Clinical global impression
The clinical global impression rating scales are measures of symptom severity, treatment response and the efficacy of treatments in treatment studies of patients with mental disorders. It is a brief 3-item observer-rated scale that can be used in clinical practice as well as in researches to track symptom changes. It was developed by Early Clinical Drug Evaluation Program Severity of Illness Global Improvement Efficacy Index. Many researchers, while recognizing the validity of the scale, consider it to be subjective as it requires the user of the scale to compare the subjects to typical patients in the clinician experience.
Validity of CGI scales
Despite its wider acceptance and consistent use in clinical trials, psychometric properties of CGI have not been established. However, the scale is found to be valid as an external criterion during the development of scales of depression and anxiety. In many studies, the clinicians' ratings of psychiatric symptoms were found to correlate significantly with self-rated and other valid scales of symptom severity. For example, Leon et al. showed that severity ratings but not improvement ratings were predicted highly and significantly by frequency of panic attacks, depression and anxiety ratings made on already valid rating scales. Similarly, another study showed that MADRS, HAM-D and CGI scales had comparable effect sizes and could be equally used in studies to assess severity and improvement of symptoms.Even though many studies have established the validity of CGI scales in relation to other commonly used robust rating scales, its efficacy in predicting treatment outcomes is highly debated. Its sensitivity is good enough to differentiate between responders and non-responders in clinical trials of depression, but its specificity is not satisfactory. It has poorer interrater reliability than HAM-D.
Many weaknesses could explain this possible lack of validity of the CGI: there is no specific interviewer guide available, and while most other symptoms scales have fairly clear and specific response options, the response format used in the CGI to assess change or severity of illness is more likely to be ambiguous.