We aimed to study if performance can be maintained. We studied the learning curve in five colonoscopists of varied experience during a prospective randomized trial on the optical diagnosis of colorectal polyps using NBI. They performed optical diagnosis based on a random assignment using either close view (CFHQ190) or standard view
(CFH180) colonoscopy. For each polyp, Romidepsin order endoscopists stated the diagnosis (neoplastic or non-neoplastic) and confidence in the diagnosis (low or high) based on validated polyp differentiation criteria. Prior to study enrollment, the endoscopists completed a computerized learning module that required a minimum accuracy of 90%, and then performed 10 colonoscopies with real-time assessment of polyp histology. Midway through the study, they completed a refresher course. We assessed learning by dividing the number of polyps diagnosed by each endoscopist into halves, and measured NPV and accuracy for each half. We used the Cochrane-Mantel-Haenszel statistic to assess for significance. Endoscopists showed overall high diagnostic performance throughout, with a non-significant trend toward higher find more NPV and accuracy in the second half, (Figure 1). In the close view arm of 530 polyps, endoscopists
had NPVs of 94.9% (95% CI: 87.5-98.6) in the first half and 96.7% (95% CI: 88.5-99.6%) in the second half, p=0.974. Three endoscopists in the first half and 4 in the second achieved > 90% NPV. Accuracy was 87.7% in the first half (95% CI: 82.7-91.7) and 90.0% in the second (95% CI: 85.3-93.7), p=0.526; 2 endoscopists in the first half and 3 endoscopists in the second achieved >90% accuracy. Overall, in the standard view arm of 445 polyps, negative predictive value was
88.0% (95% CI: 75.7-95.5) in the first half and 95.8% (88.3-99.1%) in the second, for optical diagnoses made with high confidence, p=0.714. Three of five endoscopists in the first half and four in the second achieved >90% NPV. Accuracy was 86.2% (95% CI: 79.8-91.1) in the first half and 87.8 (95% CI: 81.8-92.3%) in the second among all endoscopists, p=0.436; one endoscopist achieved > 90% accuracy in the second half. High negative predictive value for the prediction of non-neoplasms with NBI that met PIVI thresholds was achieved and maintained in this group of endoscopists Pyruvate dehydrogenase lipoamide kinase isozyme 1 who participated in standardized and continued training. Both NPV and accuracy showed continuing high performance of optical diagnosis of colorectal polyps. Negative predictive value for the first and second half of polyps assessed by each endoscopist and overall, using both colonoscope with close view (CFHQ190, L) and standard colonoscope (CFHQ180, R). “
“A paradigm shift of a “diagnose, resect and discard” strategy for diminutive (≤ 5 mm) colorectal polyps has been proposed. ASGE has established thresholds for this strategy in the recently published PIVI document.