Calculators / Psychosis and global severity
CGI-S and CGI-I
Clinical Global Impressions, Severity and Improvement. Global clinical assessment.
Members Library
The CGI-S and CGI-I calculator is in the Library.
Every rating scale calculator comes with the Library, along with the prescribing guides, medication handouts, DSM-5 criteria and sheets, and the Guide Builder.
Use it for
- A fast global rating of how ill the person is (CGI-S).
- A global rating of how much the person has changed since treatment began (CGI-I).
- Anchoring a condition-specific scale to something every clinician understands, and giving a chart-ready global impression.
Do not use it for
- Replacing a condition-specific scale. The CGI is coarse and shows little about which symptoms are changing.
- Rating change since the last visit. The CGI-I compares with baseline, not with last time.
- Use without a stable idea of what "how ill" means for this condition. It depends on the clinician's reference group.
Rate either or both. For CGI-S, consider your total clinical experience with this population and rate how ill the person is now. For CGI-I, compare the person with their condition at baseline, the point before the treatment started, and rate the change.
CGI-S and CGI-I
Answer the questions to see the result.
This calculator adds up the score in your browser. Nothing you enter is stored or sent.
More about the CGI-S and CGI-I
About the CGI-S and CGI-IWhat it is and what it is for
The Clinical Global Impressions scale, written by William Guy at NIMH in 1976, asks the clinician for one judgment of severity and one of change. The severity rating (CGI-S) is made against the clinician's experience with people who have the same condition. The improvement rating (CGI-I) compares the person's state today with their state at baseline, before the treatment being judged began.
Each runs from 1 to 7. The two are separate, and you can use either one alone. In trials, CGI-I of 1 (very much improved) or 2 (much improved) is the usual definition of a responder.
How to use itGiving it and scoring it
Rate the CGI-S at any visit from your total clinical experience with people who have the same condition. Rate the CGI-I only after treatment has started, against the baseline.
Use the full 1 to 7 range. Do not leave a rating at 4 out of caution.
Interpreting the scoreScore ranges and what they suggest
| CGI-S | What it suggests | Usual response |
|---|---|---|
| 1 | Normal, not at all illNo symptoms or impairment for this population. | |
| 2 | Borderline mentally illQuestionably ill. Subtle or doubtful symptoms. | |
| 3 | Mildly illSymptoms present but with little effect on function. | |
| 4 | Moderately illClear symptoms with some effect on function. | |
| 5 | Markedly illObvious, serious symptoms with marked effect on function. | |
| 6 | Severely illSevere symptoms with severe impairment. | |
| 7 | Among the most extremely illAmong the most severely ill people the clinician has seen with this condition. |
In linking studies in schizophrenia trials, CGI-S ratings of mildly, moderately and markedly ill corresponded to BPRS totals of about 31, 41 and 53 (Leucht 2006). The one-line meanings are plain-language glosses and are not part of the scale.
Pearls and pitfallsTips and common mistakes
- Write down the baseline CGI-S at the first visit. The CGI-I only makes sense against a recorded baseline.
- The CGI-S has a floor. A person with minimal residual symptoms is rated 2, not 1.
- A trial responder is a CGI-I of 1 or 2. Use the same definition when you describe a response in a note.
- Two clinicians often differ by a point on the CGI-S. A change of a single point can be rater variation.
- The original CGI has a third part, the efficacy index, which combines therapeutic effect and side effects. It is rarely used and is not scored here.
EvidenceAccuracy and reliability
- Guy (1976) published the CGI in the ECDEU Assessment Manual. It is the most widely used global measure in psychiatric trials.
- Linking work in schizophrenia trials found that a CGI-S of mildly, moderately and markedly ill corresponded to BPRS totals of about 31, 41 and 53, and that much improved on the CGI-I corresponded to about a 50% fall in the BPRS or PANSS (Leucht 2006).
- Busner and Targum (2007) describe how to train raters and apply the CGI in practice, and note that the two scales differ in what they ask.
Source, license and referencesWhere it comes from
The CGI was developed by William Guy at the National Institute of Mental Health (1976) and is in the public domain.
- Guy W. ECDEU Assessment Manual for Psychopharmacology. Revised. US Department of Health, Education, and Welfare; 1976:218-222.
- Busner J, Targum SD. The Clinical Global Impressions scale: applying a research tool in clinical practice. Psychiatry (Edgmont). 2007;4(7):28-37.
- Leucht S, Kane JM, Etschel E, Kissling W, Hamann J, Engel RR. Linking the PANSS, BPRS, and CGI: clinical implications. Neuropsychopharmacology. 2006;31(10):2318-2325.