Characterization of long-term prognosis in acute pancreatitis: An explorative analysis

Abstract

Background/objectives: Severity classification systems of acute pancreatitis (AP) assess inpatient morbidity and mortality without predicting outpatient course of AP. To provide appropriate outpatient care, determinants of long-term prognosis must also be identified. The aim of this study was to define clinical groups that carry long-term prognostic significance in AP.

Methods: A retrospective study that included patients admitted with AP was conducted. Determinants of long-term prognosis were extracted: These included Revised Atlanta and Determinant Based Classification (RAC), Charlson Comorbidity Index (CCI), Modified CT Severity Index (MCTSI), etiology, and local complications (LCs). Seven surrogates of morbidity up to 1 year after discharge were also collected and subsequently imputed into a clustering algorithm. The algorithm was set to produce three categories and multinomial regression analysis was performed.

Results: 281 patients were included. The incidences of morbidity endpoints were similar among the 3 RAC categories. Three clusters were identified that carried long-term prognostic significance. Each cluster was given a name to reflect prognosis. The limited AP had the best prognosis and included patients without LCs with a low co-morbidity burden. The brittle AP had a low co-morbidity burden and high MCTSI (LCs 94%). It ran a very morbid course but had excellent survival. The high-risk AP had the worst prognosis with the highest mortality rate (28%). They had a high co-morbidity burden without local complications.

Conclusion: Categories that carry long-term prognostic significance in AP have been developed. This study could help formulate appropriate follow-up and ultimately improve AP outcomes.

Keywords: Classification; Cluster analysis; Morbidity; Mortality; Outpatient; Patient discharge.