Detecting language problems: Accuracy of five language screening instruments in preschool children

To identify a simple and effective screening instrument for language delays in 3-year-old children the reliability, validity, and accuracy of five screening instruments were examined. A postal questionnaire sent to parents of 11423 children included the Dutch version of the General Language Screen (GLS), the Van Wiechen (VW) items, the Language Screening Instrument for 3- to 4-year-olds, consisting of a parent form (LSI-PF) and a child test (LSI-CT), and parents' own judgement of their child's language development on a visual analogue scale (VAS). The response rate was 78% or 8877 children. Reliability (internal consistency) was found to be acceptable (α=0.67-0.72) for all instruments. Significant correlations between the screening instruments (r=0.29-0.55, p<0.01) indicated good concurrent validity. Accuracy was estimated by the sensitivity, specificity, and receiver operating characteristic (ROC) curves against two reference tests based on parent report and specialists' judgement. If the test would classify approximately 5% of the population as screen-positive, the mean sensitivity was 50% assigning between 20% and 30% of the population as screen-positive, the mean sensitivity was 77%. The sensitivity was lowest for the LSI-CT (range 43-62%), whereas short instruments like the LSI-PF, VW, and the one-item VAS exhibited high levels of sensitivity (range 50-86%). The area under the ROC curves, ranged from 0.75 to 0.87. Apparently, short and simple parent report instruments like the LSI-PF and the one-item VAS perform remarkably well in detecting language delays in preschool children.

• View at Publisher
• View at Pubmed
• View at Scopus