ENDOSCOPIC CHARACTERISATION OF SMALL/DIMINUTIVE POLYPS BY GASTROENTEROLOGY TRAINEES: A COMPARISON BETWEEN THE NICE AND SIMPLE CLASSIFICATION

Accurate endoscopic characterisation to predict histology of small/diminutive colonic polyps can be beneficial in resect and discard strategy. Experts are able to differentiate between neoplasia and non-neoplasia with high accuracy when optical diagnosis endoscopic technologies are used. Results have not been consistently demonstrated among non-experts. The NICE (NBI International Colorectal Endoscopic) classification has an excellent multirater kappa coefficient of 0.97 in high confidence assessments and NPV 97.7% amongst experts, and similar results amongst trainees (NPV 95%). The SIMPLE (Simplified Identification Method for Polyp Labelling during Endoscopy) classification has recently been developed and has shown high degrees of accuracy and NPV 91% after training (Figure 1). The aim of this study was to compare the diagnostic accuracy to predict colonic polyp histology using NICE and SIMPLE classification and in multiple endoscopic platforms amongst gastroenterology trainees. 16 trainees (12 gastroenterology trainees and 4 naïve endoscopists) from several institutes across the West Midlands, UK, reviewed 78 videos (48 i-Scan OE and 30 NBI). Participants used NICE and SIMPLE criteria in each of the videos, both NBI and i-Scan OE. Participants completed training on endoscopic characterisation (1 hour) using a computerised power point teaching module followed by a repeat assessment of the same 78 videos but in a randomised order. Accuracy was determined by assessing neoplastic (adenoma, sessile serrated adenoma/polyp) and non-neoplastic (hyperplastic) optical diagnosis with histological diagnosis as reference standard. Interobserver agreement was determined by Kappa statistics. The overall sensitivity, specificity, and accuracy of the NICE classification were 69%, 62% and 66% before training and 77%, 55% and 68% respectively after training The overall sensitivity, specificity, and accuracy of the SIMPLE classification were 73%, 57% and 66% before training and 80%, 51%, and 67% respectively after training. There was no statistically significant difference between NICE and SIMPLE in the post-training performance (Table 1). The inter-observer agreement of polyp diagnosis using SIMPLE improved from pre-training kappa agreement of 0.34 [95% CI 0.29-0.42] to post-training 0.52 [95% CI 0.45-0.61 p<0.0001]. NICE classification also showed an improvement following training although it did not reach significance 0.29 [95% CI 0.23-0.35] vs 0.35 [95% CI 0.28-0.44 p0.064). There was no difference in diagnostic performance between NICE and SIMPLE classifications in the hands of gastroenterology trainees after computerised training module. The user-friendly SIMPLE classification may be used by non-experienced gastroenterologists on multiple endoscopy imaging platforms to differentiate neoplastic from non-neoplastic polyps.Tabled 1Comparison of diagnostic performance by gastroenterology trainees using NICE classification vs. SIMPLE classification following trainingNICESIMPLEP-valueSensitivity77.6 [95% CI 74.3-80.7]80.3 [95% CI 77.1-83.2]0.232Specificity55.4 [95% CI 51.2-59.6]51.4 [95% CI 47.2-55.7]0.206Accuracy67.7 [95% CI 65.0-70.3]67.3 [95% CI 67.3-70.0]0.898High confidence73.2 [95% CI 70.5-75.7]73.4 [95% CI 70.8-75.9]0.889 Open table in a new tab