4 for one video in the group of Mayo Clinic stratum 1 to 2, to 1.2 to 9.6 for videos in the normal stratum, to 93.4 for a video of the most severe stratum of UC, indicating that the 57 videos embraced the full range of endoscopic UC severity seen in clinical trials and practice (Figure 1). Responses also indicate that the full range of severity was assessed for each descriptor and on the VAS (Table 3). The correlation of the simple sum version of the UCEIS with evaluation of overall severity on the VAS had a median of 0.93 across investigators (minimum, 0.78; maximum, 0.99), indicating that on

average the UCEIS captured 86% (derived from 0.932) of the variance in investigators’ CDK and cancer assessments of overall severity. There was also a high level of correlation between the 3 individual descriptors and assessment of overall severity on the VAS: with a median of 0.82 (minimum, 0.55; maximum, 0.90) for vascular pattern, 0.80 (minimum, 0.45; maximum, 0.97) for bleeding, and 0.89 (minimum, 0.78; maximum, 0.96) for erosions and ulcers. The Cronbach coefficient α was 0.863

for the UCEIS overall (vascular pattern, AUY-922 supplier 0.83; bleeding, 0.80; erosions and ulcers: 0.79). One-at-a-time deletion of descriptors resulted in slightly lower α coefficients (0.79–0.83), indicating that each descriptor contributed positively to the overall UCEIS. A total of 50 repeat-pair assessments assessed intraobserver variability. The intrainvestigator reliability ratio for evaluation of overall severity was 0.87 on the VAS and 0.96 for the UCEIS. Intrainvestigator agreement for descriptors ranged from a κ of 0.47 (95% confidence interval [CI], 0.27–0.67) for bleeding to 0.87 (95% CI, 0.74–1.00) for vascular pattern (Table 4), indicating moderate to very good agreement for individual descriptors. The weighted 3-mercaptopyruvate sulfurtransferase intraobserver κ for the overall UCEIS score was 0.72 (95% CI, 0.61–0.82). A total of 548 video evaluations of 57 videos (22 per investigator, 2 missing; Table 2) assessed interobserver variability. The interinvestigator reliability ratio for overall assessment of severity was 0.78 on the VAS and 0.88 for the UCEIS. Interinvestigator agreement for descriptors ranged

from a κ of 0.48 (95% CI, 0.46–0.50) for bleeding to 0.54 (95% CI, 0.50–0.57) for vascular pattern, indicating moderate agreement for individual descriptors between investigators (Table 4). The weighted interobserver κ for the overall UCEIS score was 0.50 (95% CI, 0.49–0.52). In summary, only 4% of the variation in UCEIS scoring in the repeat evaluation data set was attributable to within-investigator variation when scoring the same video twice. Similarly, only 12% of the variation in UCEIS scoring in the main analysis data set was attributable to investigator-to-investigator differences when scoring a common video. Across investigators, the correlation between the normalized version of the UCEIS and overall severity (VAS) had a median value of 0.94 (minimum, 0.78; maximum, 0.