AbstractBackgroundThe inability to comply with enhanced recovery protocols (ERp) after pancreaticoduodenectomy (PD) is a real but understated issue. Our goal is to report our experience and a potential tool to predict ERp failure in order to better characterize this problem.MethodsFrom January 1, 2014, to January 31, 2016, 205 consecutive patients underwent PD in our center and were managed according to an ERp. Failure to comply with postoperative protocol items was defined as any of: no active ambulation on postoperative day 1 (POD1); less than 4 h out of bed on POD2; removal of nasogastric tube and bladder catheter after POD1 and POD3, respectively; reintroduction of oral feeding after POD4; and continuation of intravenous infusions after POD4. Data were collected in a prospective database.ResultsTaking in consideration the number of failed items and the length of stay, we defined failure of the ERp as no compliance to two or more items. A total of 116 patients (56.6%) met this definition of failure. We created a predictive model consisting of age, BMI, operative time, and pancreatic stump consistency. These variables were independent predictors of failure (OR 1.03 [1.001–1.06] p = 0.01; OR 1.11 [1.01–1.22] p = 0.03; OR 1.004 [1.001–1.009] p = 0.02 and OR 2.89 [1.48–5.67] p = 0.002, respectively). Patient final score predicted the failure of the ERp with an area under the ROC curve of 0.747.ConclusionsIt seems to be possible to predict ERp failure after PD. Patients at high risk of failure may benefit more from a specific ERp.