Depression detection in dementia: A diagnostic accuracy systematic review and meta analysis update

Depression is common in persons with dementia and is often under-detected and under-treated. It is critical to understand which available tools accurately detect depression in the context of dementia.

We updated our systematic review completed in 2015. The search strategy of our original review was replicated in Medline, Embase, and PsycINFO. Studies describing the use of a tool to identify depression in persons with dementia, compared to a criterion standard, and reporting diagnostic accuracy outcomes were included in the review update. Pooled prevalence estimates of major depression and pooled estimates of diagnostic accuracy outcomes (i.e., sensitivity [SN], specificity [SP]) for tools were calculated.

Three studies were included of the 8980 returned from the database search and were added to the prior 20 articles from the 2015 review. The Cornell Scale for Depression in Dementia (CSDD), Geriatric Depression Scale (GDS)-15 item, Neuropsychiatric Inventory-Depression items (NPI-D), and Depression in Old Age Scale (DIA-S) were evaluated in the three studies. Two new studies were added to the existing pooled prevalence estimate of major depression (29%, 95% confidence interval [CI] = 21.6%-36.5%, n = 17) and pooled diagnostic accuracy estimate for the CSDD at the best cut-off (SN = 0.83, 95% CI = 0.74-0.90; SP = 0.81, 95% CI = 0.69-0.89). New pooled diagnostic accuracy estimates were completed for the CSDD (cut-off ≥12) (SN = 0.61, 95% CI = 0.42-0.77; SP = 0.83, 95% CI = 0.76-0.88), GDS-15 (best cut-off) (SN = 0.65, 95% CI = 0.40-0.83; SP = 0.72, 95% CI = 0.55-0.85), and Montgomery Asberg Depression Rating Scale (MADRS) (best cut-off) (SN = 0.77, 95% CI = 0.67-0.85; SP = 0.68, 95% CI = 0.60-0.75).

The CSDD continues to have the most evidence for depression case finding in persons living with dementia. The CSDD and Hamilton Depression Rating Scale have the highest sensitivities and may be recommended for use over other common tools like the GDS-15 and MADRS. Newly identified tools like the NPI-D and DIA-S require further study before they can be recommended for use in practice.