Unidimensionality and cross-age validity of the italian EQ-5D-Y tested by a Rasch analysis

Objectives: Deepening the ongoing validation of the EQ-5D-Y questionnaire in Italian language, the main aim was to test the ability of the instrument to focus on one attribute or dimension at a time (unidimensionality), and secondly to test cross-age differences. Method: Rasch measurement was used to check measurement proprieties of Italian EQ-5D-Y questionnaire. Three different Italian samples from a general population were used: two samples of 415 and 774 children respectively collected in two different surveys and a sample of 1486 adults to compare the results. A cluster analysis was subsequently performed in order to investigate different response patterns. Main results: Preliminaries results of two Rasch analyses with children datasets pointed out high reliability of “dimensions” parameters, whereas “children” estimates had unsatisfactory reliability. It means that “dimensions” position can be generalized across different samples, but person position cannot be generalized. The same results were also obtained on the sample of adults. In-depth investigation of response patterns emerging from cluster analysis, as well as cross-age comparisons, are currently being performed. Discussion: The small number of children reporting problems, and the consequent low sample variance is surely a cause of the low person reliability. Another reason might be the presence of more than one dominant pattern of response, as emerging from cluster analysis. More detailed analyses are being run in order to investigate whether the presence of possible cross-ages differences are due to different health status or to Differential Item Functioning.