The Clinical Global Impression scale (CGI) is frequently used in clinical research because of its face validity and ease of use but data on its reliability are scarce. Our goal was to estimate the reliability of the scale and compare reliability between face-to-face and video scoring. We analyzed 50 different video interviews recorded during 5 visits of a crossover trial to study the effect of subthalamic nucleus stimulation. Six specialized clinicians rated the CGI using these videos, providing 300 different ratings. The intraclass correlation was lower at inclusion (0.30 [0.13-0.50]) than at later visits (0.68 [0.61-0.80]). Reliability was not influenced by the patients' stimulation status. The mean of at least two independent evaluations of the video is needed to achieve an ICC greater than 0.8. The video CGI is a valid clinical outcome measure suitable for clinical trials (ClinicalTrials.gov number, NCT00169377).
Copyright © 2010 Elsevier Ltd. All rights reserved.