Classification accuracy of the Quick Interactive Language Screener for preschool children with and without developmental language disorder

Amy Pace,Maura Curran,Amanda Owen Van Horne,Jill de Villiers,Aquiles Iglesias,Roberta Michnick Golinkoff,Mary S Wilson,Kathy Hirsh-Pasek
DOI: https://doi.org/10.1016/j.jcomdis.2022.106276
Abstract:Introduction: This research examined the classification accuracy of the Quick Interactive Language Screener (QUILS) for identifying preschool-aged children (3;0 to 6;9) with developmental language disorder (DLD). We present data from two independent samples that varied in prevalence and diagnostic reference standard. Methods: Study 1 included a clinical sample of children (54 with DLD; 13 without) who completed the QUILS and a standardized assessment of expressive grammar (Syntax subtest from the Diagnostic Evaluation of Language Variation-Norm Referenced; Structured Photographic Expressive Language Test-Preschool 2nd Edition; or Structured Photographic Expressive Language Test-3 rd Edition). Study 2 included a community sample of children (25 with DLD; 101 without) who completed the QUILS and the Auditory Comprehension subtest of the Preschool Language Scales-5th Edition (PLS-5; Zimmerman et al., 2011). Discriminant analyses were conducted to compare classification accuracy (i.e., sensitivity and specificity) using the normreferenced cut score (< 25th percentile) with empirically derived cut scores. Results: In Study 1, the QUILS led to low fail rates (i.e., high specificity) in children without impairment and statistically significant group differences as a function of children's clinical status; however, only 65% of children with DLD were accurately identified using the norm-referenced cutoff. In Study 2, 76% of children with DLD were accurately identified at the 25th percentile cutoff and accuracy improved to 84% when an empirically derived cutoff (<32nd percentile) was applied. Conclusions: Findings support the clinical application of the QUILS as a component of the screening process for identifying the presence or absence of DLD in community samples of preschool-aged children.
What problem does this paper attempt to address?