[Establishment of multiple predictor models of severe acute pancreatitis in elderly patients]

Objective: To investigate the prognostic factors related to the severity of acute pancreatitis and to establish the multiple predictor models of severe acute pancreatitis(SAP) in elderly patients. Methods: Clinical data of 146 consecutive elderly patients who met the inclusion criteria between January 2014 and May 2017 at Department of Pancreatic and Biliary Surgery, the First Affiliated Hospital of Harbin Medical University were retrospectively collected and analyzed, wherein 88 cases were mild acute pancreatitis, 29 cases were moderately severe acute pancreatitis and 29 cases were SAP. The patients data were subjected to univariate analysis and multiple classified Logistic regression analysis for independent prognostic factors of the severity of acute pancreatitis in elderly patients. Unweighted predictive score(unwScore) and weighted predictive score(wScore)for SAP in elderly patients were established according which the receiver-operating characteristic(ROC) curves of independent prognostic factors and predictor models were produced. The cutoff values of independeut prognostic factors and predictor models were determined. The area under the curve, the sensitivity, the specificity, the positive predictive value and the negative predictive value to verify the predictive efficiency of the independent prognostic factors and predictor models were calculated. Results: Procalcitonin(PCT)(Z=10.564, P=0.000), blood urea nitrogen(BUN)(Z=22.231, P=0.003), serum creatinine(Scr)(Z=14.151, P=0.030), serum calcium(Z=34.979, P=0.032) and pleural effusion(χ(2)=28.463, P=0.015) were independent prognostic factors of the severity of acute pancreatitis by univariate analysis and multiple classified Logistic regression analysis in elderly patients. Respectively, the area under the curve of PCT, BUN, Scr, serum calcium and pleural effusion were 0.908, 0.737, 0.701, 0.753, 0.712, the sensitivity were 0.828, 0.621, 0.552, 0.690, 0.517, the specificity were 0.915, 0.786, 0.846, 0.966, 0.906, the positive predictive value were 70.6%, 41.9%, 47.1%, 83.3%, 57.7%, the negative predictive value were 95.5%, 89.3%, 88.4%, 92.6%, 88.3%. Respectively, the area under the curve of unwScore and wScore were 0.915 and 0.953, the sensitivity were 0.759 and 0.931, the specificity were 0.889 and 0.915, the positive predictive value were 62.9% and 73.0%, the negative predictive value were 93.7% and 98.2%. Conclusions: PCT, BUN, Scr, serum calcium and pleural effusion were independent prognostic factors of the severity of acute pancreatitis in elderly patients. The multiple predictor models of SAP in elderly patients have a good predictive efficiency, which may provide valuable clinical reference for prediction and treatment.