Problems in assessment of depression in Japanese community-dwelling elderly by CES-D(Center of Epidemiologic Studies Depression) scale

This study aimed to determine the reliability and validity of items of the Japanese version of the CES-D (Center of Epidemiologic Studies Depression) scale when applied to independent community-dwelling Japanese elderly from the viewpoints of internal consistency and goodness-of-fi t for the model. Subjects were 1791 community-dwelling independent elderly people aged 60 and over. This study used the Japanese version of the CES-D and GDS(Geriatric depression scale) scales. Principle component analysis was applied to the correlation matrix of 20 items of the CES-D scale to examine consistency of the scale. The indices of goodness-of-fi t index (GFI) and the adjusted goodness-of-fi t index (AGFI) were also calculated. Reliability of the CES-D scale was examined by Cronbach alpha coeffi cients and split-half reliability by the Spearman-Brown formura. Model fi tting for the fi rst-order factorial model of the Japanese version of the CES-D scale was good (GFI = 0.883, AGFI = 0.855), and the reliability as a psychologiical scale was also suffi cient (Cronbach’s alpha coeffi cient = 0.811 and split-half reliability = 0.823). A correlation coeffi cient between total scores of the GDS and the CES-D scale was moderate but signifi cant (r = 0.653, p<0.01). However, for four items of the CES-D scale, suffi cient validity was not obtained in examining the internal consistency of the scale. A major problem as an assessment scale of depression was not recognized in the Japanese version of the CES-D scale, but the individual items should be re-examined in further studies.