Outcome heterogeneity, definition variability and trial design in randomised trials of colorectal endoscopic mucosal resection: a systematic review
Daniel M. Simadibrata, Talia F. Malik, Tavia Buysse, Zachary L. SmithObjective
Endoscopic mucosal resection (EMR) for large colorectal polyps is supported by a growing body of randomised evidence, yet the extent of outcome heterogeneity, definition variability and adherence to trial design standards has not been characterised. We assessed outcome selection, definitions, trial design and transparency across all published EMR randomised trials.
Design/method
This systematic review identified EMR randomised controlled trials (RCTs) published 2015–2024 through MEDLINE, Web of Science and Cochrane Library. Two reviewers independently extracted data into a REDCap database on primary outcome categories, definitions for delayed bleeding and adenoma recurrence, trial registration, protocol availability, outcome discrepancies, use of patient-reported outcome measures (PROMs) and stakeholder input. Diversity indices quantified outcome heterogeneity.
Results
46 RCTs were included. Four primary outcome categories were identified (en bloc resection 32.6%, safety 30.4%, efficacy 28.3%, technical success 8.7%) with a Simpson’s Diversity Index of 0.714 and no single published primary outcome exceeding 30.4% of trials. Among 43 trials reporting delayed bleeding, 30.2% did not define criteria. Among those that did, 20 unique definition combinations were identified. Only 52.2% of trials were preregistered, 26.1% had publicly available protocols, and 24.3% changed their primary outcome between registry and publication. Safety outcomes were systematically under-registered (perforation: 8 registrations, 35 publications). PROMs were used in 13.0%, and no trial documented patient input into outcome selection.
Conclusions
EMR RCTs exhibit substantial outcome heterogeneity at every level, from which outcomes are chosen, to how they are defined, to whether they are reported as planned. These findings provide the empirical foundation for developing core outcome sets in this field.