Development of a definition and rules for causal attribution of post-colonoscopy bleeding.

BACKGROUND: Post-colonoscopy bleeding (PCB) is an important colonoscopy quality indicator that is recommended to be routinely collected by colorectal cancer screening programs and endoscopy quality improvement programs. We created a standardized and reliable definition of PCB and set of rules for attributing the relatedness of PCB to a colonoscopy. METHODS: PCB events were identified from colonoscopies performed at the Forzani & MacPhail Colon Cancer Screening Centre. Existing definitions and relatedness rules for PCB were reviewed by the authors and a draft definition and set of rules was created. The definition and rules were revised after initial testing was performed using a set of 15 bleeding events. Information available for each event included the original endoscopy report and data abstracted from the emergency or inpatient record by a trained research assistant. A validation set of 32 bleeding events were then reviewed to assess their interrater reliability by having three endoscopists and one research assistant complete independent reviews and three endoscopists complete a consensus review. The Kappa statistic was used to measure interrater reliability. RESULTS: The panel classified 28 of 32 events as meeting the definition of PCB and rated 7, 8 and 6 events as definitely, probably and possibly related to the colonoscopy, respectively. The Kappa for the definition of PCB for the three independent reviews was 0.82 (substantial agreement). The Kappa for the attribution of the PCB to the colonoscopy by the three endosocopists was 0.74 (substantial agreement). The research assistant had a high agreement with the panel for both the definition (100% agreement) and application of the causal criteria (kappa 0.95). CONCLUSIONS: A standardized definition of PCB and attribution rules achieved high interrater reliability by endoscopists and a non-endoscopist and provides a template of required data for event adjudication by screening and quality improvement programs.

Prophylactic Endoscopic Clipping Does Not Prevent Delayed Postpolypectomy Bleeding in Routine Clinical Practice: A Propensity Score-Matched Cohort Study.

INTRODUCTION: Delayed postpolypectomy bleeding (DPPB) is a relatively common adverse event. Evidence is conflicting on the efficacy of prophylactic clipping to prevent DPPB, and real-world effectiveness data are lacking. We aimed to determine the effectiveness of prophylactic clipping in preventing DPPB in a large screening-related cohort. METHODS: We manually reviewed records of patients who underwent polypectomy from 2008 to 2014 at a screening facility. Endoscopist-, patient- and polyp-related data were collected. The primary outcome was DPPB within 30 days. All unplanned healthcare visits were reviewed; DPPB cases were adjudicated by committee using a criterion-based lexicon. Multivariable logistic regression was performed, yielding adjusted odds ratios (AORs) for the association between clipping and DPPB. Secondary analyses were performed on procedures where one polyp was removed, in addition to propensity score-matched and subgroup analyses. RESULTS: In total, 8,366 colonoscopies involving polypectomy were analyzed, yielding 95 DPPB events. Prophylactic clipping was not associated with reduced DPPB (AOR 1.27; 0.83-1.96). These findings were similar in the single-polyp cohort (n = 3,369, AOR 1.07; 0.50-2.31). In patients with one proximal polyp ≥20 mm removed, there was a nonsignificant AOR with clipping of 0.55 (0.10-2.66). Clipping was not associated with a protective benefit in the propensity score-matched or other subgroup analyses. DISCUSSION: In this large cohort study, prophylactic clipping was not associated with lower DPPB rates. Endoscopists should not routinely use prophylactic clipping in most patients. Additional effectiveness and cost-effectiveness studies are required in patients with proximal lesions ≥20 mm, in whom there may be a role for prophylactic clipping.