Effect of prophylactic endoscopic clip placement on clinically significant post-endoscopic mucosal resection bleeding in the right colon: a single-centre, randomised controlled trial

Endoscopic mucosal resection (EMR) is a cornerstone in the management of large (≥20 mm) non-pedunculated colorectal polyps. Clinically significant post-EMR bleeding occurs in 7% of cases and is most frequently encountered in the right colon. We aimed to assess the use of prophylactic clip closure in preventing clinically significant post-EMR bleeding within the right colon.We conducted a randomised controlled trial at a tertiary centre in Australia. Patients referred for the EMR of large non-pedunculated colorectal polyps in the right colon were eligible. Patients were randomly assigned (1:1) into the clip or control (no clip) group, using a computerised random-number generator. The primary endpoint was clinically significant post-EMR bleeding, defined as haematochezia necessitating emergency department presentation, hospitalisation, or re-intervention within 14 days post-EMR, which was analysed on the basis of intention-to-treat principles. The trial is registered with ClinicalTrials.gov, NCT02196649, and has been completed.Between Feb 4, 2016, and Dec 15, 2020, 231 patients were randomly assigned: 118 to the clip group and 113 to the control group. In the intention-to-treat analysis, clinically significant post-EMR bleeding was less frequent in the clip group than in the control group (four [3·4%] of 118 patients vs 12 [10·6%] of 113; p=0·031; absolute risk reduction 7·2% [95% CI 0·7-13·8]; number needed to treat 13·9). There were no differences between groups in adverse events, including delayed perforation (one [1%] in the clip group vs one [1%] in the control group) and post-EMR pain (four [3%] vs six [5%]). No deaths were reported.Prophylactic clip closure can be performed following the EMR of large non-pedunculated colorectal polyps of 20 mm or larger in the right colon to reduce the risk of clinically significant post-EMR bleeding.None.