Is the geriatric depression scale a reliable screening tool for depressive symptoms in elderly patients with cognitive impairment?

To determine the reliability of the 30-item Geriatric Depression Scale (GDS-30) for the screening of depressive symptoms in dementia and mild cognitive impairment (MCI) using the Cornell Scale for Depression in Dementia (CSDD) as the 'gold standard'.Diagnosed according to strictly applied clinical diagnostic criteria, patients with MCI (n = 156) and probable Alzheimer's disease (AD) (n = 247) were included. GDS-30, CSDD, Mini Mental State Examination (MMSE) and Global Deterioration Scale were assessed in all patients at inclusion. The AD group was divided in three subgroups: mild AD (MMSE>or=18) (n = 117), moderate AD (MMSE< 18 and >or=10) (n = 89) and severe AD (MMSE<10) (n = 38).In MCI patients, moderate but highly significant correlations were found between GDS-30 and CSDD scores (Pearson: r = 0.565; p < 0.001). In mildly (r = 0.294; p = 0.001), moderately (r = 0.273; p = 0.010) and severely (r = 0.348; p = 0.032) affected AD patients, only weak correlations between GDS-30 and CSDD scores were calculated. ROC curve analysis showed that sensitivity and specificity values of respectively 95% and 67% were achieved when a GDS-30 cut-off score of 8 was applied in MCI patients. In AD patients, too low sensitivity and specificity values did not allow selecting an optimal cut-off score by means of ROC curve analysis.Using the CSDD as 'gold standard', we demonstrated that the GDS-30 is a reliable screening tool for depressive symptoms in MCI but not in AD patients.