Accuracy and reliability of the endoscopic classification of chronic radiation-induced proctopathy using a novel...

To assess a novel grading method of radiation proctitis for intraobserver and interobserver agreement among endoscopists.

There are no established criteria for the endoscopic classification of chronic radiation-induced proctopathy. We introduce a classification system based on telangiectasia density and vascular coalescence. Accuracy and reproducibility of this system were examined.

A total of 131 endoscopic images of the rectum in 74 consecutive patients undergoing lower endoscopy who had received pelvic radiation therapy were analyzed. Each image was duplicated, reversed, and rotated 90 degrees for a total of 262 images. These were shown in random order to 13 endoscopist evaluators (6 attending physicians, 7 gastroenterology fellows) using an online computer testing program. Each image was scored from grade 0 to 3 using criteria from the rectal telangiectasia density (RTD) classification we developed. Kappa (kappa) statistics and percent agreement were used to quantify the reproducibility and level of agreement.

Intraobserver agreement: The mean (SD) for kappa among the 13 raters was 0.58 (0.09); 95% confidence interval [CI] = 0.527-0.636. Interobserver agreement: The estimated kappa across all 13 raters was 0.518 (95% CI = 0.506-0.530). For the 7 trainees, kappa was 0.547 (95% CI = 0.523-0.571). For the 6 attending physicians, the kappa was 0.481 (95% CI = 0.453-0.509). As another indicator of agreement, all 13 evaluators agreed on 30 (22.9%) of images, differed by no more than 1 grade on 60 (45.8%) images, no more than 2 grades on 33 (25.2%) of images, and no more than 3 grades on 8 of the images (6.1%); 73% of patients referred for bleeding control were RTD grade 2 or 3.

The RTD grading scale for radiation proctopathy is reproducible among endoscopists. Hematochezia is associated with high RTD grade.