Who Do We Fail to Rescue after Pancreatoduodenectomy? Outcomes Among >4000 Procedures Expose Windows of Opportunity

Objective: Our investigation on in-hospital mortality after 4474 pancreatoduodenectomies aimed to identify time-dependent risks as well as windows of opportunity to rescue patients from complications. Background: Pancreatoduodenectomy is generally considered a safe procedure with a 1-10% perioperative mortality based on complexity and surgical volume. Yet, patients are susceptible for life-threatening complications particularly with extended resections. Recognition of distinct vulnerabilities over time while patients recover is required to permit focused monitoring, sophisticated resource allocation, and greatest surgical safety. Methods: Patients who deceased in-hospital after pancreatoduodenectomy between 2003-2021 were retrieved from the institutional pancreatectomy registry and analyzed in detail with respect to their postoperative course. Results: Among 4474 pancreatoduodenectomies, 156 patients deceased in-hospital (3.5%). When assessing root causes of mortality, we observed 3 different clusters of complications which were postpancreatectomy-specific (47.4%), visceral vasculature-associated (25.6%), or cardiopulmonary in origin (23.7%). The median times of root cause onset in the 3 categories were postoperative day (POD) 9, POD 4.5 ( P =0.008) and POD 3 ( P <0.001), and medians of in-hospital mortality were POD 31, POD 18 ( P =0.009) and POD 8 ( P <0.001), respectively. Intervals between root cause onset and mortality varied with medians of 23 days, 11 days ( P =0.017), and 1 days ( P <0.001). The 3 categories were similarly distributed between different types of surgical complexity. Conclusion: Postpancreatectomy-specific complications prompt almost half of in-hospital mortalities after pancreatoduodenectomy, with rather long intervals for interventions to prevent failure to rescue. In contrast, visceral vasculature-related events and cardiopulmonary complications dominate early in-hospital mortalities with short intervals until mortality, demanding rigorous management of such events or preoperative conditioning. These data externally validate a previous high-volume initiative and highlight distinct windows of opportunity to optimize perioperative safety.