Average treatment effect of pancreatico-duodenectomy versus total pancreatectomy

Abstract Purpose The effect size between total pancreatectomy (TP) over pancreatico-duodenectomy (PD) remains not well established due to the lack of high-level evidences. Aim of the study was to evaluate the average treatment effect on the patients treated (ATT) with PD in the case they received TP. Methods A reweighting scheme was applied to 202 TPs patients to equal 415 PDs (2016–2020), providing what would have happened to PD patients if they received TP. Superiority, non-inferiority and equivalence of PD and TP were assessed for outcomes’ risk ratios (RR). Results Balancing was perfect providing a robust estimation of ATT. If PDs were treated with TP they would more frequently experienced difficult glycemic control (RR: 1.44; 95%C.I.:1.21–1.73; p = 0.001) and received red blood cell (RBC) transfusion (RR: 1.23; 95%C.I.:1.04–1.46; p = 0.017). Conversely, TP would have reduced delayed gastric emptying (DGE; RR: 0.68; 95%C.I.:0.42–0.88; p = 0.008) as the likely avoidance of post-operative pancreatic fistula. Major morbidity and failure-to-rescue were within equivalence margins. In subgroups, TP would have not worsened glycemic control in patients already diabetics, transfusions would have not increased in patients with worse general conditions, and TP would have not reduced DGE in older, diabetics and female patients, more susceptible to gastric motility disorders. Post-operative stay and 30-day readmission were within equivalence margins. Conclusion If PD is feasible, the adoption of TP would be potentially more harmful than beneficial except in some clinical circumstances, supporting this choice in these cases. Both procedures finally provided similar major morbidity, failure-to-rescue, mortality, in-hospital stay and readmission rate.