Measuring Self-report Symptoms of Depression in Organ Transplant Patients: Validation of a Multidimensional Model of CES-D Item Data

• Activity Number: 21
• Type: Contributed
• Date/Time: Sunday, August 8, 2004 : 2:00 PM to 3:50 PM
• Sponsor: Section on Health Policy Statistics
• Author(s): Irene D. Feurer*+ and Hongxia Liu and Kenneth T. Thomas and Ronald M. Salomon and Theodore Speroff and C. Wright Pinson
• Companies: Vanderbilt Transplant Center and Vanderbilt School of Nursing and Transplant Center and Vanderbilt Transplant Center and Vanderbilt University Medical Center and Vanderbilt University Medical Center and Vanderbilt Transplant Center
• Keywords: CES-D ; instrument scaling ; latent variable structure ; depression ; organ transplantation

Abstract

This research evaluated the latent variable structure and implications for scoring a depression (DEP) screening instrument in organ transplant patients. Adult organ transplant patients completed the Center for Epidemiologic Studies Depression Scale (CES-D) and the 36-item Short Form (SF-36). Responses (N=1264) were randomly assigned to exploratory (EXP, n=651) and confirmatory (CON, n=613) samples. Statistical methods included EXP principal component analysis, and CON bifactor and discriminant analysis. An EXP three-component solution that accounted for 54% of the variance was interpreted as somatic symptoms, isolation, and positive affect. A CON bifactor model (60% variance) with a general factor (alpha=.84) and three group factors was superior to a two-group model (chi-square p<.001). A CON discriminant function based on the three factor scores (model p<.001) successfully classified 87% of cases (sensitivity=70%, specificity=93%, chi-square p<.001) in relation to the SF-36 mental component norm (50+/-10). Conclusion: A multidimensional CES-D scoring system is indicated for transplant patients. Funding: AHRQ R03-HS13036 and Roche Laboratories, Inc.