Premium
Beyond pass‐fail: Examining the potential utility of two thresholds in the autism screening process
Author(s) -
Roberts Megan Y.,
Stern Yael,
Hampton Lauren H.,
Grauzer Jeffrey M.,
Miller Amanda,
Levin Amy,
Kornfeld Benjamin,
Davis Matthew M.,
Kaat Aaron,
Estabrook Ryne
Publication year - 2019
Publication title -
autism research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.656
H-Index - 66
eISSN - 1939-3806
pISSN - 1939-3792
DOI - 10.1002/aur.2045
Subject(s) - autism , autism spectrum disorder , intervention (counseling) , psychology , false positive paradox , categorical variable , logistic regression , clinical psychology , medicine , developmental psychology , psychiatry , machine learning , computer science
Access to early intervention as early in development as possible is critical to maximizing long‐term outcomes for children with autism spectrum disorders (ASD). However, despite the fact that ASD can be reliably diagnosed by 24 months, the average age of diagnosis is 2 years later. Waitlists for specialized developmental evaluations are one barrier to early diagnosis. The purpose of this study was to examine one potential approach to reducing wait time for an ASD diagnostic evaluation by examining the utility of using more than one threshold for an autism screening tool, the Screening Tool for Autism in Toddlers and Young Children (STAT). Participants included 171 children between 24 and 36 months of age who received a medical diagnostic evaluation through Illinois' Early Intervention Program. This study directly compared the performance of the STAT when scored: (a) using the original single threshold, (b) using seven equally weighted items using a single threshold, and (c) using all items differentially weighted based on how strongly that item predicts a later ASD diagnosis. In addition, this study explored the potential utility of using two thresholds rather than a single threshold for each scoring method. Results of this study suggest that using a two‐threshold logistic regression method has potential psychometric advantages over a single threshold and categorical scoring. Using this approach may reduce the wait time for specialty ASD diagnostic evaluations by maximizing true negatives and true positives, such that specialty evaluations may be reserved for those cases that are more ambiguous or more complex. Autism Research 2019, 12: 112–122 . © 2018 International Society for Autism Research, Wiley Periodicals, Inc. Lay Summary This study examined the benefits of using two versus one cutoff score when screening for autism. Results indicate that having two scores and weighting test items based on predictive association with an autism diagnosis is better than using a single score and weighting each item equally. Using such an approach may reduce the wait time for specialty autism diagnostic evaluations, such that specialty evaluations may be reserved for those cases that are more ambiguous or more complex.