Acute Pancreatitis Severity Prediction Model Based On Common Clinical Indicators

Objective:To establish prediction model of acute pancreatitis severity by collecting common clinical indicators.Methods:A total of patients diagnosed with AP from Yijishan Hospital of Wannan Medical College from April 1,2016 to December 31,2021 were retrospectively collected.According to the following inclusion and exclusion criteria,a total of 1105 AP patients were included in this study.AP patients were divided into mild acute pancreatitis(MAP)group and non-mild acute pancreatitis(NMAP,namely MSAP+SAP)group.Common clinical indexes of the two groups were compared.Age,sex,cost,BMI,vital signs,etiology,time from onset to admission,length of stay,length of ICU,final diagnosis,prognosis,previous comorbidities(hypertension,diabetes,coronary heart disease,hyperuricemia,fatty liver),adverse addictions(smoking,drinking);Laboratory indicators(first measurements)within 48 hours of admission: serum amylase(AMY),blood lipase(LIP),albumin(ALB)and other serological data.Other clinical data: Pleural effusion,Ascites,imaging examination(abdominal CT,chest CT,abdominal B-ultrasound),SIRS score,Ranson score,BISAP score,MCTSI score.The least absolute shrinkage and selection operator(LASSO)was used to screen variables,incorporate multi-factor logistic regression and establish a logistic regression model.The model was visualized by line charts and formulas,and the AUC area of ROC curve,calibration chart and Decision curve analysis(DCA)were used to evaluate the differentiation,accuracy and clinical practicability of the model.Results:1.In terms of general data,a total of 1105 patients were included in this study,including 741 patients in the MAP group and 364 patients in the NMAP group(common cases of MSAP and common cases of SAP).There were 0 deaths in the MAP group and 45 deaths in the NMAP group.There was no statistical significance in BMI between the two groups(P>0.05).The mean hospital stay of MAP patients was 10.22±5.36 days,and that of NMAP patients was 19.53±15.42 days,with statistical significance between groups(P<0.05).The average length of ICU care in the MAP group was 0.65±2.71 days,and the average length of ICU care in the NMAP group was 6.68±13.82 days,which was statistically significant(P<0.05).The average hospitalization cost of MAP was 19650.95±18416.41 yuan,and the average hospitalization cost of NMAP was 32329.46±40716.15 yuan,which was statistically significant between groups(P<0.05).In terms of hypertension history,210 cases(28.3%)in MAP group and 151 cases(41.4%)in NMAP group had statistical significance between groups(P<0.05).In terms of coronary heart disease history,237 cases(31.9%)in the MAP group.2.Etiology analysis: The comparison of etiology between MAP and NMAP groups concluded that there was statistical significance between hyperlipidemia etiology and idiopathic etiology(P<0.05).There was no statistical significance in bile,alcohol,diet,mixed etiology or other etiology(P>0.05).3.Serological analysis.MAP and NMAP,ALB,GLOB,TBIL,BUN,Cr,CYS-C,GLU,TC,TG,HDL,APOA1,LDH,Ca2+,WBC,NEUT,LYM,MO,RBC,HB,HCT,PCT,PT,APTT,FDP,DD,CRP,NLR,LMR and PLR were statistically significant(P<0.05).There was no significant difference in PA,TBA,ALT,AST,GGT,ALP,UA,SOD,LDL,APOB,LP,MCV,MCH,MCHC,RDW,PLT,MPV,PDW,TT,Fib and PNI between MAP and NMAP(P>0.05).4.Comparison of pleural effusion,ascites and various scores: In the MAP group,260patients(35.1%)had pleural effusion,70 patients(9.5%)had ascites,and 84 patients(11.3%)had SIRS.In NMAP,255 patients(70.1%)had pleural effusion,138 patients(37.9%)had ascites,and 127 patients(34.8%)had SIRS.Ranson,BISAP and MCTSI scores in MAP group were significantly lower than those in NMAP group,with statistical significance(P<0.05).5.Screening variables.Among 34 candidate predictors,SIRS,pleural effusion,ascites,BUN,LDH,NEUT and CRP were selected by LASSO regression as predictors of NMAP.6.The establishment and evaluation of the model are based on multi-factor logistic regression.Draw a Logit regression equation of the model(P)= 1.262 +(0.503)*SIRS +(0.78)+(0.865)* * pleural effusion ascites LDH + 0.09 + 0.002 * * BUN NEUT + 0.007 + 0.041 * * CRP(= 1 have a pleural effusion,pleural effusion,No pleural effusion =0,ascites =1,no ascites =0).The AUC value of the new prediction model was 0.8355(95%CI=0.810-0.861),which was higher than that of BISAP(AUC= 0.747),MCTSI(AUC=0.761)and Ranson(AUC= 0.696).Conclusion:1.The length of hospital stay,cost and mortality in NMAP group were significantly higher than those in MAP group.There were no significant differences between the NMAP group and the MAP group in gender,age,BMI,smoking and drinking habits,and coronary heart disease history.2.The most common etiology in MAP and NMAP groups was biliary origin.3.SIRS,pleural effusion,ascites,BUN,LDH,NEUT and CRP were independent predictors of NMAP.4.The new clinical prediction model based on SIRS,pleural effusion,ascites,BUN,LDH,NEUT and CRP has higher sensitivity and specificity,and its prediction value is better than Ranson,BISAP and MCTSI scores,with better differentiation and calibration.5.The new prediction model can predict the severity of AP effectively,cheaply,quickly and simply,which is helpful for clinicians to make scientific decisions in early stage.