Does Size Really Matter?
The first question is whether we should accept that inaccurate measurement is a real phenomenon. Sakata et al1 use density distribution curves to show that endoscopist observations favor certain sizes over others. Some might dismiss their claims because they do not compare observed with actual measured polyp size. However, their findings make intuitive sense, and this statistical approach adds to an already strong direct- measurement literature implicating polyp size assessment as inaccurate. Studies have shown errors in as many as 87% of cases,4 poor relative performance of both simple optical estimation and use of open biopsy forceps as a guide,5 and variation in measurement of the same simulated polyp by the same endoscopist over time.6 Indeed, the more published studies one reviews, the more difficult it becomes to make the case that polyp measurement, as currently practiced by almost all endoscopists, is acceptably accurate.
Having established that inaccuracy likely is real, the next question becomes whether the impact of defective measurement is important. The key inflection point in polyp size is 10 mm. Many studies have suggested that, above this size, polyps become markers of significant increased lifetime colorectal cancer risk, whereas below this size, the risk is increased minimally or not at all.7,8 Surveillance guidelines reflect this and recommend colonoscopy at 3-year intervals for patients with ≥10 mm polyps and 5- to 10-year intervals for sizes below this.3 Based on current evidence, inaccuracy at 5 mm, 20 mm, or at any size other than 10 mm does not importantly impact risk assessment or treatment. Thus, to determine the impact of measurement error, we only need focus on how often subcentimeter polyps are estimated at ≥ 10 mm and how often ≥10 mm polyps are estimated to be ≤9 mm. Schoen et al4 looked at inaccuracy around the 10-mm threshold and found that 8.9% of polyp size estimates resulted in misclassification of high-risk patients as low risk or vice versa. Gopalswamy et al5 did the same and found that 13.0% were misclassified. If we assume an adenoma detection rate of 40.0%, then 3.6% to 5.2% of the colonoscopies in these studies would have had an overestimate or underestimate that would alter risk determination and patient management. In reality, the percentage whose care might be affected is even lower. At index colonoscopy, ≈35% of patients with polyps will have advanced histology, and 20% will have 3 or more adenomas.8,9 With either of these findings, polyp size becomes irrelevant. The bottom line is that probably only 2% to 3% of patients undergoing colonoscopy are affected by polyp measurement inaccuracy.
Although the small percentages may be reassuring, the absolute number affected by mismeasurement still demands our attention.