Introduction Variceal size is a major predictor of the risk of bleeding in patients with liver cirrhosis. Guidelines from the American Association for the Study of the Liver (AASLD) recommend the use of a simplified system (small vs large) for grading of oesophageal varices (OV), but data on the reliability of this system compared to the conventional 3- grade classification remains lacking. We aimed to assess inter-rater reliability of these two widely used grading systems for OV.
Method High-resolution endoscopy recordings of 108 patients (n=8 training cohort and n=100 evaluation cohort) with chronic liver disease were prospectively collected using standardised criteria. Nine Gastroenterologists of variable experience performed independent evaluations of the videos in a random order. The cases were scored according to the presence or absence of OV. If OV were present they were scored using a 2-grade system (small and large) as well as a 3-grade system (small, medium and large). Overall agreement was analysed using intraclass correlation coefficient or kappa statistic as required.
Results Agreement on the presence or absence of varices was good (k=0.6, p<0.001). Agreement between observers using the 2-grade (0.74, 95% Confidence interval (CI) 0.68–0.80) was very similar to that using the 3-grade scoring system (0.76, 95% CI 0.70–0.81). The agreement between experts (2-grade: k=0.78, 95% CI 0.71–0.84 and 3-grade: k=0.79, 95% CI 0.72–0.84) was slightly higher than that between trainees (2-grade: k=0.7, 95% CI 0.61–0.77 and 3- grade: k=0.72, 95% CI 0.637–0.792). The agreement between observers and the reference endoscopist was slightly higher for the 2-grade system (k=0.55, p<0.001) in comparison to the 3-grade system (k=0.47, p<0.001).
Conclusion Observers had a good to substantial level of agreement using 2 different scoring systems of OV. We conclude that the simpler 2-grade system recommended by AASLD appears to be equivalent to the 3-grade system in terms of inter-observer agreement. This is the first study to support its use in clinical practice.
Disclosure of Interest None Declared
- Inter-rater reliability