Identifying Depression in Dementia: A Meta-Analysis

There is a need for an accurate tool that is quick to use and easy to implement to diagnose depression in dementia.

Results from a systematic review and meta-analysis examining tools to detect depression in dementia show that tools incorporating collateral histories as well as a clinician interview with the patient, such as the Cornell Scale for Depression in Dementia (CSDD) or the Hamilton Depression Rating Scale (HDRS), have higher sensitivity for detecting depression, thus reducing false negatives.

Zahra S. Goodarzi, MD, from the Department of Community Health Sciences and the Department of Medicine at the University of Calgary and Alberta Health Services in Canada, and colleagues narrowed 11,539 citations down to 20 studies to include in a qualitative synthesis and to 15 studies to include in a meta-analysis.

The researchers investigated the following tools for detecting depression:

  • Montgomery Asberg Depression Rating Scale
  • CSDD
  • Geriatric Depression Scale (GDS)
  • Center for Epidemiologic Studies Depression Scale
  • HDRS
  • Single Question
  • Nijmegen Observer-Rated Depression Scale
  • Even Briefer Assessment Scale–Depression

The pooled prevalence of depression in those with dementia was 30.3% (95% CI, 22.1%-38.5%), the average age was 75.2 years (95% CI, 71.7-78.7 years), and the mean Mini-Mental State Examination scores ranged from 11.2 to 24.8.

The researchers found that the CSDD had a sensitivity of 0.84 (95% CI, 0.73-0.80) and a specificity of 0.80 (95% CI, 0.65-0.90), the 30-item

GDS (GDS-30) had a sensitivity of 0.62 (95% CI, 0.45-0.76) and a specificity 0.81 (95% CI, 0.75-0.85), and the HDRS had a sensitivity of 0.86 (95% CI, 0.63-0.96) and a specificity of 0.84 (95% CI, 0.76-0.90).

“Part of the diagnostic challenge for depression in dementia is related to overlapping symptomatology, such as apathy, poor memory or concentration, which precludes accurate or timely diagnosis,” the researchers wrote. “This is compounded by the potentially decreased ability to convey symptoms or a lack of insight, which can be common in dementia.”

As dementia worsens, caregivers often have to report symptoms on behalf of the patient, and they tend to report more depressive symptoms than patients do. This may be because they themselves are depressed or stressed from the burden of caregiving, or it may be because the patients with dementia underreport depression symptoms.

This demonstrates a need for an accurate tool that is quick to use and easy to implement to diagnose depression in dementia.

Analysis of the Tools for Identifying Depression in Dementia

The researchers identified 14 tools that were validated in this study population.

Brief tools such as the GDS-4 (sensitivity, 38%-76%), EBAS-DEP (sensitivity, 46%), and Single Question (sensitivity, 59%) appeared to have low sensitivity in this population. “It is possible that these brief tools do not capture the complexity of symptoms in individuals with dementia, resulting in lower sensitivity,” the investigators wrote.

Results from the meta-analysis show that the HDRS and CSDD have similar sensitivities (86% and 84%, respectively), and that they both also have areas under the summary receiver operating characteristic curve (0.89) that indicate higher discriminatory ability.

“This higher discriminative ability of the HDRS and CSDD may be related to incorporation of an interview as part of the tool, which obtains information from individuals with dementia and their caregivers,” the researchers wrote. “The CSDD has the added benefit of being designed specifically for individuals with dementia, whereas the HDRS and GDS were not.”

The GDS’s lower sensitivity (62%) indicates a high false-negative rate. This may be a result of a problem with individuals lacking self-awareness when they self-report.

“The [collateral source GDS] has been examined to address this. For the [collateral source GDS]-15 and -30, the sensitivity increases to 70%,” the researchers wrote. “This again is still less discriminating than some of the other tools identified and perhaps relates to the differing nature of depression in older adults with and without dementia.”

Summary and Clinical Applicability

The CSDD and HDRS were identified as tools that were more sensitive in detecting depression in outpatients with dementia. This may be a result of included interviews with the patient and surrogate.

“Practitioners should become comfortable with an arsenal of tools so that they can adapt to different clinical settings. Any positive case-finding tool should be followed by an in depth assessment to establish a diagnosis and follow-up,” the investigators concluded.


  • Most studies excluded patients with severe dementia or communication deficits. This may bias the sample to mild or moderate dementia and may exclude certain types of dementia.
  • A lack of consistent reporting of dementia subtype or severity of deficits prevented further subgroup analysis of those variables.
  • Diagnostic accuracy studies may have certain biases, such as observer variation and differential verification bias.

Disclosures: Dr Zahra S. Goodarzi reports receiving funding from the Canadian Institutes for Health Research (CIHR) Canadian Graduate Student Scholarship, Alberta Innovates Health Solutions Clinician Fellowship Award, and Western Regional Training Centre affiliate award. Dr Derek J. Roberts reports support by an Alberta Innovates Health Solutions Clinician Fellowship Award, a Knowledge Translation Canada Strategic Training in Health Research Fellowship, a Knowledge Translation Canada Research Stipend, and funding from the CIHR. Dr Jayna Holroyd-Leduc reports receiving grant funding unrelated to this project from the CIHR and Alberta Innovates Health Solutions.

Related Articles


Goodarzi ZS, Mele BS, Roberts DJ, Holroyd-Leduc J. Depression case finding in individuals with dementia: a systematic review and meta-analysis [published online February 2, 2017]. JAGS. doi: 10.1111/jgs.14713