Home / Resources / Psychological Measures / HAM-D-17
Hamilton Depression Rating Scale:
Scoring and Interpretation
The Hamilton Depression Rating Scale (HAM-D, HDRS-17) is a clinician-rated measure of depression severity built from 17 scored items. Hamilton Depression Rating Scale scoring is a plain sum of item-specific ratings, but the version and interview guide matter, so this guide covers the 0 to 52 total, the usual severity bands, SPSS and R code, the evidence on rater and version effects, and a model reporting paragraph.
What the Hamilton Depression Rating Scale Measures
The Hamilton Depression Rating Scale is a clinician-administered measure of depression severity, originally developed for patients who already had an established depressive disorder. Its long history as an outcome in antidepressant trials makes it familiar to reviewers, but it is a severity rating and not a stand-alone diagnostic interview (Hamilton, 1960; Williams, 2001).
Unlike a self-report questionnaire, the HAM-D relies on a clinician who integrates the interview and observed clinical features into each item rating, so rater training and consistency are design issues rather than footnotes. Reviews have also documented meaningful differences among versions and interview guides. For that reason, a methods section should name the exact version and guide used and not simply refer to “the Hamilton scale” (Williams, 2001; Rosenberg, 2022).
Structure of the HAM-D-17
- Items: seventeen scored symptom items form the usual HAM-D-17 total. Response categories and maximum values differ by item, with some items rated 0 to 2 and others 0 to 4.
- Extended forms: several versions, such as the 21-item form, contain additional items. Do not silently append their ratings to a HAM-D-17 total.
- Subscales: no universally standardized subscale score. HAM-D-6 is a separate, derived short form and its total is not interchangeable with the 17-item total.
- Direction: no reverse-scored items in the conventional scoring. Higher item ratings mean more severe features.
- Administration: ratings require a trained interviewer and a specified interview and scoring guide. The recall period is typically the past week, so check the guide.
- Translations: use a validated version for the intended population and name the language and its source. If no validated translation exists, say so and describe how the version was adapted.
The instrument itself is available from the original Hamilton publication and an authorized interview and scoring manual.
How to Score the HAM-D-17
- Document the exact 17-item form and the interview and scoring guide the raters used.
- Confirm that every item already holds its correct, item-specific rating. The code below does not check that the rating was assigned properly.
- Sum the 17 scored item values to obtain hamd17_total. Under commonly used conventions the maximum is 52.
- Do not replace an unrated item with zero. The sources reviewed here set no universal prorating rule across forms, so the code requires all 17 ratings. Follow the rule in your manual or study protocol, and report how missing ratings were handled.
* Illustrative syntax: hamd1 to hamd17 already hold the final item ratings.
COMPUTE hamd17_total = SUM.17(hamd1 TO hamd17).
EXECUTE.
# data: one row per rating occasion; items hamd1 to hamd17 hold final item ratings.
items <- paste0("hamd", 1:17)
data$hamd17_total <- rowSums(data[items], na.rm = FALSE)
Interpreting HAM-D-17 Scores
| Score band | Interpretation |
|---|---|
| 0 to 7 | Often reported as within the normal range |
| 8 to 13 | Mild |
| 14 to 18 | Moderate |
| 19 to 22 | Severe |
| 23 to 52 | Very severe |
Source: published score-band convention summarized in the NCBI Bookshelf outcome-measures chapter; descriptive, not diagnostic, and to be confirmed for the version used.
A higher total indicates greater clinician-rated depressive symptom severity. The 0 to 7, 8 to 13, 14 to 18, 19 to 22 and 23 or higher bands are descriptive conventions and not diagnostic criteria. State whether a value is a baseline, follow-up or change score. A 17-item total also has narrow content, so a person can have substantial symptoms that it captures only partly.
Before comparing HAM-D scores across sites or studies, confirm that the interview guide, rater training and assessment period are comparable. Differences in operational definitions across guides have been documented in the literature.
Psychometric Properties
| Property | Published finding | Source |
|---|---|---|
| Reliability | Evaluated extensively, but estimates differ by rater, protocol and study; no single coefficient is asserted | Williams (2001) |
| Construct coverage | Item content includes sleep and somatic features, and one total may not be unidimensional in every population | Williams (2001) |
| Version effects | Documented differences among traditional, structured and modified scoring guides | Williams (2001); Rosenberg (2022) |
| Short form | HAM-D-6 is a distinct derived instrument and cannot substitute for the HAM-D-17 total | Williams (2001) |
These are source-specific findings and cannot be assumed to hold as reliability estimates in a new sample.
Using the HAM-D-17 in a Study
In a randomized trial, prespecify the follow-up time point and the estimand, for example an adjusted mean outcome, a difference in mean change, or a longitudinal treatment effect. A mixed-effects model can accommodate repeated assessments when its covariance structure and missing-data assumptions are justified. In cross-sectional or multi-site work, consider rater and site effects and document how the interview was standardized. Because the total is bounded and discrete, check its distribution and consider an ordinal sensitivity analysis if a Gaussian model is hard to defend.
For measurement questions, remember that a clinician-rated scale calls for rater agreement, such as an intraclass correlation from a design with multiple raters, and not only an internal-consistency coefficient. If the HAM-D is used to define remission or response, the protocol must prespecify and justify that definition. For a reliability or factor-analytic objective, plan sample size around precision and model complexity and not around a fixed participant-to-item ratio.
Reporting the HAM-D-17
Depression severity was assessed with the 17-item Hamilton Depression Rating Scale (HAM-D-17; Hamilton, 1960), administered by [trained rater type] using [interview version]. The total score was computed by summing the 17 item-specific severity ratings (range = 0 to 52), with higher scores indicating greater symptom severity. Ratings were obtained at [time points]. The analysis accounted for [repeated measures, site or rater, as applicable], and missing ratings were handled by [prespecified rule].
References
Hamilton, M. (1960). A rating scale for depression. Journal of Neurology, Neurosurgery & Psychiatry, 23(1), 56-62. https://doi.org/10.1136/jnnp.23.1.56
Rosenberg, L. I. (2022). The Ham-D is not Hamilton’s Depression Scale. Psychopharmacology Bulletin, 52(2), 117-153. https://doi.org/10.64719/pb.4435
Williams, J. B. W. (2001). Standardizing the Hamilton Depression Rating Scale: Past, present, and future. European Archives of Psychiatry and Clinical Neuroscience, 251(Suppl. 2), II6-II12. https://doi.org/10.1007/BF03035120
Frequently Asked Questions
Is the HAM-D-17 a depression diagnosis?
No. It is a clinician-rated severity instrument. Diagnostic status should come from a separately specified clinical evaluation or diagnostic interview.
Why do some forms list 21 items?
The Hamilton family includes expanded forms. Only the 17 designated items contribute to the HAM-D-17 total, so report the form and the scoring guide used.
Can I use the severity bands to define remission?
Only if the protocol prespecifies and justifies that endpoint. Remission definitions differ across studies, and the descriptive bands were not derived as a remission criterion.
Can I combine HAM-D and PHQ-9 scores?
No, not by simple addition or substitution. They differ in respondent, content and metric, so analyze them separately unless a validated linking procedure exists.
NEXT STEP
Need HAM-D-17 Data Scored
or Checked?
OnlineSPSS is a statistical consultancy for researchers. Share your design and dataset, and a statistically trained Research Coordinator reviews it. The scope and one fixed project fee are confirmed before work begins.
No Commitment Required
Strictly Confidential
Fixed Project Fee
In-House PhD-Level Statisticians
Same-Day Response
Serving Researchers Since 2014