Rasch analysis of the Center for Epidemiologic Studies Depression scale used for the assessment of community-residing patients with stroke

Purpose. To use Rasch analysis to validate a Center for Epidemiologic Studies Depression scale (CES-D) for measuring the level of depression in community-residing patients with stroke. Method. This study involved 183 community-residing stroke survivors. Depression was measured using the CES-D. The data were analysed with Winsteps version 3.62 using a Rasch model to investigate the unidimentionality of the items' fit, the distribution of items' difficulty and reliability and appropriateness of the rating scale. Result. Three items for CES-D showed a misfit with the Rasch model. Misfitting items included 'I did not feel like eating; my appetite was poor', 'I felt hopeful about the future' and 'My sleep was restless'. The person separation was 2.58, reliability was 0.87 and the rating scale of the CES-D was appropriate for stroke patients. The cut-off point of 16 out of 60 equated to a logit score of 0.85, which when applied to the 17-item scale (with four categories) equated to a cut-off point of 14. Conclusion. At a new cut-off point of 14, the CES-D with 17 items was found to be reliable and valid as a screening tool for assessing depression in community-residing stroke patients. Moreover, it may not be appropriate to use a CES-D as it contains somatic factors for a population with physical dysfunctions.