The GDS-15 is a short yes/no depression screen designed for older adults. This page covers its structure, scoring key and cut-offs, and how to run it as a clinician-administered eCRF or a participant questionnaire.
Free sandbox · No credit card · 21 CFR Part 11 aligned
Week 12 total
6 / 15
GDS-15 at a glance
Licensing
Free to useRights holder: Yesavage, Sheikh and colleagues (Stanford)Stanford's GDS page states the original scale is in the public domain because it was partly the result of federal support. The 15-item short form is Sheikh and Yesavage (1986). Translations may be hosted with their own contacts, so confirm the terms of the version you use.
In Capture: No licence is needed. Enter the published wording and use official translations.
Stanford GDS pageInstrument names belong to their rights holders, and Capture is not affiliated with or endorsed by them. Terms change, so confirm them with the rights holder before your study starts. This is general information, not legal advice.
The instrument
The GDS was developed by Yesavage and colleagues as a 30-item screen for depression in older adults, and Sheikh and Yesavage derived the 15-item short form in 1986 from the items that correlated most strongly with depressive symptoms. Its yes/no answers take a few minutes, need no rating scale to explain and can be read aloud, which makes it practical in clinics, care homes and community cohorts.
In trials it screens for depression at enrolment (often as an exclusion or a stratification factor), tracks mood as a safety or secondary outcome in frailty and cognitive studies, and supports observational cohorts. Where a clinician rating is the endpoint, the Hamilton Depression scale is the usual choice; for general adults the PHQ-9 is free to reproduce and widely used. In older adults with cognitive impairment, check validity: the GDS-15 is less reliable as cognition falls, so pair it with a cognitive measure such as the MoCA and state your rule in the protocol.
Each answer is stored as the coded value yes or no. The point for each item depends on its position in the key, so the total is derived in analysis from the exported codes. It is not one of Capture's built-in calculated fields (BMI, eGFR, QTc and similar clinical derivations). A common rule is to score the scale only when most items are answered; whichever rule you choose, write it in the statistical analysis plan, and make every item required to avoid the issue.
Structure
| Element | Detail |
|---|---|
| Items | 15, each answered yes or no |
| Item scoring | 1 point for a depression-indicating answer: "no" on items 1, 5, 7, 11, 13 and "yes" on the other ten |
| Total | 0 to 15, higher is worse |
| Recall period | The past week |
| Common bands | 0 to 4 normal, 5 to 8 mild, 9 to 11 moderate, 12 to 15 severe |
| Screening threshold | Often above 5; some studies use 5 or more |
| Who completes it | Self-report, or read aloud by site staff |
| Rights status | Public domain per the Stanford GDS page |
Take the official item text and any translation from the Stanford GDS page or the translation's own source, and keep order and wording as published.
Two ways to run it
Because the GDS-15 is read aloud as often as it is self-completed, build it the way your protocol collects it.
Evening diary
Question 3 of 8
How would you rate your fatigue today?
0 = no fatigue, 10 = worst imaginable
Licensing
Capture's licence controls exist for participant questionnaires only. If you run the GDS-15 as an ePRO and want to record where your version came from, you can switch on "Licence required" and enter a number, holder and expiry, but a public-domain scale does not need one. If you run it as a site-completed eCRF, the form has no licence controls at all, and the provenance of your translation stays with the study documents. Capture does not verify licences with rights holders and does not supply instrument wording.
Copyright & licence
Approval pendingLicence number
Not recorded
Expiry date
Not recorded
Watermark shown to staff only. Participants never see it.
Setup
Take the 15-item English form from Stanford or your approved translation, and note the source in the study documents.
Add 15 single-choice items with coded values for yes and no, in the published order. Add "completed by" and "administered by site staff" fields.
Required fields stop partial forms that cannot be scored.
Add a custom-value rule that prompts follow-up per your protocol, and range checks on any date or time fields.
Place it at screening and the follow-up visits, no more often than the one-week recall period allows.
Put the reversed items (1, 5, 7, 11, 13) and the cut-offs in the analysis plan and data dictionary notes.
Worked example
A participant answers "no" to items 1, 5, 7 and 13 and "yes" to item 11. On those five keyed-no items, four score a point. Across the other ten items, two "yes" answers score a point each. The total is 6, which falls in the mild band and above a screening threshold of above 5. The same participant scored 9 at baseline. The analyst reads the change of three points alongside the clinical context and does not treat a screening tool as a diagnosis.
This arithmetic is the reason to export item codes and score in a script that is checked once against a few hand-scored forms. See calculated fields and skip logic for what Capture derives in the form and what stays in analysis.
Before go-live
English or translated form, with where it came from.
Hand-score five forms and compare with the analysis script.
Self-completed or read aloud, recorded per administration.
Protocol says when the GDS-15 is not used or is interpreted with caution.
Who acts on a positive screen and how it is recorded.
Build it in the free sandbox, schedule it and open the participant link before you go live.
Stanford's GDS page states the original scale is in the public domain because it was partly federally funded. Translations may carry their own terms, so confirm the version you use.
Give one point for each depression-indicating answer: "no" on items 1, 5, 7, 11 and 13, and "yes" on the other ten. The total runs from 0 to 15.
Common bands are 0 to 4 normal, 5 to 8 mild, 9 to 11 moderate and 12 to 15 severe. Screening thresholds vary between studies (above 5, or 5 or more), so state yours in the protocol.
No. Capture stores each answer and exports a data dictionary. The total is derived in analysis, because Capture's calculated fields are a fixed set of clinical derivations such as BMI and eGFR.
Yes, the format allows it. Build it as a site-completed eCRF and record who administered it, or as an ePRO if participants answer alone.
No. It is a screening tool. Diagnosis needs clinical assessment.
No. It gives structure and scoring. Take the item text from the Stanford page or your translation's source so the wording stays exactly as published.
Keep exploring
Build the form, test the scoring export and open it on your phone. Free sandbox, no credit card.