm and extend these findings in a larger sample of children.

Objectives: The major objectives of this prospective, longitudinal study of the FIRST WORDS® Project were: 1) to determine which red flags differentiate children with ASD from children with developmental delays (DD) and children with typical development (TD) between 18 and 24 months of age, and 2) to determine a total score and a number of red flags cut-off with high sensitivity and specificity for group membership.

Methods: Participants were selected from a general population sample of 6,581 children recruited with a broadband screener for communication delays. Videotapes of the Communication and Symbolic Behavior Scales (CSBS; Wetherby and Prizant, 2002) behavior sample were collected from 150 children 18–24 months of age who were later diagnosed with ASD (n = 60), DD in which ASD was ruled out (n = 30), and TD (n = 60); 54 of these children were included in the initial study. The archived video samples were reanalyzed using the SORF. The 29 behaviors included 0–2 ratings of the absence of typical behaviors and presence of atypical behaviors in the DSM IV domains.

Results: Of the 29 items on the SORF, a one-way ANOVA indicated 20 behaviors that significantly differentiated the ASD group from the DD and TD groups with medium to large effect sizes (Cohen’s d), including the 13 red flags originally identified. The red flags included both a lack of typical and presence of atypical behaviors falling in all three domains of the DSM IV. An analysis of receiving-operator curves (ROC) using the 20 items resulted in a cut-off score predicting group membership with an area under the curve exceeding 94% and a sensitivity of 92% and a specificity of 81%. Based on a count of the number of the 20 red flags present, a cut-off of 8 resulted in a sensitivity of 87% and specificity of 84%. Large significant correlations were observed between the SORF total score for these items and the total score of both the CSBS (r = -.83) and the ADOS (r = .66).

Conclusions: The results indicate that 20 of 29 behaviors on the SORF can meaningfully differentiate children with ASD from children with DD and TD. The findings support previous research showing that early indicators for ASD are present and can be observed in the second year of life. Further validation of the cut-off score is necessary to confirm the utility of the SORF as an interactive screening instrument with at-risk children at 18-24 months following a general population screening for social communication delays.