Abstract
This study aimed to develop from the Korean Form of Geriatric Depression Scale (KGDS) a 15‐item short version (KGDS‐15); to analyse the internal consistency and criterion validity of KGDS and KGDS‐15; and to investigate the possible factors contributing to misclassification. The KGDS for screening of depression and interviews with the Hamilton Rating Scale for Depression (HAM‐D) for diagnosis of depression were performed in community‐dwelling, cognitively impaired elderly Koreans. The KGDS‐15 items were chosen from those in the KGDS showing high item‐total score coefficients calculated using Cronbach's alpha. For both KGDS and KGDS‐15, the internal consistencies were good, and areas under the receiver operating characteristics curves served by HAM‐D as a gold standard, were acceptable. Lower level of education, being currently unemployed, living in rented accommodation and having chronic medical illness were independently associated with the outcome and the misclassification of the KGDS. The KGDS‐15 was proved to be equally or more valid and less affected by above factors than the KGDS. The newly developed KGDS‐15, as well as the KGDS, could be used for screening for depression overall in the community‐dwelling cognitively impaired elderly, but the resulting scores should be interpreted carefully. Copyright © 2001 Whurr Publishers Ltd.