Diagnostische Zuverl ä ssigkeit einer Kurzversion des Elternfragebogens ELFRA-2 zur Fr ü herkennung von Sprachentwicklungsverz ö gerungen Validity of a Short Form of the ELFRA-2 (German Version of the CDI-Toddler Form) for Early Identifi cation of Late Talkers

Background: Parents ' questionnaires are assumed to be reliable instruments for early identifi cation of language retardation. The gen- eral use of the ELFRA-2 in practice, however, is limited by its length. The goal of the study was to construct and to evaluate a short version. Subjects and Method: The ELFRA-2 was sent to 1397 parents shortly before their child ' s second birthday. The return rate of the questionnaire was 71 % . Questionnaires of monolingual German speaking children were included in the study (n = 866) and randomly divided into two samples. From the fi rst sample the most informative word categories were extracted by means of regres- sion analysis and were used to compose a short form. With the second independent sample the validity of the short version was determined. Results: By means of stepwise regression anal- ysis within the fi rst sample a short version con- sisting of 42 items was developed. The correlation between scores of the short and the long version was high (r = 0.95) and the classifi cation (no lan- guage retardation vs. late talker) was congruent for most children. The sensitivity and the specif- ity of the short form were 83 % and 97 % , respec- tively. For the prediction of language impairment at the age of three years the short and the long version reached similar detection rates. Conclusions: The results show that the validity of the short and the long form are comparable with respect to detection of late talkers. The time necessary to complete the short version allows its application during standard check-up exami- nations of children at the age of two years, and also to immediately discuss its result with the parents. Before the short form can be applied in pediatric practice, however, a standardization is