RESUMO
BACKGROUND AND AIMS: The clinical course of necrotizing pancreatitis (NP) is variable and unpredictable, with some patients managed conservatively, but a significant proportion become symptomatic and needing intervention for drainage and/or necrosectomy. The aim of this study was to identify patients based on baseline clinical and imaging metrics who will likely need intervention and therefore closer follow-up. METHODS: All NP patients managed in our institution between 2010 and 2019 were identified from a prospective database and those who did not undergo intervention during initial hospitalization were followed longitudinally post discharge until clinical and imaging resolution of necrosis. Patients were categorized into a conservative arm or intervention arm (endoscopic/percutaneous/surgical drainage and/or necrosectomy) for criteria defined according to IAP/APA guidelines. Clinical and imaging characteristics during initial presentation were analyzed between the two groups to identify independent predictors for eventual intervention using multivariable logistic regression. A nomogram was designed based on factors that were significant as defined by P value < 0.05. RESULTS: Among 525 patients, 340 who did not meet criteria for intervention during initial admission were included for study and followed for an average 7.4 ± 11.3 months. 140 were managed conservatively and 200 needed intervention (168 within 6 months and 32 after 6 months). Independent predictors of need for eventual intervention were white race [OR 3.43 (1.11-10.62)], transferred status [OR 3.37 (1.81-6.27)], and need for TPN [OR 6.86 (1.63-28.9)], necrotic collection greater than 6 cm [OR 8.66 (4.10-18.32)] and necrotic collection with greater than 75% encapsulation [OR 41.3 (8.29-205.5)]. A prediction model incorporating these factors demonstrated an area under the curve of 0.88. CONCLUSIONS: Majority of NP patients do not need intervention during initial admission but may require drainage/necrosectomy mostly in the first 6 months following discharge. Need for subsequent intervention can be accurately predicted by a combination of clinical and imaging features on index admission.