Validity of the Geriatric Depression Scale-30 against the gold standard diagnosis of depression in older age: The GreatAGE Study

Introduction Depression is a common disorder in late-life. Structured clinical interviews may be less efficient compared to self-administered questionnaires, but provide more accurate findings in terms of diagnosis. No population-based studies with both these depression assessment instruments have been ever performed. Objectives To estimate the GDS-30 accuracy for depression assessment against the gold standard [Semi-structured Clinical Diagnostic Interview for DSM-IV-TR Axis I Disorders (SCID)] in subjects 65+ years in a random sampling of the general population. Methods The sample was collected in a population-based study (GreatAGE) conducted among elderly residents in Castellana, Southeast Italy. It includes 597 participants (57.62% males, mean age 73 years). Depression was assessed through the GDS-30 and the SCID, both double-blinded administered respectively by a trained neuropsychologist and psychiatrist. The GDS-30 screening performances were analyzed using ROC curves. Results According to the gold standard SCID, the rate of depressive disorder was 10.22% (15.81% of women; 6.1% of men) while with GDS-30 instrument 12.06% of the residents met the depression cutoff. Only 36.1% of GDS cases were true positive. At the optimal cutoff score (> 5), GDS had 62% sensitivity and 81% specificity. Using a more conservative cutoff (> 9), the GDS-30 specificity reached 91% while sensitivity dropped to 43%. Conclusions These preliminary results from the first population-based study that compares GDS-30 and SCID showed that the GDS-30 identified adequate levels of screening accuracy (AUC 0.76) compatible with scores established in community settings. Funding PRIN2009E4RM4Z.