Figure 4.

Predicted mean assessment of severity compared with reported mean assessment of severity. To construct the index after excluding the second assessment of repeat video pairs and the videos with a Contact Friability Test (CFT), each of the 30 investigators evaluated 21 independent videos, leading to 630 evaluations. Each video was scored by 10–12 investigators, except for Mayo Clinic score 0 videos, which were scored by 15 investigators (making up the 630). Twenty-one evaluations with missing data were excluded from the index construction (making 609 evaluations overall). Thus, for each video, evaluations by 10–15 investigators were available, allowing the mean of the evaluations of overall severity to be calculated, as well as the mean of the severity evaluations predicted from the generalised linear mixed model using the three descriptors—vascular pattern, bleeding and erosions and ulcers—according to the levels of these predictors reported by each investigator. VAS, visual analogue scale.