Inaccuracy and Bias in Endoscopic Size Estimation of Large Colorectal Polyps; Insights From a Prospective Cohort of 1828 En-Bloc Resections.

Abstract

Background and aims: Endoscopic size estimation of large colorectal polyps influences treatment decisions and clinical outcomes; however, its precision remains unclear. This study aimed to assess the accuracy of endoscopic size estimation for colorectal lesions ≥ 20 mm utilizing data from an endoscopic submucosal dissection (ESD) cohort.

Methods: This post hoc analysis included only en bloc resected lesions treated by ESD. Patients with neuroendocrine tumors, recurrent lesions, colitis-associated dysplasia, or insufficient data were excluded. Size accuracy was defined as a margin of error < 5 mm. Outcomes included the frequency of size errors ≥ 10 mm and ≥ 20 mm, terminal digit preferences in estimated size, and predictors for lesions estimated endoscopically at 20 mm but pathologically ≥ 25 mm. The reference standard was pathological size.

Results: Among 1889 lesions (1809 patients), 61 lesions (60 patients) were excluded. Finally, 1828 lesions (1749 patients) were evaluated. The accuracy of endoscopic size estimation was 53.4%. Errors ≥ 10 and ≥ 20 mm occurred in 19.1% and 4.5% of lesions, respectively. Endoscopic size estimation showed a strong terminal digit preference for 0 (65.2%) and 5 (30.0%). Among 366 lesions estimated at 20 mm, 97 (26.5%) were pathologically ≥ 25 mm. Polypoid lesions [odds ratio (OR) 2.8, 95% confidence interval (CI) 1.1-6.8] and laterally spreading tumors granular type (OR 2.0, 95% CI: 1.1-3.5) were predictors of underestimation.

Conclusions: Endoscopic size estimation of large colorectal lesions can be inaccurate and influenced by digit bias, underscoring the need for improved measurement techniques (UMIN000010136).