LAUSR

BACKGROUND The Geriatric Depression Scale (GDS) has been widely used for late-life depression, but it lacks validation in psychological autopsy research. This study aimed to assess the validity and establish the optimal cut-off values of the GDS-30 and the GDS-15 with proxy-based data in rural China. METHODS We applied psychological autopsy to collect data from 242 consecutive suicide cases and 242 paired living community controls. RESULTS Subject-proxy concordance for the GDS-30 (ICC = 0.590) and the GDS-15 (ICC = 0.539) were fair in the living controls. Based on proxy-data, we found that the suicide cases had higher scores of depression than the living controls did; the values of Cronbach's alpha demonstrated good internal consistency of the GDS-30 and the GDS-15; the Spearman correlation analysis indicated that the GDS scores were correlated with hopelessness, loneliness, and quality of life. For suicide cases, the GDS-30 showed the highest Youden's index as 34.86% with a cut-off value at 22, when its sensitivity and specificity was 0.78 and 0.56, respectively; the score of 12 on the GDS-15 showed the highest Youden's index of 31.39%, and its sensitivity and specificity was 0.74 and 0.58, respectively. LIMITATIONS This study is limited to its generalizability to Chinese urban elderly with psychological autopsy method. CONCLUSIONS The GDS-30 and the GDS-15 were both valid tools for measuring the severity of depressive symptoms rather than screening for major depression in psychological autopsy research in rural China. The GDS-15 can be considered as a good substitute for the GDS-30.