Construction Of The Prediction Model Of Burn Sepsis With Clinical Laboratory Combined Scoring Index

Background and purpose:In the past few decades,the main cause of death of large area burn patients has changed from hypoxia to burn sepsis.Although post-burn wound infection and burn sepsis have a great impact on the diagnosis and treatment of burns,the prediction and diagnosis of burn sepsis is still a major challenge.Therefore,establishing a reasonable and effective early screening method for burn sepsis,identifying high-risk groups of burn sepsis,early detection and early treatment are the key to reduce the mortality of critically burned patients.The purpose of this study is to focus on the analysis of the risk factors of early onset of burn sepsis in adult severe and extremely severe burn patients,build a nomogram risk prediction model,provide a simple and effective screening method for adult severe burn patients with burn sepsis,improve the early prediction diagnosis rate and survival rate,and provide ideas for clinical diagnosis and treatment of burn sepsis.Methods:(1)According to the inclusion and exclusion criteria,78 adult patients with burn sepsis and 118 adult patients with non-burn sepsis after severe and extremely severe burns who were treated in Nantong University Affiliated Hospital from January 2012 to October 2022.10 were randomly divided into training set(70%)and test set(30%).(2)Baseline analysis was carried out on the general clinical data of patients and clinical indicators of laboratory serological tests,and single-factor logistic regression analysis was carried out on the data of patients in the training set.The selected indicators with statistical significance(P<0.05)were included in the multivariate binary logistic regression analysis.Finally,Lasso regression screening was carried out to confirm that the variables screened by logistic regression did not have over-fitting,Finally,the independent risk factors of sepsis after burn in adult critical patients were screened.(3)Logistics2.0.0 software was used to construct a nomograph prediction model for the risk of burn sepsis after severe and extremely severe burns.The area(AUC value)under the ROC curve(receiver operating characteristic curve)is used to evaluate the differentiation of the model.Hosmer-Lemeshow test,calibration curve and clinical decision curve were used to evaluate the fitting,calibration and clinical practicability of the model.At the same time,internal validation(bootstrap self-sampling 1000 times)and external validation(test set)are used to verify the predictive performance and clinical practical value of the model.(4)The Baux score and ABSI score were used as independent variables,and death was used as dependent variable for Cox regression analysis,and the ROC curve was drawn.After the critical value was obtained,the measurement data was converted into grouping data.The survival analysis of adult severe and severe burn sepsis was performed using Medcalc software.Results:(1)Through logistic regression analysis,three independent risk factors of burn sepsis in adult severe and extremely severe burn patients were screened out: venous blood D-dimer(p=0.022,OR=2.300,95% CI(1.275-5.619)),antithrombin-Ⅲ(p=0.001,OR=0.679,95% CI(0.493-0.809)),and ABSI score(p=0.001,OR=9.174,95%CI(3.263-5.182)on the third day after injury,P<0.05.(2)Based on the above risk factors,a nomograph prediction model for the risk of burn sepsis in adults after severe and extremely severe burns was successfully constructed.The AUC values of the model in the training set and the test set were 0.935(95% CI: 0.895-0.974)and 0.865(95% CI: 0.763-0.986),respectively,indicating that the model has good prediction performance for the risk of burn sepsis after severe and extremely severe burns in adults.The critical value of the ROC curve in the training set is 0.383,the sensitivity is 82.1%,and the specificity is 92.6%.The P values of the Homer Lemeshow test in the training set and the test set are 0.423 and 0.612 respectively,both P values are>0.05,indicating that the model fitting effect is good.The calibration curve showed that the model predicted the incidence risk of burn sepsis after severe and extremely severe burns in adults and the actual incidence risk were in good agreement.The clinical decision curve of the model in the training set and test set shows that the model has certain clinical practical value.(3)Cox regression analysis was conducted with Baux score and ABSI score as independent variables and death as dependent variables.Both were independent risk factors for death of burn and sepsis in adult severe patients.The ROC curve was drawn.The critical values of Baux score and ABSI score were 117 and 9 respectively;After changing the two kinds of measurement data into counting data,the adult patients with sepsis after severe burn were divided into three groups: low,medium and high risk.The survival curve drawn showed that the combination of the two scores was more effective in distinguishing the prognosis of patients.Conclusion:(1)ABSI score,D-dimer and antithrombin-Ⅲ are independent risk factors for predicting sepsis in severe burn patients.(2)The nomograph prediction model for the incidence risk of burn sepsis in adults after severe and extremely severe burns has good prediction ability and clinical practicability,and provides a simple and effective screening method for adult severe and extremely severe burn sepsis.(3)The combination of Baux score and ABSI score is more effective in distinguishing the prognosis of adult patients with burn sepsis after severe and extremely severe burns.