The Predictive Value Of RDW/SC Combined With NLR In Severe Acute Pancreatitis

Objective:The prognosis of Acute Pancreatitis(AP)was correlated with its severity.Twenty percent of those with AP will develop severe disease(mortality is about 40percent).Therefore,Identification in early stage of the risk in patients with severe or fatal AP is critical to improve outcomes.At present,many indicators were used to evaluate the severity of AP in clinic,each of which has its own advantages and disadvantages.Acute Physiology and Chronic Health Evaluation(APACHE II),Imire score and Ranson score all have high sensitivity and specificity,but the Evaluation content and the operation are very complicated.Some indexs,such as Procalcitonin(PCT),C-reactive Protein(CRP),interleukin-6,lactate dehydrogenase(LDH),are often not applicated in the clinical to diagnosis SAP,which is not conducive to promotion in primary hospitals.However,total bile acid has low sensitivity and specificity in evaluating the prognosis of AP.Therefore,it is particularly important to explore a single or combined index that can be applied in conventional hospitals and to evaluate the severity of AP in a short time quickly and accurately.The purpose of this experiment is to analyze the data from AP patients of our hospital in the past 5 years.On the one hand,to calculate and explore the relationship between common indicators and the severity of AP as well as calculate the optimal diagnostic threshold,specificity and sensitivity to screen out better indicators.On the other hand,the combination of several better indicators as combined predictors was compared with the traditional BISAP scoring system to further prove the value and significance of the combined predictor so as to provide better guidance for clinical application.Methods:According to the purpose of our experiment,the criteria of inclusion and exclusion from research objects.Finally,116 AP patients were selected,including 43 Mild Sever Acute Pancreatitis(MAP),34 Moderately Sever Acute Pancreatitis(MSAP),and 39 Sever Acute Pancreatitis(SAP).MAP patients and MSAP patients were classified into non-SAP group.We utilized the software SPSS 25.0 to calculate and analyze the data from non-SAP group and SAP group,P<0.05 was defined as be significant statistically.First,the data of clinical general characteristics(including gender,age,Body Mass Index(BMI),smoking and drinking history,history of hypertension,etc.)and etiology(hypertriglyceridemia,biliary,alcoholic,and other)were compared between the two groups to explore what the relationship between above factors and the severity of AP is.Enumeration data was calculated by chi-square test.Quantitative data conforming to normal distribution was calculated by independent sample T test,and those inconsistent with normal distribution was calculated by non-parametric test.Second,the differences of indictors(such as RDW,RDW/SC,NLR,D-dimer and BUN)in the two groups were analyzed by univariate analysis.Then,multivariate analysis was performed to determine whether these predictors were independent factors affecting the severity of AP.Next,the best indicators with statistical differences were distinguished after univariate and multivariate analysis,the Receiver Operating Characteristic(ROC)curve was used to calculate Area under curve(AUC),Jordon index,specificity and sensitivity corresponding to this critical value.Eventually,we constructed the prediction model.At the same time,the ROC curves of the combined predictors and BISAP traditional score were drawn to compare the AUC values.Results:1.In the non-SAP group and SAP group,both the clinical general characteristics(gender,age,BMI,history of smoking and drinking,history of hypertension,etc.)and etiology traits(lipogenic,biliary,alcoholic,and other)of these patients all were analyzed.The distribution of quantitative data was conforming to normal distribution,but there was no statistical difference between the above data of two groups(P>0.05).2.In univariate analysis,the indexes of RDW,RDW/SC,NLR,D-dimer and BUN of AP patients did not conform to normal distribution,and the non-parametric test analysis showed that there were significant differences between these predictors in the two groups(P<0.05).Moreover,when compared above indicators in the two groups,the level of those were higher in SAP group.3.In the multivariate analysis,there were significant differences in RDW,RDW/SC and NLR(P<0.05)in the two groups.4.As seen in the ROC curves,the AUC of RDW was 0.861,the corresponding optimal threshold was 13.15(95%CI 0.795-0.927),the sensitivity was 0.872,the specificity was 0.74.The AUC of RDW/SC was 0.922,the corresponding to the optimal threshold was 6.495(95%CI 0.8744-0.97),and the sensitivity,specificity were 0.974,0.818,respectively.The AUC of NLR was 0.741,the corresponding optimal threshold was 8.175(95%CI 0.065-0.0833),the corresponding sensitivity was 0.769,the specificity was 0.636.5.Based on the purpose of this study and the above results,RDW/SC and NLR were finally defined as the joint predictor to construct the prediction model and verify.At the same time,the ROC curves of the combined predictors and BISAP traditional score were drawn to compare the AUC values.It was found that the optimal threshold of the prediction mode was-1.0769(95%CI 0.876 ～ 0.970)with sensitivity of 0.974,specificity of 0.805.However,the sensitivity and specificity of BISAP score were0.744 and 0.883,respectively.Meantime,appropriate degree of the prediction model was P > 0.05,AUC was 0.923,and BISAP AUC was 0.831,indicating that the prediction model had good predictive value.Conclusions:1.There was no significant correlation between these clinical traits of gender,age,BMI,history of smoking and drinking,history of hypertension,and the severity of AP.2.RDW,RDW/SC and NLR are independent factors can influence the prognosis of AP and can be applicated as predictors for SAP with high accuracy.And the predicted effect of RDW/SC was better than RDW.3.The SAP risk prediction model based on RDW/SC and NLR has high consistency and differentiation,and has good prediction value.4.The AUC and sensitivity of SAP risk prediction model were higher than BISAP,indicating that the predictive value of model for AP severity was higher than BISAP.