[A method for phenotyping specific language impairment (SLI) in children aged 6-8]

In recent years possibilities of genotyping have increased rapidly. The more accurate and valid the methods of phenotyping are, the greater is the chance to find genotype-phenotype correlations. For a complex disorder like specific language impairment a dichotomous decision (language impaired - normal developing) is necessary in addition to characterisation of endophenotypes concerning research and clinical practice.

A total of 61 (specific language impaired and normal developing) children aged 6-8 years were examined in a diagnostic study phase I/II. The diagnostic value of a language test battery including 8 subtests derived from German norm-referenced, standardized language tests was evaluated on the basis of different classification analyses.

Comparing the means, the subtests of the test battery showed significant differences between the 2 groups. Several analyses of classification validity resulted in acceptable to good (≥ 80%) diagnostic accuracy. Best classification rates for sensitivity (92.3%) and specificity (96.0%) were obtained by using logistic regression (method: forward conditional).

The results empirically support the diagnostic value of the test battery regarding differentiation between children with normal language and those with specific language impairment. However these findings cannot result in recommendations in routine clinical and scientific practice as the study was designed as a Phase I/II diagnostic study. For dichotomous diagnostic decisions a norm-referenced, standardized test should only be used, if there is empirically derived information about its classification validity.