Abstracts

Rationale: Given the importance of measuring health-related quality of life (HRQL) in children with epilepsy, having a psychometrically sound measure that is feasible to apply across a variety of clinical and research settings is essential. Item Response Theory (IRT) methods have been increasingly applied in improving existing health outcome measures as it offers the opportunity to examine both scale and item quality. The primary aim of this study was to develop a short-form version of the widely used Quality of Life in Childhood Epilepsy Questionnaire (QOLCE). Methods: Data come from the Health-Related Quality of Life in Children with Epilepsy Study (HERQULES, N=373), a multi-centre prospective cohort study examining determinants and trajectories of HRQL in children ages four to twelve years with epilepsy during the first two years after diagnosis. Multidimensional IRT modeling was used to identify lower quality items and to assess the overall coverage of information provided by each item to produce a short version of the QOLCE with robust measurement properties. As our goal was to maintain the multidimensional construct of health-relayed quality of life, items were compared within each domain in order to determine which items to keep.  Results: A bifactor multidimensional IRT model was obtained, with 16-items representing functioning in four domains of HRQL: Physical, Emotional, Social, and Cognitive. The IRT parameter estimates for the 16-item bifactor model appeared reasonable and had better fit then competing IRT models. Thirty-nine items were removed due to lack of fit, low slope parameter, or insufficient additional coverage provided. The 16 items varied in their discrimination, with mean slope parameter estimates ranging from 1.39 to 4.87. The 16 items reflected a considerable range of health-related quality of life (thresholds: -10.40 to 5.52). Evidence suggests similar coverage of information is obtained with the 16 items compared to the 55-item QOLCE. Preliminary analyses comparing determinants and trajectories of HRQL measured using the QOLCE-55 and QOLCE-16 were similar. Conclusions: Results offer evidence regarding validity of the proposed 16-item short-form QOLCE (QOLCE-16). These items provide robust psychometric information for a wide range of functional levels in each domain. Our results illustrate how IRT methods can be a powerful tool in not only item and scale development and evaluation, but also in scale refinement in order to produce shorter measures to minimize respondent burden. Given the fewer items, QOLCE-16 may be a viable option, when assessing HRQL in children with epilepsy, both for researchers and clinicians.  Funding: Funding for this project was provided in part by a Canadian Institutes of Health Research reseach grant.