Clinical Global Impression (CGI)

The CGI is a 3-item, clinician-rated scale used to assess global illness severity, overall improvement from the start of treatment, and therapeutic response. The scale was designed to be universal and can be administered to a wide variety of patient populations.
Description: The CGI is a 3-item, clinician-rated scale used to assess global illness severity, overall improvement from the start of treatment, and therapeutic response. The scale was designed to be universal and can be administered to a wide variety of patient populations.
Disease States: General
Validated Uses: Treatment Monitoring & Evaluation, Symptom Severity
Administration Method: Clinician-report
Time to administer: Less than 5 minutes
Commonly used in: Clinical Trials & Research, Clinical Practice
Detailed Description: The CGI was originally developed as a brief, standalone tool for evaluating a patient’s global functioning before and after initiation of a new medication in a clinical trial.1 Today, the CGI is also used in clinical practice1 and consists of 3 items that assess global illness severity, overall improvement from the start of treatment, and therapeutic response.2,6 These items are commonly referred to as subscales (“CGI-S,” “CGI-I,” and “efficacy index” or “CGI-E,” respectively), and can be used individually when appropriate.2,6 All ratings are made with respect to the past week of the patient’s life1 and depend, in part, on the clinician’s experience with patients of the same diagnosis.2,5 Because each item is rated independently, there is no global score for the CGI.3 Instead, each score should be recorded and tracked at every visit.1 While changes in the CGI-S score are thought to be understood intuitively by clinicians,7,8 the CGI-I score is often treated as a categorical measure, with scores ≤2 used to classify patients as “treatment responders”.8 See table below for more information about each subscale and its intended use.
Item/Subscale Description How the Measure Is Used
CGI-S (“severity scale”)
  • 1–7 scale according to severitya2
  • Possible ratings range from “not at all ill” (ie, score of 1) to “among the most extremely ill” (ie, score of 7)2,3
Compare with previous severity score1
CGI-I (“improvement scale”)
  • 1–7 scale, according to the degree of improvement from baselinea2
  • Possible ratings range from “very much improved” (ie, score of 1) to “very much worse” (ie, score of 7)2,3
Compare with the initial assessment of the patient1
CGI-E (“efficacy index”)
  • 0–4 scale, according to the degree of therapeutic response3
  • Final index score depends on separate ratings of therapeutic benefits versus side effects2,3
  • This subscale is not commonly used today3
Compare with the previous efficacy score1
aScores of 0 can be given if no assessment is made.
Scale Validity: Properties of the CGI have been examined in different populations. The first study of the scale’s reliability was done in patients with dementia. Duplicate ratings of 12 adult patients with dementia found that inter-rater reliability of the CGI was adequate, with the CGI-S showing an inter-rater reliability of 0.66 and the CGI-I showing an inter-rater reliability of 0.51.9 Inter-rater reliability of the CGI-S was similar in a later study of 30 adult patients with major depressive disorder, ranging from 0.64 to 0.70.10
Alternative Versions: While the CGI was designed to be universal, alternative versions that are specific to different mental health disorders have also been developed.4 Versions exist for use in patients with bipolar disorder, major depressive disorder, schizophrenia, anxiety disorders, and more.10-13 These disorder-specific versions tend to have improved instructions for clinicians along with more detailed behavioral anchors (behavioral anchors are descriptive examples of symptoms or behaviors that would qualify a patient for a given rating score).10-14 A self-report version of the CGI called the Patient Global Impressions scale has also been developed.4,15
Cited Limitations: A frequently mentioned limitation of the CGI is that it lacks well-defined behavioral anchors and asks the rater to compare the current patient with other patients who have the same diagnosis.5 As a result, scoring depends on clinical experience and best judgment, and can result in poor inter-rater reliability.5,16 Additional criticisms of the CGI include that it lacks extensive psychometric validation, includes redundant information, and is overly reliant on a clinician’s memory.4,6,7,16


†Although the original CGI is provided here, we include information about alternative versions for your general awareness and convenience. These alternative versions may be used by different practitioners or in clinical trial or research settings.

Supporting references for the filters are as follows:
Disease States: General1,2; Validated Uses: Treatment Monitoring & Evaluation1,2, Symptom Severity1,2; Populations: Adult2,3; Administration Method: Clinician-report1,3; Time to Administer: < 5 minutes1,4; Commonly used in: Clinical Trials & Research1,4, Clinical Practice1,5


  1. Busner, J & Targum, SD. Psychiatry (Edgmont). 2007;4(7):28-37.
  2. Guy, W. ECDEU Assessment Manual for Psychopharmacology. 1976. Education U.S. Department of Health, and Welfare. Rockville, MD.
  3. Lam, R et al. Assessment Scales in Depression, Mania and Anxiety.  (Taylor & Francis Group, 2005).
  4. Rush, AJ et al. Handbook of Psychiatric Measures.  (American Psychiatric Publishing, Inc., 2008).
  5. Dunlop, BW et al. Behav Sci (Basel). 2017;7(3).
  6. Forkmann, T et al. BMC Psychiatry. 2011;11:83.
  7. Leucht, S et al. Br J Psychiatry. 2005;187:366-371.
  8. Spielmans, GI & McFall, JP. J Nerv Ment Dis. 2006;194(11):845-852.
  9. Dahlke, F et al. Psychopharmacol Bull. 1992;28(4):425-432.
  10. Kadouri, A et al. BMC Psychiatry. 2007;7:7.
  1. Spearing, MK et al. Psychiatry Res. 1997;73(3):159-171.
  2. Haro, JM et al. Acta Psychiatr Scand Suppl. 2003(416):16-23.Za
  3. Zaider, TI et al. Psychol Med. 2003;33(4):611-622.
  4. Goldstein, G et al. Handbook of Psychological Assessment. 4th edn (Elsevier, 2020).
  5. Mohebbi, M et al. Eur Psychiatry. 2018;53:17-22.
  6. Beneke, M & Rasmus, W. Pharmacopsychiatry. 1992;25(4):171-176.

Welcome To NP Psych Navigator

This website is intended for healthcare professionals inside the United States.
Please confirm that you are a healthcare professional inside the US.

No, I am not US Healthcare professional

You are now leaving NP Psych Navigator

Links to sites outside of NP Psych Navigator are provided as a resource to the viewer. AbbVie Inc accepts no responsibility for the content of non-AbbVie linked sites.

Redirect to:

Please to bookmark content