The language screening instrument SNEL

Abstract

Objectives: The aim of this study was to design a valid and accurate language-screening instrument (SNEL) (Spreek Normen Eerste Lijn = Speech Norms primary Healthcare), for identifying possible language problems in children from one to six years of age, which is intended to improve the referral of at-risk children for further diagnosis.Methods: A sample of 527 children from one to six years of age were randomly selected from day nurseries, playgroups, and primary schools in order to scale milestones in language development. The questionnaire regarding these milestones was completed by parental report. The milestones were scaled according to the nonparametric Mokken Item Response Model for dichotomous responses, and the assumption of sample independence was evaluated.The external validation was examined by means of the relationship between age and the language-screening instrument SNEL and the relationship between the gold standard for language development and SNEL. For this latter, a weighted sample of 84 children from the large sample of 527 was used together with 14 children with language problems. The accuracy of the instrument in terms of sensitivity and specificity for different percentiles (ROC-curves) was studied in order to identify children with language problems. The accuracy was summarized by the area under the ROC curve (AUROC).Results: A sample independent scale of 14 milestones was found, with a strong scalability (H = 0.95), and a high reliability (Rho = 0.96). The SNEL scores increased with age. The sensitivity and specificity for the tenth percentile were respectively 0.94 and 0.83. The accuracy of SNEL was good (AUROC = 0.94).Conclusions: The unidimensional and sample-independent scale SNEL constitutes a new screening instrument that is short, sensitive, and easy to use in primary health and educational settings.