Validity and test–retest reliability of the Swedish version of the Geriatric Depression Scale among very old adults

Abstract

Abstract Background The Geriatric Depression Scale (GDS) has shown good validity and reliability, but few studies have examined the GDS among very old adults or the Swedish translation. Objectives Evaluate the validity and reliability of the Swedish version of GDS-15 among very old adults. Methods In the Umeå85 + /GErontological Regional DAtabase (GERDA) study, 387 participants were assessed with both the GDS-15 and the Montgomery-Åsberg Depression Rating Scale (MADRS). The mean age was 91 years. Concurrent validity between the scales was calculated using Spearman's correlation. We used the Diagnostic and Statistical Manual of Mental Disorders (DSM) V symptom criteria for depression based on MADRS item scores to define depression. We calculated the Area Under the Curve (AUC) and found an optimal cut-off. A convenience sample with 60 individuals was used to calculate test–retest reliability with Cohen’s kappa and Intraclass Correlation Coefficient (ICC). Results Spearman's correlation coefficients between total scores for GDS-15 and MADRS were 0.60. Cronbach's alpha for the whole scale was 0.73. The AUC was 0.90 for distinguishing major depression, and the recommended cut-off of ≥ 5 showed a sensitivity of 95.2% and specificity of 65.8%. The test–retest showed that Cohen’s kappa was substantial (0.71) and the ICC was excellent (0.95). Conclusions The Swedish version of the GDS-15 showed good validity and reliability among very old adults. The generally recommended cut-off of ≥ 5 seems reasonable to use with the Swedish version and among very old adults.

Keywords

• Aged 80 and over
• Depression
• Psychometrics
• ROC curve
• Psychiatric status rating scales