Introduction The validity of the 30-item Geriatric Depression Scale (GDS-30) in detecting late-life depression (LLD) requires a certain level of cognitive functioning. Further research is needed in population-based setting on other socio-demographic and cognitive variables that could potentially influence the accuracy of clinician rated depression. Objective To compare the diagnostic accuracy of two instruments used to assess depressive disorders [(GDS-30) and the Semi-structured Clinical Diagnostic Interview for DSM-IV-TR Axis I Disorders (SCID)] among three groups with different levels of cognitive functioning (normal, Mild Cognitive Impairment – MCI, Subjective Memory Complain – SMC) in a random sampling of the general population 65+ years. Methods The sample, collected in a population-based study (GreatAGE Study) among the older residents of Castellana Grotte, South-East Italy, included 844 subjects (54.50% males). A standardized neuropsychological battery was used to assess MCI, SMC and depressive symptoms (GDS-30). Depressive syndromes were diagnosed through the SCID IV-TR. Socio-demographic and cognitive variables were taken into account in influencing SCID performance. Results According to the SCID, the rate of depressive disorders was 12.56%. At the optimal cut-off score (≥ 4), GDS-30 had 65.1% sensitivity and 68.4% specificity in diagnosing depressive symptoms. Using a more conservative cut-off (≥ 10), the GDS-30 specificity reached 91.1% while sensitivity dropped to 37,7%. The three cognitive subgroups did not differ in the rate of depression diagnosis. Educational level is the only variable associated to the SCID diagnostic performance ( P = 0.015). Conclusions At the optimal cut-off, GDS-30 identified lower levels of screening accuracy for subjects with normal cognition rather than for SMC (AUC 0.792 vs. 0.692); educational attainment possibly may modulate diagnostic clinician performance. Disclosure of interest The authors have not supplied their declaration of competing interest.
原文链接
分享
引用
