Prognostic nomogram for acute pancreatitis patients: An analysis of publicly electronic healthcare records in intensive care unit

The mortality rate of severe acute pancreatitis (AP) is 20–30% even after admission to intensive care unit (ICU). Thus we aimed to develop a laboratory-based nomogram to identify AP patients at high risk for mortality. The primary and validation cohorts were extracted from the Medical Information Mart for Intensive Care III database (MIMIC-III). Independent predictors were determined using multiple Cox analysis and then assembled to predict survival. The performance of proposed nomogram was evaluated by Harrell's concordance index (C-index) and area under the receiver operating characteristic (AUC) analysis, and subsequently compared with conventional scoring systems. A total of 342 AP patients admitted to ICU were enrolled, with 30-day, 180-day and 1-year mortality rate of 10.8%, 16.1% and 17.5%, respectively. Independent factors from multivariate Cox model to prognosticate 30-day and 1-year mortality were retrieved. The C-index of 1-year prediction nomogram (0.758, 95%CI: 0.676–0.840) were superior to several prediction approaches, and these findings were further confirmed by applying time-specific AUC analysis. Decision curve analysis indicated our nomogram was feasible in clinical practice. Similar results were observed in the validation cohort. The proposed nomogram gives rise to accurately prognostic prediction for critically AP patients admitted to ICU. •Up to 25% of AP patients proceed to severe conditions, in whom the mortality rate is 20-30% even after admission to ICU.•It is pivotal to predict patient outcomes, thus advantageous in permitting clinicians for selection of care level and therapeutic allocation.•It is imminent to develop a model incorporating routine, readily accessible and reliable measurements without extra cost.•More recently, several inflammation-based markers have been intensively verified for prognostication in AP.•Medical nomogram leads to rapid computation, together with enhanced accuracy and more concrete demonstration, allowing for seamless implementation of results to help treatment allocation.