Combined response to letters to the editors

We thank Gupta et al 1 and Khalaf et al 2 for their thoughtful and detailed comments on our work and for the opportunity to further clarify its methodological and clinical implications. Both letters rightly emphasise that our registry represents a referral-enriched cohort of large non-pedunculated colorectal polyps managed in expert endoscopic submucosal dissection (ESD) centres. This was clearly acknowledged in our manuscript and is intrinsic to the clinical question we aimed to address. Our objective was not to estimate population-level submucosal invasive cancer (SMIC) prevalence, but to assess whether rectal location independently predicts SMIC among advanced lesions selected for endoscopic resection after accounting for established morphological and optical high-risk features.3 Regarding the apparent discrepancies among propensity score matching (PSM), inverse probability weighting (IPW) and multivariable regression, it is essential to recognise that these methods address different estimands. PSM was deliberately chosen to answer a lesion-level, clinically pragmatic question: when lesions …