Introduction Delayed bleeding is the most common severe adverse event after p-Endoscopic Mucosal Resection (p-EMR) but factors that may predict delayed bleeding are not well-defined.

Methods Data were analysed from a prospective single centre observational cohort study of patients with complex NPCPs ≥20 mm who underwent p-EMR between January 2010 and August 2012. Patient, polyp, and procedure-related data were collected. Four post p-EMR resection defect factors were evaluated for interobserver agreement and included in risk analysis. A telephone interview was conducted with patients 14 days post p-EMR. Delayed bleeding severity was reported in accordance with guidelines. Predictors of bleeding were identified by univariable and multivariable analyses.

Results Delayed bleeding requiring hospitalisation occurred in 22 of 330 (6.7%) patients. Eleven patients required blood transfusion; of these 4 underwent urgent colonoscopy, 1 underwent radiological embolisation and 1 required surgery to achieve haemostasis. Interobserver agreement for identification of the 4 post p-EMR resection defect factors was moderate (kappa >0.5 for each). Predictors of delayed bleeding on multivariable analysis were a previous EMR attempt (odds ratio = 3.13, P = 0.05) and visible muscle fibres in the post p-EMR resection defect (OR 3.64, P = 0.03). Factors not predictive on multivariable analysis included patient age, ASA class, aspirin use, polyp site, polyp size or APC use.

Conclusion Visible muscle fibres in the p-EMR resection defect and a previous failed p-EMR attempt are predictors of delayed bleeding after p-EMR. These findings emphasise the importance of technical factors in ensuring good outcomes after p-EMR for complex NPCPs.

Disclosure of Interest None Declared