Evaluation of Prognostic Criteria for Severe Acute Pancreatitis in Patients With Early Nasogastric and Nasojejunal Feeding

Abstract

   Objective: To evaluate the prognostic value of predictors of severe acute pancreatitis in patients with nasogastric (NG) and nasojejunal (NJ) feeding.   Materials and methods: Our open-label randomized controlled study was carried out in the intensive care unit of Neftyanik Occupational Health Facility (Tyumen, Russian Federation) between November 2012 and October 2018 and included 64 patients who had predictors of severe acute pancreatitis (APACHE II score > 8, C-reactive protein [CRP] > 150 mg/L, SOFA score > 2) and could start early NG and NJ feeding. We evaluated the prognostic value of APACHE II, SOFA, BISAP, blood urea, and CRP in the general group and for different enteral feeding methods during the first 24 hours and in 48 hours. Data were statistically processed using the SPSS 26 software.   Results: Only one prognostic model was statistically significant for the first 24 hours and included APACHE II (AUC = 0.705, 95 % CI 0.577-0.834, P = 0.005, cutoff threshold of 8.5, sensitivity of 0.394, and specificity of 0.935 for all patients). In 48 hours models of the following groups were statistically significant: that of all patients, which included APACHE II, CRP, and blood urea (AUC = 0.904, 95 % CI 0.826-0.983, P < .001, cutoff threshold of 0.65, sensitivity of 0.818, and specificity of 0.935), NJ feeding group with CRP (AUC = 0.775, 95% CI 0.611-0.94, P < .001, cutoff threshold of 209.0 mg/L, sensitivity of 0.529, and specificity of 1.0), and NG feeding group with APACHE II (AUC = 0.767, 95% CI 0.599-0.934, P < 0.001, cutoff threshold of 9, sensitivity of 0.625, and specificity of 0.882).   Conclusions: Enteral feeding method does not affect the prognostic criteria for severe acute pancreatitis. 48 hours after admission is the best period period for severity prognosis, and the model with such independent predictors as APACHE II, blood CRP, and urea showed the best quality prognosis. If these predictors could not be used at once, we recommend APACHE II for NG feeding and blood CRP for NJ feeding.