Supplementary Material for: The association between surgical patient selection and hospital variation in failure to cure in esophageal cancer surgery. A nation-wide cohort study.

Introduction; Failure to cure describes: 1)non-resectional(‘open-close’) surgery, 2)non-radical surgery(R1-R2) and/or 3)postoperative mortality. This study aims to investigate whether hospitals offering surgery to a large proportion of patients have higher failure-to-cure rates than hospitals operating fewer patients. Methods; From the Netherlands Cancer Registry, all cT1-cT4a/cTx-any cN-cM0 esophageal cancer patients diagnosed in 2015-2018 were included. For each center, the expected(E) proportion of patients undergoing surgery was established and divided by the observed(O) proportion. Hospitals were categorized into three groups: 1. hospitals treating relatively many patients with surgery, 2. average hospitals and 3. hospitals treating relatively few patients with surgery. Multilevel multivariable regression investigated the association between these hospital groups and failure to cure. Results; Some 3,437(53.2%) of 6,457 patients underwent surgery, ranging from 45-64% among 16 hospitals. The failure-to-cure rate was 15.0%(hospital variation[4.6-23.7%]). After categorizing, 1,003 patients underwent surgery in hospitals with low surgery rates (O/E ratio<0.94/corrected percentage<50%), 1,297 patients in average hospitals and 1,137 patients in hospitals treating many patients surgically (O/E ratio>1.01/corrected percentage>54%). Failure-to-cure rates were 16.8%, 12.2% and 14.0%, respectively. This was non-significant in multilevel analyses (aOR:0.63, 95%CI:0.38-1.05; aOR:0.76, 95%CI:0.46-1.24). Discussion/Conclusion; Failure-to-cure rates were similar in hospitals with a high surgery rate and hospitals with a low rate. Increasing the proportion of patients undergoing a resection may offer more patients a chance for cure.