| Objective To investigate the significance of early laboratory indicators in predicting the severity of acute pancreatitis(AP),and establish the prediction model,so as to provide a basis for the early assessment and treatment of acute pancreatitis.Methods A total of 779 adult(> 18 years old)and 99 children(≤18 years old)AP patients who were hospitalized in the Affiliated Hospital of Qingdao University from January2016 to December 2020 were enrolled.The patients were divided into mild acute pancreatitis(MAP)group,moderate severe acute pancreatitis(MSAP)group and severe acute pancreatitis(SAP)group,and the former two groups were non-severe acute pancreatitis(NSAP)group.The general clinical data,laboratory examination and imaging examination results of patients were collected.The inflammatory indicators,blood routine,biochemical indicators and blood coagulation routine were compared between SAP group and NSAP group within 24 hours of admission or within 24 hours of onset during the hospitalization.SPSS25.0 statistical software was used to analyze and process the data.Univariate analysis was used to determine the indicators related to predicting the severity of AP.Multiple factor binary Logistic regression analysis was used to determine the independent predictors of SAP.The early laboratory indicator prediction model of SAP was built based on Logistic regression model.The prediction ability of each independent predictor and prediction model was evaluated by receiver operating characteristic(ROC)curve,area under the curve(AUC)and optimal cut-off value.Results1.General clinical data: there were no significant differences in gender and age between the SAP group and the NSAP group(P > 0.05).The length of hospital stay in the SAP group was significantly longer than that in the NSAP group,with statistically significant differences(P < 0.05).The hospital mortality in the SAP group was higher than that in the NSAP group,with statistically significant differences(P < 0.05).2.Univariate analysis of early laboratory indicators for predicting SAP: In adult patients,the levels of c-reactive protein(CRP),procalcitonin(PCT),white blood cell count(WBC),neutrophil count(N),red blood cell distribution width(RDW),platelet distribution width(PDW),aspartate aminotransferase(AST),lactate dehydrogenase(LDH),triglyceride(TG),blood glucose(GLU),D-dimer and prothrombin time(PT)in SAP group were higher than those in NSAP group,while the levels of lymphocyte count(L),albumin(ALB),high density lipoprotein cholesterol(HDL-C)and antithrombin-Ⅲ(AT-Ⅲ)were lower than those in NSAP group,differences were statistically significant(P< 0.05).In children’s patients,the levels of CRP,PCT,RDW,direct bilirubin(DBIL),alanine transaminase(ALT),LDH,D-Dimer and PT in SAP group were higher than those in NSAP group,while the levels of L,ALB,total cholesterol(TC),low density lipoprotein cholesterol(LDL-C),HDL-C and AT-Ⅲ in NSAP group were lower than those in NSAP group,and the differences were statistically significant(P < 0.05).3.Multivariate binary Logistic regression was used to analyze the independent predictors of SAP: In adult patients,CRP(OR=1.013,95%CI=1.008~1.019),RDW(OR=1.089,95%CI=1.016~1.166),LDH(OR=1.005,95%CI=1.003~1.008),GLU(OR=1.098,95%CI=1.013~1.227)and HDL-C(OR=0.433,95%CI=0.165~0.835)were independent predictors of SAP(P < 0.05).In children’s patients,CRP(OR=1.021,95%CI= 1.008~1.034),ALB(OR=0.889,95%CI=0.795~0.995),PT(OR=1.881,95%CI=1.229~ 2.878)were independent predictors of SAP(P < 0.05).4.Construction of early laboratory index prediction model of SAP based on lo gistic regression model: The formula of the SAP prediction model constructed base d on the independent predictors of adult SAP is: Logit P1=﹣8.935+0.013×(CRP)+0.085×(RDW)+0.005×(LDH)+0.093×(GLU)﹣0.837×(HDL-C),the probability model for predicting adult SAP is: P1=1/{1+exp(﹣Logit P1)}.The formula of the SAP prediction model constructed based on the independent predictors of children’s SAP is: Logit P2=﹣7.660+0.021×(CRP)+0.632×(PT)﹣0.117×(ALB),and the probability model for predicting children’s SAP is: P2=1/ {1+exp(﹣Logit P2)}.5.The ROC curve was used to analyze the predictive value of each independent predictor and prediction model for SAP: In adult patients,the AUC of CRP,LDH,RDW,GLU and HDL-C alone predicting SAP were 0.808,0.823,0.649,0.662,and 0.643,respectively.The AUC of the adult SAP prediction model constructed based on the above indicators is 0.899,and the prediction performance is better than that of a single indicator.The optimal cut-off value of Logit P1 corresponding to the maximum Youden index was-2.9,and the prediction sensitivity and specificity were 95.2% and 72.0% respectively.In pediatric patients,the AUC of CRP,ALB and PT alone for predicting SAP were 0.752,0.765,and 0.820,respectively;the AUC of the children’s SAP prediction model constructed based on the above indicators was 0.938,and the predictive performance was better than a single indicator.The optimal cut-off value of Logit P2 corresponding to the maximum Youden index was-2.2,and the prediction sensitivity and specificity were93.8% and 79.5% respectively.Conclusions1.Among the early laboratory indicators,CRP,LDH,RDW,GLU and HDL-C were independent predictors of adult SAP.CRP,LDH,RDW and GLU were independent risk factors for SAP,and HDL-C was independent protective factor for SAP.2.Among the early laboratory indicators,CRP,ALB,and PT were independent predictors of SAP in children.Among them,CRP and PT were independent risk factors for SAP,and ALB was an independent protective factor for SAP.3.Based on independent predictors,the adult and pediatric SAP prediction models constructed by Logistic regression model have good performance in assessing the severity of AP,which is better than a single laboratory indicators.It can be tried to be applied in clinical practice to assist clinicians to identify and diagnose SAP earlier so as to take active intervention measures timely to improve prognosis. |
PDF Full Text Request