Pilot Analyses of the Feasibility and Validity of Autism Spectrum Disorder Screening Instruments: Social Communication and Interaction Screening in Toddlers (SISO)

Background: Early diagnosis is the critical first step for early intervention for autism spectrum disorders (ASD), which is crucial for prognosis. However, there have been limited early ASD screening instruments available in Korea.

Objectives: The objectives of this study are 1) to develop an early screening instrument of ASD for infants/toddlers in Korea, named “Social Communication and Interaction Screening in tOddlers (SISO),” and 2) to verify feasibility and validity of the instrument by comparing with clinical best estimate diagnosis.

Methods: 1) We primarily identified the lists of social and communication behaviors that might be manifested in infants/toddlers with ASD by age range, each of which is relevant to DSM-5^TM criteria. Behaviors in the lists were paralleled by those observed in typically developing infants/toddlers in each age range. We developed preliminary interview questions and play-based activities to detect such behaviors clearly. 2) Focus group interviews were conducted on 30 daycare center/kindergarten teachers and 11 professionals, in order to select the most relevant items, draw appropriate examples regarding children’s play in everyday life, and refine item wordings and expressions for better understanding. 3) Finally, we developed SISO-P (Play-based observation of toddlers, 3~14 items) and SISO-I (Interview questions for caregivers/teachers, 11~19 items). Both instruments covers five age ranges: 9~11, 12~17, 18~23, 24~35, and 36~42 months. 4) Subjects included for validity analyses were toddlers at age 12 to 42 months with no significant physical/neurological/sensory problems interfering the diagnostic procedures. For all participants, SISO-P and SISO-I were conducted along with the diagnostic procedures using CARS, ADOS, and ADI-R. We analyzed sensitivity, specificity, positive predictive value, and negative predictive value of the SISO Classification in relation to the clinical best estimate diagnosis that provides ASD/non-ASD classification. χ²-test and t-test were used to analyze validity of individual items.

Results: Total 251 infants/toddlers participated (9~11, n=3; 12~17, n=20; 18~23, n=30; 24~35, n=110; 36~42, n=88). 1) For overall subjects, sensitivity was 97.84% in SISO-P, and 95.00% in SISO-I. Specificity was 34.55% in SISO-P, and 56.25% in SISO-I. For SISO-P, PPV was 65.38 and NPV was 92.68. For SISO-I, PPV was 73.08 and NPV was 90.00. 2) For SISO-P, individual item analyses on 24~35 month-olds (ASD, n=61; non-ASD, n=48) and 36~42 month-olds (ASD, n=71; non-ASD, n=17) showed significant differences in scores for all items (p<.05) and significant differences in frequencies of screen-positive and -negative scores between ASD and non-ASD groups for all items (p<.05). The results showed that, for SISO-I, differences in scores for ‘Social Interest to others’ item in 24 months and 'Eye Contact', ‘Social Smile', and 'Social interest to others' items in 36 months are not statistically significant across two groups, suggesting that the interview questions might need to be revised in a way that better differentiates toddlers’ behaviors towards others and caregivers.

Conclusions:

The present study suggests that SISO might be feasible and highly sensitive but is a less specific screening instrument for ASD in toddlers. Further study is needed on a larger number of subjects to find out which items are adequately specific in screening ASD.