Geriatric Depression Scale (GDS)

This instrument measures depression in the elderly. (Ref: 1) It was specifically designed for older populations, which require items that are easier to read and understand. Depression screening with this measure in older adults with generalized anxiety disorder has also been tested. (Ref: 2) There are 30 items that generally fit on one page.

Free (original version in public domain)

10-15 minutes

1982

Flesch-Kincaid grade level of 4.1. Items are simple sentences, usually less than 15 words each.

Yes/No answer choices for all items

Typically administered by clinicians, researchers, or trained personnel, psychiatric expertise is not necessary.

"Yes" items are summed. Higher scores indicate more depression. Using a receiver operating characteristic (ROC) curve, researchers in one study found that the optimal cutting score for depressed and non-depressed patients was 19, with sensitivity and specificity of 61% and 58%, respectively. (Ref: 3) Other researchers found that a score of 11 or higher is considered as depressive, achieving 84% sensitivity and 95% specificity at this level. (Ref: 1)

There are multiple language translations available. There are also short forms available. See the developer’s website below for more information on other forms.

Jerome A. Yesavage, T. L. Brink, Terence L. Rose, Owen Lum, Virginia Huang, Michael Adey and Von Otto Leirer

Department of Psychiatry and Behavioral Sciences
Stanford University of Medicine
Stanford, CA 94305

Veterans Affairs Medical Center
3801 Miranda Ave
Palo Alto, CA 94304

Geriatric Treatment Team
Santa Clara County Mental Health Department
828 S. Bascom Ave., Suite 200
San Jose, CA 95128 USA

yesavage@stanford.edu

www.stanford.edu/~yesavage/GDS.html

top

Initially, 100 items were administered to 47 male and female subjects over 55 years of age, comprised of those with no history of mental illness or complaints of depression and those hospitalized for mental illness (including depression.) The final 30 items were chosen based on the rationale that the 100-item scale should have prima facia validity for depression and that the items that correlated best with the total score would indicate depression. (Ref: 1) Items were chosen to assess topics relevant to depression, such as cognitive complaints, future/past orientation, self-image, losses, agitation, obsessive traits, and motivation. (Ref: 1) In a principal components analysis using data from 326 community-dwelling elderly subjects, five factors were identified after varimax rotation. Factor 1 reflected sad mood and pessimistic outlook. Factor 2 reflected lack of mental or physical energy. Factor 3 related to positive or happy mood, and Factor 4 represented agitation or restlessness. Factor 5 was described as a social withdrawal factor. (Ref: 5)

No information found. All studies cited used less than 100 persons in their respective samples. (Ref: 1-3) The exception was the study conducted among Veterans, which used 314 patients. (Ref: 4)

Using 20 subjects twice, one week apart, was 0.85 (p<0.001). (Ref:1) The agreement between two administrations of the T-CS-GDS was above 70% for all 30 items. (Ref: 3)

No information found.

Median correlation with the total score was 0.54; mean inter-item correlation was 0.36; Cronbach’s alpha coefficient was 0.94; split half reliability was 0.94. (Ref: 1) In older adults with GAD, Cronbach’s alpha was found to be 0.73. (Ref: 2)

The telephone administered GDS, or T-CS-GDS, was found to correlate at 0.87 with the original GDS. (Ref: 3)

The researchers provided validity evidence for the GDS based on subject classification of normal, mildly depressed, or severely depressed based on Research Diagnostic Criteria for major affective disorder. Non-depressed persons were expected to receive the lowest sores while the severely depressed individuals were expected to receive the highest scores. Analysis of variance was conducted using the classification as the between-subject factor with total scores as dependent variables to test this hypothesis. Results indicated that a significant difference existed between the groups, F (2, 97)=99.48, p<0.001 (Ref: 1) In patients with GAD, validity evidence was presented as different GDS scores for groups with and without depression [t(52) = -3.5, p<0.01]. Using t-tests and chi square analyses there was no significant difference between the two groups on age, education or gender variables. (Ref: 2) Correlation between the T-CS-GDS and the CS-GDS was found to be 0.87 (Pearson’s r; p<0.0001.) (Ref: 3)

The GDS was found to be positively correlated with other tested measures for depression. GDS and the Zung Self-Rating Depression Scale = 0.84; GDS and the Hamilton Rating Scale for Depression = 0.83; Zung Self-Rating Depression Scale and Hamilton Rating Scale for Depression = 0.80 (p<0.001 for all). (Ref: 1) The GDS correctly identified 81.3% of GAD subjects without depression and 59.1% of GAD subjects with depression based on two items [X² (2) = 11.4, p<0.005]. (Ref: 2)

By comparing the T-CS-GDS with clinical diagnosis of depression by psychiatrists , the receiver operating characteristic (ROC) was shown to have an area under the curve (AUC) of 0.76 (Wilcoxon z=2.7, p<0.001.) (Ref: 3)

Correlation with the classification variables (following Ferguson (1971)) was 0.82 for GDS, 0.69 for the Zung Self-Rating Depression Scale and 0.83 for Hamilton Rating Scale for Depression (for all, p<0.001). (Ref: 1)

Original GDS items created by team of clinicians and researchers involved in geriatric psychiatry. Scores on GDS correlated to Research Diagnostic Criteria for depression. (Ref: 1)

No information found.

Yes. (Ref: 4)

Yes. (Ref: 1-3)