Depression among older adults is under-recognized either in the community or in general hospitals in Chinese culture. This study aimed to develop a culturally appropriate screening instrument for late-life depression in the non-psychiatric settings and to test its reliability and validity for a diagnosis of depression.

Using a Delphi method, we developed a geriatric depression inventory (GDI), consisting of 12 core symptoms of depressive disorder in old age. We investigated its reliability and validity on 89 patients with late-life depression and 249 non-depression controls. Both self-report (GDI-SR) and physician-interview (GDI-RI) versions were assessed.

Cronbach's α coefficient was 0.843 for GDI-SR and 0.880 for GDI-RI. Both GDI-SR and GDI-RI showed good concurrent validity with the 15-item Geriatric Depression Scale (GDS-15) (GDI-SR: r = 0.750, p < 0.001; GDI-RI: r = 0.733, p < 0.001). The area under the curve of the receiver operating characteristic (ROC) was 0.938 for GDI-SR and 0.961 for GDI-RI, suggesting good to excellent discrimination of depression versus non-depression. Using a cut-off of three items endorsed, sensitivity and specificity were 92.1% and 81.9% for GDI-SR, and 93.3% and 87.1% for GDI-RI.

The GDI, either based on self-report or rater interview, is a reliable and valid instrument for the detection of depression among older adults in non-psychiatric medical settings in Chinese culture.