Background: Depression is highly prevalent in nursing home residents living with moderate to severe dementia. However, assessing depressive symptoms in residents with dementia can be challenging and may vary by rater perspective. We aimed to investigate the concordance of, and factors associated with self- and informant-rated depressive symptoms in nursing home residents with dementia.
Methods: Cross-sectional data was collected from N= 162 nursing home residents with dementia (age: 53-100; 74% women). Self-ratings were assessed with the Geriatric Depression Scale, while the depression and anxiety items of the Neuropsychiatric Inventory were used for informant-ratings. Cohen's Kappa was calculated to determine the concordance of both measures and of each with antidepressant medication. Multivariate associations with sociodemographic variables, self- and informant-rated quality of life, dementia stage, neuropsychiatric symptoms, functional status and antidepressant medication were analysed with linear mixed models and generalized estimating equations.
Results: Concordance between self- and single item informant-rated depressive symptoms was minimal (Cohen's Kappa = .22, p= .02). No concordance was found for self-reported depressive symptoms and the combined informant-rated depression-anxiety score. Self-reported depression was negatively associated with self-rated quality of life (beta=-.32; 95%CI: -.45 to -.19, p< .001), informant-rated quality of life (beta=-.25; 95%CI: -.43 to -.07, p= .005) and functional status (beta=-.16; 95%CI: -.32 to -.01, p= .04), whilst single item informant-rated depression revealed negative associations with informant-rated quality of life (beta =-.32; 95%CI: -.52 to .13, p=.001) and dementia stage (beta=-.31; 95%CI: -.52 to -.10, p = .004). The combined informant-rated depression-anxiety score showed negative associations with self-rated quality of life (beta=-.12; 95%CI: -.22 to -.03, p = .01) and dementia stage (beta = -.37; 95%CI: -.67 to -.07, p= .02) and a positive association with neuropsychiatric symptoms (beta = .30; 95%CI: .10 to .51, p= .004). No concordance was found with antidepressant medication.
Conclusions: In line with our expectations, low agreement and unique association patterns were found for both measures. These findings indicate that both instruments address different aspects of depression and underline the need for comprehensive approaches when it comes to detecting signs of clinically relevant depressive symptoms in dementia.
