Practice-Based Learning to Predict Polyp Histology at Colonoscopy

A) Study Purpose and Rationale Most polyps removed at colonoscopy are small. The natural history of these polyps is not understood completely, but the risk of subsequent cancer in persons with small rectosigmoid adenomas may not be higher than in persons without rectosigmoid adenomas [1]. With improvements in colonoscopic imaging, experienced endoscopists can detect polyps in a large fraction of patients. Removal of all small polyps followed by formal histopathological examination increases the costs associated with colorectal cancer screening, and may increase the risk of complications, depending on the technique that is used for polypectomy.

New technologies such as narrowband imaging (NBI) offer the possibility of in vivo differentiation between adenomatous and hyperplastic polyps. Policies to leave in place small polyps that appear to be hyperplastic, or to remove and discard small polyps after in vivo histologic categorization without formal histopathology review could significantly decrease the costs and potentially the risks of screening and surveillance colonoscopy.

Multiple studies have demonstrated the ability of experienced endoscopists to achieve high accuracy in differentiating adenomatous from hyperplastic polyps using NBI [2, 3, 4, 5]. The level of confidence associated with in vivo histologic categorization of a particular polyp is a valuable adjunct measure in determining subsequent clinical management. Dissemination to the community setting of policies that promote in vivo histologic categorization is likely to require practice-based learning.

B) Hypotheses The investigators hypothesize that community-based endoscopists can learn to identify polyp histology at colonoscopy with the aid of NBI through the use of an introductory didactic program, followed by practice based-learning, and that representative learning curves can be generated that can serve as guidelines for wider dissemination.

C) Purpose The purpose of this study is to test an educational program combining a didactic program followed by practice-based learning that is designed to allow community-based endoscopists to become proficient at in vivo histologic characterization of small polyps with the aid of NBI. This study will not require any changes in endoscopists' decisions regarding the indications and methods for polypectomy.

This study will not address directly whether polyps predicted to be hyperplastic or even diminutive adenomas should be left in place, or discarded and not submitted for formal histopathological review.

D) Specific Aims This study has two primary and two secondary aims

1. One primary aim is to assess the proficiency of community-based endoscopists at ex vivo histologic characterization of polyps using NBI based on photographs before and after a didactic program designed to familiarize them with in vivo histologic characterization.
2. The central primary aim is to assess whether the currently designed program is effective at training endoscopists to classify adenomatous vs. hyperplastic polyps in practice with at least 90% accuracy.
3. A secondary aim is to characterize endoscopists' individual and group average learning curves for in vivo histologic characterization using NBI during practice-based learning, with attention to level of confidence, accuracy, sensitivity, specificity and positive and negative predictive values.
4. Another secondary aim is to determine whether surveillance recommendations that would be made based on in vivo histologic characterization using NBI are comparable to those based on formal histopathological assessment.

E) Timeline for assessments:

Endoscopists' accuracy will be determined at three pre-specified points: after assessment of 50, 70 and 90 independent diminutive polyps (defined as <=5mm polyps, one per study colonoscopy, with random selection in cases of >1 diminutive polyp per study colonoscopy). We estimate that in order to assess 90 independent diminutive polyps, endoscopists will need to participate for 6-12 months.