Predictive models to calculate individual surgical mortality are a critical element of “personalized surgery.” Risk modeling is particularly suited to technically challenging operations with a narrow therapeutic index, such as radical pancreaticoduodenectomy for pancreatic cancer, and will transform surgery from a “practice” to reproducible performance.
The current report by Venkat et al1 objectifies an emotionally charged “go, no go” decision for patients and surgeons dealing with pancreatic cancer and promises to replace medical “opinion” with transparency. “Predicting the risk of perioperative mortality in patients undergoing pancreaticoduodenectomy” is an elegant distillation of readily obtained patient factors affecting perioperative mortality at an expert center for pancreatic surgery. The risk predictions are statistically significant and clinically meaningful and merit prospective multicenter validation.