Measurement Bias of Polyp Size at Colonoscopy: What Is the Big Deal?

Abstract

International post-polypectomy surveillance guidelines and optical diagnoses depend on accurate determination of polyp size at colonoscopy. Study 1 was a retrospective, population-based study of all colonoscopies performed within the state-funded health system in Brisbane over a 12-month-period: 12,597 electronic colonoscopy reports were individually analyzed to obtain the frequencies of polyp measurements. Study 2 introduced a novel, proof-of-concept measurement device to reduce measurement bias. Fifty endoscopists used 3 measurement strategies (M1: no visual cues; M2: a snare tip; M3: a novel 1x1 mm grid) to evaluate 120 endoscopic images of 40 simulated polyps (1 to 10 mm) presented in random order. In study 1, 65 endoscopists individually sized 8,591 from 18,276 detected polyps (47%). Most polyps were 5 mm in size, and endoscopists were likely biased toward measurements in 5-mm increments. In study 2, the exact accuracy of polyp measurement was significantly different between methods: 20%, 33%, and 90% (p < 0.001). Endoscopists had limited insight to their accuracy because >70% of measurements were made with high confidence. Using M3, endoscopists were 99.8% accurate for high-confidence classifications of polyps into clinically relevant size categories. Endoscopists are inaccurate in measuring polyps at colonoscopy. Improvements in measurement accuracy are urgently needed.