Assessment and diagnosis of social phobia in the clinic and the community.

I have been asked to comment on the three articles on social phobia in this issue of Psychological Medicine. The three papers in one way or another concern themselves with important issues in the assessment and diagnosis of social phobia (also known as social anxiety disorder; Liebowitz et al. 2000). I do so with the caveat that I am an author on one of the papers and do not claim the same degree of objectivity in examining my own work as I do in examining the work of others. The three papers concern the psychometric evaluation of a modified version of the Clinical Global Impression (CGI) (Guy, 1976) in a sample of persons with social phobia (Zaider et al. 2003), the development of a self-report screening questionnaire for social phobia (Newman et al. 2003) and the characteristics of social phobia in the Australian National Survey of Mental Health and Wellbeing (NSMHWB) (Lampe et al. 2003). Zaider et al. (2003) examine the validity of an adapted version of the CGI Severity and Improvement Scales for the assessment of social phobia and their change over the course of treatment. As many readers are likely aware, the CGI is a ubiquitous measure of outcome in studies of the pharmacotherapy of many disorders. It is included in virtually all industry studies of the efficacy of new medications and is designated as the one of the primary outcome measures in many of these studies. This is certainly the case in studies of social phobia and the other anxiety disorders, despite the almost total lack of studies of its psychometric characteristics. Zaider et al.’s study is of great importance on that basis alone. In this study, additional information was provided to evaluators about the meaning of scale points on the CGI Severity and Improvement Ratings. This is a change from standard operating procedure, but one that we might do well to adopt more broadly. It is likely that this increased specificity will result in greater inter-rater agreement. Although this question was not addressed by Zaider et al. (2003), it is not trivial because a measure can be no more valid than it is reliable. In fact, the modified CGI scales performed quite nicely in this study, giving us a modicum of hope that our previous studies that used this measure may themselves be valid. Scores on the Severity and Improvement scales were highly correlated with each other and normally distributed. They were also significantly (and often substantially) correlated with both self-report and clinician-administered measures of social anxiety symptoms, depression, disability and quality of life. In this study, we also sought to discover the factors that influenced CGI ratings. Thankfully, the patient’s level of social anxiety was an important source of influence. However, the severity rating did not show itself to be a pure measure of the severity of social anxiety symptoms, as significant variance was also accounted for by measures of depression and disability. Thus, the Severity rating seems to provide a summary index of a patient’s impairment, not limited to the symptoms of the specific disorder. It remains for future research to examine whether this is true for disorders other than social phobia or whether this would be the case without the specific rating instructions used in this study. It is interesting that Hope et al. (1997) came to a similar conclusion in a study of ratings of distress and impairment associated with a specific diagnosis that are a part of the Anxiety Disorders Interview Schedule. In addition to social anxiety symptoms, behavioural avoidance and dysphoria accounted for significant variance in these ratings. Interestingly, in the study by Zaider et al. (2003), the CGI Improvement ratings were not influenced by any factors other than change in social anxiety