An investigation into aspects of the Mayo early language screening test

The concurrent validity and inter-tester agreement on the Mayo early language screening test (MELST), and the concurrent validity of parental anxiety about speech and language development as a screening measure were investigated. A sample (aged 18 months to 5 years) comprising 120 children referred to the Mayo speech and language therapy department during 1995 and 80 controls, was assessed on the MELST by their local public health nurse. Within a calendar month each child was reassessed on the MELST by the therapist, who also administered the Reynell developmental language scales (RDLS) and the Edinburgh articulation test (EAT) as criterion measures. Overall, the MELST achieved moderately high sensitivity and specificity measures. Phonological problems were identified with more accuracy than language problems and the test was equally sensitive to impairments in comprehension and expression. The test tended to over-refer and was least accurate in screening the language skills of children at the highest age level (4-5 years). A significant association was found between parental concern and pass/failure on each of the MELST, RDLS and EAT. As a screening measure, parental anxiety achieved high sensitivity, moderate specificity and a moderate positive predictive value. However, it also tended to over-refer, and was also least accurate in identifying speech and/or language disorders at the highest age level. Moderately high inter-tester agreement measures were found, with greater agreement between scoring for controls than referrals. The phonology sections were scored significantly higher by the public health nurses than by the therapist. The results of this study imply that the MELST is sufficiently accurate and reliable to recommend its continued use in child surveillance programmes.