Screening and Prediction of Autism in Toddlers Using SORF in Videos of Brief Family Interactions.

Researchers tested a new screening tool called SORF that watches how toddlers (15-24 months old) interact with their parents during 10-minute videos. They studied 54 children, including those with autism, developmental delays, and typical development. The tool was quite good at identifying which children had autism, correctly identifying about 79% of children with autism and 80% of children without autism using a simple scoring system.

This validation study examined the Systematic Observation of Red Flags (SORF) as an autism screening tool using 10-minute parent-child interaction videos in toddlers aged 15-24 months. The study included 54 children: 19 with autism spectrum disorder, 23 with developmental delays, and 12 typically developing children. Trained coders evaluated various SORF measures including total scores, social communication scores, restricted repetitive behavior scores, and composite scores. The composite score demonstrated the strongest discriminative ability (AUC = 0.884) for distinguishing between children with and without autism.

Using a cutoff score of 7, the composite measure achieved 78.9% sensitivity and 80.0% specificity for predicting autism diagnosis.

SORF shows promise as an observational screening tool for autism in toddlers during brief parent-child interactions. The tool's moderate accuracy suggests potential utility as part of comprehensive screening protocols, though further validation with larger samples and comparison to existing measures is needed before clinical implementation.

Small sample size (54 children total) limits generalizability. Study lacks comparison to established screening tools. Coder reliability and training requirements not specified. Long-term predictive validity unknown. Population characteristics and recruitment methods not detailed.

The purpose of this study was to validate the utility of the Systematic Observation of Red Flags (SORF) for autism screening during 10-minute parent-child interactions at ages 15-24 months. A total of 54 children participated in this study, including 19 with autism spectrum disorder (ASD), 23 with developmental delay, and 12 typically developing children. Coders coded 10-minute videos of parent-child interactions based on the defined scoring criteria. The discriminative ability for outcome diagnosis was evaluated for total score, social communication score, restricted repetitive behavior score, number of red flags, and composite score.

SORF scores demonstrated good discriminative ability between ASD and non-ASD children, with the composite score (AUC = 0.884) showing the best discriminative ability for outcome diagnosis and predicting likelihood of ASD in young children. The composite score represented a simplified measurement, with the cutoff score of 7 and sensitivity and specificity of 0.789 and 0.800, respectively.