Prevalence and Predictors of Early Termination, Non-Publication, and Primary Endpoint Discrepancy in 2,125 Minimally Invasive Gastrointestinal Surgery Randomized Controlled Trials (2015–2024): A Three-Repository Meta-Epidemiological Study

Background: The prevalence of early termination, non-publication, and primary endpoint discrepancies has not been assessed across multiple subspecialties of minimally invasive gastrointestinal (MIS GI) surgery in a single registry-linked cohort. Aims: To quantify prevalence and predictors of early termination, non-publication, time-to-publication, and primary endpoint discrepancy across MIS GI RCTs (2015–2024). Methods: This meta-epidemiological study examined 2,125 randomised MIS GI trials first submitted to ClinicalTrials.gov between 2015 and 2024. Records were linked to 5,373 candidate PubMed identifiers and 5,153 Crossref digital object identifiers. Trial status, publication outcomes, time to publication, and agreement between registered and reported primary endpoints were examined using multivariable logistic regression, Kaplan–Meier analysis, and a token-based endpoint audit. Results: Among 2,125 RCTs (1,294 [60.9%] laparoscopic, 146 [6.9%] robotic, 685 [32.2%] hybrid/other; 960 in 2015–2019, 1,165 in 2020–2024), 973 (45.8%) were completed and 124 (5.84% overall; 11.30% of 1,097 closed trials) stopped early, mainly from slow accrual (32.3%). In closed RCTs, sample <100 (aOR = 6.52, p < 0.0001) and Western location (aOR = 3.97, p < 0.0001) predicted early termination, while blinding was protective (aOR = 0.51, p = 0.0012). Among 881 mature completed RCTs (≥24 months post-completion), 531 (60.3%) were unpublished; industry funding (aOR = 2.34, p = 0.0081) and sample <100 (aOR = 1.62, p = 0.0011) increased non-publication, while multicentre design (aOR = 0.54, p = 0.0039), Western location (aOR = 0.57, p = 0.0004), and robotic modality (aOR = 0.50, p = 0.0490) reduced non-publication. Median time-to-publication was 26.9 months for multicentre and 28.0 months for robotic RCTs. Across 584 published RCTs with structured abstracts, 364 (62.3%) had endpoint discrepancies (45.5% timeframe modification, 13.7% outcome switching, 3.1% selective multi-primary reporting). Conclusion: Substantial accrual failure, non-publication, and endpoint modification affect MIS GI RCTs, requiring multicentre networks and registry-to-manuscript concordance audits.