BACKGROUND: Endoscopic resection (ER) offers organ-preserving, potentially curative treatment for early (T1 N0) oesophagogastric (OG) adenocarcinoma, yet the extent of interhospital variation in practice is unclear. This study assessed variations in clinical practice and outcomes of ER for OG cancer across centres using a large multicentre data set. METHODS: A retrospective analysis was conducted using the CONGRESS database, a UK-based international multicentre registry of patients with T1 N0 OG cancer between 2015 and 2022. Demographics, pathology, and outcomes for patients treated with ER were analysed. Centres were ranked according to ER volume, and patients were stratified into tertiles (low-, medium- and high-volume) with comparable numbers of patients in each group. Outcomes between low- and high-volume centres were compared using non-parametric tests and multivariable regression. The primary outcomes of interest were rates of R1 resection, procedural complications, and progression to surgery. RESULTS: In all, 1215 patients from 28 centres were included. The median ER volume per centre for OG cancer was 72 (interquartile range 43-150) over the 8-year period. R1 resection rates ranged from 0 to 67% (mean 28.1%, median 26.5%), and complication rates ranged from 0 to 50% (mean 7.8%, median 4.0%). Patients in the high-volume tertile had lower rates of R1 resection (17.3% versus 26.4%; P = 0.001), complications (3.8% versus 8.2%; P = 0.007), and progression to surgery (9.8% versus 20.7%; P < 0.001) than patients in the low-volume tertile. These differences remained after adjustment for patient and tumour variables, with odds ratios of 0.63 (95% confidence interval (c.i.) 0.42 to 0.94; P = 0.023), 0.42 (95% c.i. 0.22 to 0.79; P = 0.007), and 0.40 (95% c.i. 0.26 to 0.63; P < 0.001) for R1 resection, procedural complications, and subsequent surgery, respectively. CONCLUSION: This study highlights significant interhospital variation in clinical outcomes for ER in OG cancer. A greater understanding of underlying factors is needed to optimize patient outcomes.
Cole et al. (Mon,) studied this question.