| Objiective:To develop and internally validate a visualized model with early predictive value for the severity of acute pancreatitis(AP).Methods:The research utilized a retrospective study including 600 patients with AP who received treatment in Department of Gastroenterology,First Affiliated Hospital of Kunming Medical University within 48 hours were enrolled from September 2017 to August 2018.Those 600 patients with AP were divided into non severe acute pancreatitis(NSAP,included mild acute pancreatitis and moderately severe acute pancreatitis)group and severe acute pancreatitis(SAP)group,according to the Atlanta classification of AP in 2012.Compared the general clinical data(age,sex,body mass Index,etc),etiology and laboratory test indicators[(amylase(AMY),lipase(LIP),Triglyceride(TG),Totalcholesterol(TC),High density lipoprotein(HDL),serum albumin(ALB),blood glucose(GLU),serum calcium(Ca),blood urea nitrogen(BUN),blood creatinine(Scr),C-reactive protein(CRP),D-dimer,fibrinogen(FIB),white blood cell count(WBC),neutrophil percentage(NEU),Neutrophil to lymphocyte ratio(NLR),Platelet distribution width(PDW)]within 48 hours after admission APACHEII score,Balthazar CT score and BISAP score,complications(ascites,pleural effusion),between the two groups.For the potential predictors of SAP,using the least absolute shrinkage and selection operator(LASSO)regressionto screen predictor variables;the variables selected by LASSO regression were included in the multivariable logistic regression analysis and estabilished a logistic regression:performance of the model was assessed with respect to its discrimination,calibration,and clinical usefulness.Bootstrap was used internally verified the model by the method of repeated sampling 1000 times.Results:1 General clinical data1.1 Comparison of prognosis between SAP group and NSAP group A total of 600 patients with acute pancreatitis were included in this study.Of them,335 patients were classified as mild,137as moderately severe and 128 as severe AP.The incidence of SAP was 9.7%in the study.Deaths were documented in 128 SAP patients,accounting for 21.09%.Among them,15 of which were due to severe sepsis,7 cases due to multiple organ failure,4cases because of abdominal compartment syndrome,and the remaining one from abdominal bleeding.However,none of hospital deaths in the NSAP group.1.2 Age The average age of patients in the SAP group were 47.38±15.56 years,and in NSAP group were 46.64±14.25 years.There were no significant difference in the age between the two groups(P>0.05)1.3 Gender Among the 128 patients in the SAP group,98(76.56%)males and 30(23.44%)females,with a male-to-female ratio of 3.3:1;among 472 patients in the NSAP group,313(66.31%)males and 159(33.69%)females were females.It were 2.0:1;there were statistical significance in gender between the two groups(P<0.05).1.4 BMI The average BMI(25.42±3.95)of patients in the SAP group was not significant different from that of the NSAP group(24.81±3.88).1.5 Hospitalization expenses and length of stay The length of stay in the SAP group was 20.33± 16.21(days),and the cost of hospitalization was 102930.70±136672.15(yuan).The length of stay in the NSAP group was 8.42±4.26(days)and the cost of hospitalization was 15586.76± 11821.71 (yuan).The difference was statistically significant(P<0.001)1.6 Previous history In the SAP group,a history of acute pancreatitis accounted for 35.94%,with a history of diabetes was16.41%,which was lower than that in the NSAP group(39.83%,19.07%),however,there were no statistical difference between the groups.2 Comparison between SAP group and NSAP group for etiology The most common cause of SAP group is chologenic,and there is a statistically significant difference in alcoholic factors between the two groups(P<0.05);while biliary,hyperlipidemia,diet,two or more causes,unknown causes,and others there were no statistically significant difference in cause between the two groups(P>0.05).3 Comparison of clinical indicators between the two groups The TG,GLU,BUN,HDL,GLU,BUN,Scr,CRP,D-dimer,FIB,WBC,NEU,NLR,APACHEII and BISAP scores in the SAP group were significantly higher than those in the NSAP group,with TC,HDL,ALB Both were significantly lower than the NSAP group.There were significant differences in pleural effusion and abdominal effusion between the two groups(P<0.05);there was no statistically significant difference inAMY,LIP,PDW,and fatty liver between the two groups4 Variables selection U sing LASSO regression to screen out CRP,BUN,D-dimer,and abdominal fluid as predictors of SAP,multivariate logistic regression analysis found that CRP(OR=1.009,95%CI:1.006-1.012,P<0.01),BUN(OR=1.185,95%CI:1.097-1.280,P<0.01),D-dimer(OR=1.166,95%CI:1.082-1.256,P<0.01),abdominal fluid(OR=4.848,95%CI:2.829-8.307,P<0.01)is the independent predictor of SAP.5 Model establishment and evaluation The equation of LR model is Logit(P)=-4.274+0.009*CRP(mg/L)+0.17*BUN(mmol/L)+0.153*D-dimer(ug/mL)+1.579*abdominal fluid(Note:When there is abdominal fluid,abdominal fluid=1,the another is 0),by using Logistic regression.Compared with 2 existing scoring systems,the model with a C statistic(AUC value)0.895(95%CI=0.865-0.926),which was superior to the APACHEII score(AUC 0.835,95%CI=0.791-0.878)and BISAP score(ACU 0.803,95%CI=0.760-0.846),the difference was statistically significant(P<0.05).The nomogram has a good ability to distinguish acute pancreatic severity.Hosmer-Lemesshow test 2=2.940,degrees of freedom=8,P=0.938,P value>0.05,the model showed favorable calibration.Application of the nomogram in the internal verification by Bootstrap still gave good discrimination(C-index,0.892)and calibration.The DCA demonstrated that the model was clinically useful,when using the model to predict and identify AP development to SAP and intervention was decided between the threshold probability of 2%-76%.Conclusion:1.Compared to the in-hospital mortality,length of stay and hospitalization expenses in the SAP group were significantly higher than those in the group of NSAP.2.Compared to the NSAP group,there were more male patients in the SAP group,and the statistical difference were significant.3.In terms of the etiology of SAP,biliary was the most common cause;There was a statistically significant difference in alcoholic factors between the SAP group and the NSAP group.4.CRP,BUN,D-dimer and abdominal fluid were independent predictors of SAP.5.The combination of CRP,BUN,D-dimer and abdominal fluid,translated into a model with good discrimination and calibration.The DCA demonstrated that the model was clinically useful.In the internal verification model performed well.As well,the model with high sensitivity and specificity,the predictive value of which is superior to complicated score systems such as APACHE Ⅱ score and BISAP score.The model allows a reliable estimation of AP,whichih might assist clinicians in tailoring precise therapy. |
