Examining the predictive accuracy and stability of the Shaywitz DyslexiaScreen.

Abstract

Accurate processes to identify individual risk for dyslexia are needed to identify students who require intensive interventions and prevent text reading difficulties. The Shaywitz DyslexiaScreen (SDS) is a teacher rating tool available for use in schools and clinics to screen for dyslexia risk. In this pre-registered study, we evaluated the stability and accuracy of a prototype of the SDS administered in fall of Kindergarten, 1st, and 2nd Grades for predicting reading difficulties at the end of 1st and 2nd Grades with a sample of 683 students tracked across these grades. Paired comparisons of teacher ratings on the SDS K-2nd Grades demonstrated moderate overall stability, but stability for at-risk students was much lower and approximately two in three students changed risk category once across the 3 years. The SDS identified a high percentage of truly at-risk students (sensitivity range .63-1.0). However, the SDS yielded a high number of false positives (PPV range .17-.47). When thresholds for risk were optimized for this sample, the SDS performed comparably with other candidate screening measures. We conclude that risk screening based on current SDS norms would result in large numbers of students receiving unnecessary interventions. Additional research on the technical adequacy of the SDS, including the potential for bias, is necessary before it can be considered an evidence-based tool.