Agreement Between Early Intervention Staff Impressions and Screening Tools for Autism Spectrum Disorder
Early diagnosis of autism spectrum disorder (ASD) is important so that early intervention (EI) can begin as soon as possible. Early identification of ASD allows for earlier access to EI, during a time when neural plasticity is at its peak. Further research is needed to optimize identification of children with possible ASD through EI systems, including evaluating the role of standardized screening tools.
The objective of this study was to determine the agreement between ASD-targeted screening tools and the clinical impressions of EI clinicians.
This study was carried out through York Region Early Intervention Services (YREIS), in Ontario, Canada. The screening tools used were the Modified Checklist for Autism in Toddlers – Revised with Follow-Up Interview (M-CHAT-R/F) and the CSBS DP Infant-Toddler Checklist (ITC). 235 packages containing the ASD-specific screening tools were distributed to families through YREIS’ intake staff according to three age groupings: children 9-15 months received the ITC, children 16-24 months received the ITC and MCHAT-R/F, and children 25-30 months received the MCHAT-R/F. The MCHAT-R/F questionnaires were scored and those over the cut-off (score of 3 or higher) underwent the follow-up interview. During the intake process, the EI clinician (blinded to the results of the questionnaire) also filled a clinical opinion sheet indicating whether they suspected the child had ASD. The consistency between the EIS clinical opinions and the screening tools were measured using a kappa statistic.
Of the 235 distributed packages, 57 were returned back to YREIS (24% response rate). Of the 57 packages, 51 were completed. YREIS clinicians identified concerns about ASD in 25 of the 51 participants (49%). Of the completed MCHAT-R/F’s, 11 of 39 (28%) were positive (indicating concern), compared with 14 of 26 ITC questionnaires (54%). In the comparison between clinical opinion and the ITC in 26 participants, kappa was calculated to be 0.429, demonstrating fair agreement. Agreement was calculated for the MCHAT-R only and clinical opinion in 46 participants, which showed a kappa of 0.136 (slight agreement). The kappa for the full M-CHAT-R/F (39 participants) and clinical opinion was 0.238 (fair agreement). The agreement between one positive screening tool and clinical opinion (51 participants) was 0.280, demonstrating fair agreement.
The results of this study show a disappointing level of agreement between EI staff and ASD-targeted questionnaires. There was a higher rate of concern in the EI staff when compared with the M-CHAT-R/F. The ITC has lower specificity for ASD, which may explain the higher proportion of identified concerning cases when compared with the M-CHAT-R/F. These results suggest that combined use of clinical opinion and screening tools is likely to maximize the proportion of cases of suspected ASD identified through EI intake processes, with a trade-off of lower specificity.