Background: Few studies have evaluated the validation of 15-item Geriatric Depression Scale (GDS-15) in a heterogeneous population with different age, ethnicity and comorbidities of elderly users of social services in the community.
Aims: To assess the criterion validity and reliability of the GDS-15 and its equivalence across different gender, age groups, ethnicity and different comorbidities in community living elderly and nursing homes residents.
Method: A validation sample of non-demented 4253 elderly (age > or = 60 years), who regularly use community based care corner, senior activity center, day care center, sheltered homes and nursing homes were interviewed using the GDS-15. Structured clinical interview (SCID) was used to make DSM-IV diagnosis of major depressive disorder (MDD).
Results: The overall sensitivity and specificity were 0.97 and 0.95, respectively (area under curve, AUC was 0.98). The overall Cronbach's alpha was 0.80, and intraclass coefficient of test--retest reliability over 2 weeks was 0.83 and inter-rater reliability was 0.94 (intra-class) and 0.99 (Cohen's kappa). Although some items in the GDS-15 appeared to be biased by gender, age and ethnicity, there were no clinically significant differences in test performance among different age, gender, ethnicity and comorbidities at cutoff of 4/5.
Conclusions: The GDS-15 was a reliable and valid screening for MDD across different age, gender, ethnicity and chronic illness status in the community and social service setting.