Objective: To assess the screening performance of a specific language-screening instrument at 18 and 24 months of age and to assess its effect on the early detection and prognosis of language delay.
Design: Child health care physicians were randomised to the intervention group, in which specific language screening was conducted twice (at age 18 months and 24 months), or to the control group (usual care). The specific screening instrument consisted of a uniform set of questions for the parents and test elements for the child, with scaled scores to assess responses.
Setting: Child health care in the Netherlands and referral of screen-positive children.
Subjects: 5734 children in the intervention group and 4621 in the control group.
Main outcome measures: Test characteristics and disorders at 24 months, and confirmed diagnoses of a language disorder before 36 months in both groups. Gold standard based on reports of parents, specialists and expert panel. Prognosis estimated from two diagnostic language development performance scores at 36 months (in questionnaire).
Results: In the intervention group, 3147 of the 5734 children (55%) were screened with the specific screening instrument and 73 of the screened children (2.3%) were screen-positive. Of the screen-positive children, 41 (55%) had confirmed language delay (diagnostic assessment and/or reported treatment). The estimated sensitivity of the test ranged between 24–52% depending on the severity of language disorders. The prevalence of language disorders in three-year olds was estimated to be 2.4–5.3%. In the intervention group, 1.25–2 times more children with language delay had been diagnosed before 36 months. The assessment of language development at 36 months showed no statistically significant differences between the intervention and the control groups.
Conclusions: The inclusion of a specific language-screening instrument in child health centre activities resulted in the earlier detection of children with language delay. Short-term health benefits could not be demonstrated. Large-scale introduction cannot be recommended on the basis of this information alone.
