CGI: Clinical Global Impression

The CGI (Clinical Global Impression) is a clinician-rated scale of psychiatric illness severity and treatment response, made of two single-item 1–7 ratings. The CGI scales are simple measures widely used as outcomes in psychiatric clinical trials ^[4].

What does the CGI measure?

The CGI was published by William Guy in 1976 in the ECDEU Assessment Manual for Psychopharmacology, from the US National Institute of Mental Health (NIMH) ^[1]. It gives a brief, standardized global measure that can be used across clinical conditions and treatment settings.

The CGI consists of two primary independent ratings:

1. CGI-S (Severity of Illness): A 1–7 rating of the client's current condition relative to the clinician's total experience with that population.
2. CGI-I (Global Improvement): A 1–7 rating of the client's change from baseline condition.

Unlike self-report measures such as the PHQ-9 or GAD-7, the CGI asks the clinician to make a global judgment based on their overall clinical experience with people presenting with similar conditions. In a survey of clinical trial investigators, symptom severity was the most important element in their CGI-S scores and functional status the second ^[4].

The ECDEU manual is a US government publication, and the CGI is in the public domain. Its brevity, simplicity and independence from any single condition have kept it in use in clinical trials for nearly five decades, including in generalized anxiety disorder. It is also practical in routine care, where a clinician can rate it quickly in a busy practice ^[2].

How do you give the CGI?

• Rater: Must be completed by a clinician; not for self-report.
• Time: Brief: two single ratings.
• Baseline: CGI-I requires a prior CGI-S baseline rating to be validly completed.
• Global Judgment: Scores are based on overall clinical impression, not a sum of symptoms.

Clinician-rated tool. Rate current illness severity (CGI-S) and, if a baseline is available, global improvement (CGI-I).

Severity of Illness

Considering your total clinical experience with this particular population, how mentally ill is the patient at this time?

Global Improvement

Compared to the patient's condition at baseline, how much has the patient changed?

Requires a baseline CGI-S rating for valid interpretation.

Clinician reference tool only. Cannot replace individualized clinical judgment.

How do you score and interpret the CGI?

The anchors below are the definitions in the NIMH ECDEU Assessment Manual ^[1]. CGI-S and CGI-I are independent ratings.

The CGI-S can stand alone; the CGI-I always requires a prior CGI-S baseline.

CGI-S (Severity):

• 1: Normal, not at all ill
• 2: Borderline mentally ill
• 3: Mildly ill
• 4: Moderately ill
• 5: Markedly ill
• 6: Severely ill
• 7: Among the most extremely ill

CGI-I (Improvement):

• 1: Very much improved
• 2: Much improved
• 3: Minimally improved
• 4: No change
• 5: Minimally worse
• 6: Much worse
• 7: Very much worse

Clinical benchmarks: A CGI-I score of 1 or 2 is commonly used to define a "responder" in clinical trials.

CGI-S: Severity of Illness

The clinician rates the client's current illness severity relative to their total experience with people who have the same clinical condition. The rating is a single global judgment on the 1–7 scale, not a sum of symptoms.

Can be completed at intake or any visit without a prior score. The anchors run from 1 (normal, not at all ill) through 3–4 (mildly to moderately ill) to 5–7 (markedly ill to among the most extremely ill).

CGI-I: Global Improvement

The clinician rates total improvement since the baseline assessment, regardless of whether it is due to drug treatment. A score of 1 ("Very much improved") or 2 ("Much improved") is typically considered a positive clinical trial response. Scores of 4 indicate no meaningful change; 5–7 indicate deterioration.

Always requires a prior CGI-S baseline rating to be interpretable. In clinical trials, CGI-I is often co-primary or key secondary endpoint alongside disorder-specific scales.

Efficacy index (optional third component)

The original CGI also included an optional Efficacy Index, a ratio of therapeutic effect to side-effect burden, scored on a 4x4 grid. This component is rarely used today because it conflates efficacy and tolerability, which most modern trial designs treat as separate endpoints. The CGI-S and CGI-I remain the standard.

When clinicians or protocols refer to "the CGI," they almost always mean CGI-S and CGI-I only. The Efficacy Index is largely a historical artifact.

Psychometric properties

• Reliability: Inter-rater agreement depends on rater training and anchor definitions. In a depression study, averaging ratings from 4 independent raters gave intraclass correlation coefficients of about 0.9 ^[3].
• Validity: Psychometric evaluations have reported good internal consistency and concurrent validity, alongside problems with validity, scaling and test-retest reliability in some disease populations ^[4].
• Sensitivity to change: The CGI scales have been used as the primary outcome in clinical trials for a variety of conditions, including major depression, social phobia, post-traumatic stress disorder and panic disorder ^[4].

Limitations

• Reliant on the clinician's subjective experience and "internal norms" for that population.
• The original scales lack well-defined anchor points, a source of rater variability ^[4].
• Not a substitute for detailed, disorder-specific symptom assessment.
• Not a diagnostic tool: it rates severity and change, not the presence of a specific condition.