Validation and Comparison of Three Short Depression Screening Tools Among Chinese Community-Dwelling Older Adults

The current study validated and compared three short Geriatric Depression Scales (GDS), including the GDS-5, D'Ath GDS-4, and van Marwijk GDS-4, among 917 Chinese community-dwelling older adults. The GDS-5, D'Ath GDS-4, and van Marwijk GDS-4 presented satisfactory accuracy against the GDS-15 (area under the curve [AUC] = 0.872 to 0.952), and the GDS-5 and D'Ath GDS-4 had better accuracy than the van Marwijk GDS-4. Satisfactory accuracy (AUC = 0.842 to 0.979) for the three scales was also observed across subgroups by age, sex, education, cognitive function, and multimorbidity. The GDS-5 but not D'Ath GDS-4 and van Marwijk GDS-4 retained a 2-point optimal cutoff for depressive symptoms across subgroups. The GDS-5 (average inter-item correlation coefficient [AIIC] = 0.233) and the D'Ath GDS-4 (AIIC = 0.171) but not van Marwijk GDS-4 (AIIC = 0.128) had acceptable internal consistency. Three scales had stable test-retest reliability within a 1- to 2-week interval (intraclass correlation coefficient = 0.670 to 0.885). The GDS-5 is an accurate and reliable depression screening tool with an invariable optimal cutoff among Chinese community-dwelling older adults. The variable optimal cutoffs for the D'Ath GDS-4 and van Marwijk GDS-4 across subgroups may limit their applicability in this population. [Research in Gerontological Nursing, 15(6), 283-291.].