Using Narrow and Global Measures to Identify Children with LI in a Population Sample

Appropriate identification of children with language impairment (LI) is a persistent clinical issue in the field of communication disorders. Supplemental measures derived from the clinical markers of LI have been evaluated for diagnostic accuracy and clinical utility to improve diagnostic practices. Three of such measures are the percent of grammatical communication units (PGCU), the finite verb morphology composite (FVMC) and errors per communication unit (ECU). Although there is promising initial data to support the use of these supplemental measures, they have only been evaluated by two-gate studies to date, which creates spectrum bias. Spectrum bias is the methodological flaw with the greatest potential to overestimate the diagnostic accuracy of a measure (Dollaghan & Horner, 2011; Lijmer et al., 1999; Rutjes, et al., 2005). Therefore, this one-gate study aims to evaluate the diagnostic accuracy and clinical utility of the supplemental measures in a participant pool that reflects a clinical population. Participants included 141 monolingual children ages four through six recruited from a public school in upstate New York. Findings indicate that PGCU, FVMC, and ECU do not have acceptable diagnostic accuracy to identify LI in children ages four through six. Findings from this study may contribute to recommendations for identification of LI.