Abstract
BACKGROUND: The 15-item Geriatric Depression Scale (GDS-15) shows good performance in detecting depression among older persons, but its applicability has not been well studied in non-Western oldest-old adults and centenarians. This study aimed to evaluate the psychometric property of the GDS-15 and a simplified version among a large representative longevous population in China. METHODS: A total of 1624 individuals (786 oldest-old persons aged from 80 to 99 years; 838 centenarians aged 100+ years) participated in this study. Home interviews with structured questionnaires were conducted to collect sociodemographic data. Depressive symptoms were measured using the Chinese GDS-15 version. We implemented mixed methods for the psychometric evaluation of the GDS-15. Cronbach's α coefficient and item-total correlation coefficients were used to evaluate the internal consistency. A standard expert consultation was conducted to test the content validity of each item. Multiple factor analyses were used to explore the optimal factor structure and measurement invariance. RESULTS: The α coefficient of the GDS-15 was 0.745, while two items impaired the overall consistency reliability. Nineteen experts rated the applicability for each item and provided removal suggestion. Five items with less validity were removed, and a simplified 10-item GDS model with three-factor structure was proposed as an optimal solution. The GDS-10 model showed factorial equivalence across age, sex, residence, and education in multi-group confirmatory factor analyses. CONCLUSIONS: The original GDS-15 has acceptable internal reliability, known-group validity, and concurrent validity among Chinese community-dwelling oldest-old and centenarians; however we provided preliminary evidence indicating that individual items related to somatic function or social activities may not be applicable for this population. The modified GDS-10 can be proposed as a potentially more practical and comprehensible instrument for depression screening.