Short Version Of The Geriatric Depression Scale: Exploratory Study Of Its Validity In a Portuguese Sample

Background: To explore the validity of the short Geriatric Depression Scale (GDS) version – 5 items – in Portuguese community-dwelling older adults. Depression is reported to affect between 1% and 3% community-dwelling older adults, whereas depressive symptoms are two to four times more prevalent. Community surveys targeting older adults should have reliable and valid instruments to assess depressive symptoms. Methods: Two samples were considered: a convenience sample (N=71) to compare GDS5 with the 15-item GDS version and the Beck Depression Inventory (BDI); and a population-based sample (N=1030) to compare GDS5 with self-reported depression diagnosis. Internal consistency (Cronbach's alpha coefficient), criterion validity (sensitivity and specificity), diagnostic capacity (area under the ROC curve) and factorial analysis (tetrachoric) were analyzed. Results: The cut-off point was taken as 2 or more. GDS5 was found to be internally consistent: Cronbach Alpha=0.69. GDS-5 and GDS15: Sensitivity= 85.7%; Specificity=83.3%, AUC=94.3. GDS5 and BDI: Sensitivity=70%; Specificity=85.3%, AUC=84.1. GDS5 and self-reported depression: Sensitivity=79%; Specificity=47.4%, AUC=72.2. Conclusions: Results are similar to other GDS5 studies. The low specificity results for the depression diagnose indicating false positives can be due to self-reporting. GDS5 may be a useful alternative to other longer instruments not adequate for research surveys.FCT - Foundation for Science and Technology (PDTC/CS-SOC/110311/2009