Psychometric properties of the Center for Epidemiologic Studies Depression (CES-D) scale in stroke survivors

Purpose: This study aimed to evaluate the psychometric properties of the Center for Epidemiologic Studies Depression (CES-D) scale in adults with stroke. Methods: A secondary analysis of the Stroke Recovery in Underserved Populations Cohort Study. The CES-D was administrated to 828 stroke patients at discharge from inpatient rehabilitation facilities and at 3- and 12-month follow-ups. Data were analyzed using classical test theory (CTT) and Rasch measurement model. Results: Confirmatory factor analyses of the CES-D items showed excellent fit of a four-factor model (CFI = 0.98; TLI = 0.98; RMSEA = 0.05). CTT analyses revealed satisfactory reliability and validity. Rasch analyses also supported the unidimensionality of each factor (subscale). Wright maps indicated a floor effect and item gaps. A few items displayed differential item functioning: 3 items (1 depressed affect and 2 somatic symptoms) across gender, 1 item (depressed affect) across time of assessment and all # somatic symptom items across time of assessment. Conclusion: The four-factor structure of the CES-D was confirmed and its psychometric properties were validated, supporting the use of four subscales to characterize depressive symptomatology in adults with stroke. Supplementary assessments are needed for evaluating and comparing somatic symptoms over time. A refinement of the CES-D was recommended to better differentiate stroke survivors with subtle depressive symptoms.