Approaches to denote treatment outcome: Clinical significance and clinical global impression compared

• Published in: International journal of methods in psychiatric research Vol. 28; no. 4; pp. e1797 - n/a
• Main Authors: Beurs, Edwin, Carlier, Ingrid V.E., Hemert, Albert M.
• Format: Journal Article
• Language: English
• Published: United States John Wiley & Sons, Inc 01.12.2019; John Wiley and Sons Inc
• Subjects: Adult; clinical global impression; Clinical significance; Discordance; Female; Humans; Male; Mental disorders; Mental Disorders - diagnosis; Mental Disorders - therapy; Middle Aged; Original; Outcome Assessment, Health Care - standards; Psychiatric Status Rating Scales - standards; Psychometrics - standards; Reliable Change Index; treatment outcome; clinical global impression; Reliable Change Index; clinical significance; treatment outcome
• ISSN: 1049-8931; 1557-0657
• DOI: 10.1002/mpr.1797

Objectives The authors of a previous study proposed a statistically based approach to denote treatment outcome, translating pretest and posttest scores into clinically relevant categories, such as recovery and reliable improvement. We assessed the convergent validity of the Jacobson–Truax (JT) approach, using T‐score based cutoff values, with ratings by an independent evaluator. Methods Pretest and retest scores on the Brief Symptom Inventory (BSI) and clinical global impression improvement (CGI‐I) ratings were collected repeatedly through routine outcome monitoring from 5,900 outpatients with common mental disorders. Data were collected in everyday practice in a large mental health care provider. Results Continuous pretest‐to‐retest BSI change scores had a stronger association with CGI‐I than the categorical variable based on JT. However, JT categorization and improvement according to CGI converged substantially with association indices (Somers' D) ranging from D = .50 to .56. Discordance was predominantly due to a more positive outcome according to JT than on CGI‐I ratings. Conclusion Converting continuous outcome variables into clinically meaningful categories comes at the price of somewhat diminished concurrent validity with CGI‐I. Nevertheless, support was found for the proposed threshold values for reliable change and recovery, and the outcome denoted in these terms corresponded with CGI improvement for most patients.