Construction and validation of a predictive model for disease outcomes in the patients with traumatic hemorrhagic shock caused by multiple fractures

Cai Haitao; Wang Jianqiang

doi:10.3969/j.issn.1002-1949.2024.09.003

Abstract：Objective To construct a predictive model for disease outcomes in the patients with traumatic hemorrhagic shock caused by multiple fractures,and to conduct internal and external validation.Methods A total of 126 patients with traumatic hemorrhagic shock caused by multiple fractures who visited Dezhou Hospital,Qilu Hospital of Shandong University from January 2023 to January 2024 were selected as the modeling set for model construction,and 100 patients with traumatic hemorrhagic shock caused by multiple fractures were selected as the validation set for model validation.The patients in the modeling set were divided into survival group(n=95)and death group(n=31)based on their in-hospital disease outcomes.Clinical data from the two groups were compared,and univariate and multivariate Logistic regression analysis were used to determine the risk factors affecting disease outcomes in the patients with traumatic hemorrhagic shock caused by multiple fractures.A prediction model was established based on the selected risk factors,the diagnostic effectiveness of the prediction model was evaluated by using the receiver operating characteristic(ROC)curve and calibration curve,and a clinical decision curve analysis(DCA)was used to analyze the benefit rate of the prediction model.Results The age ≥60 years old(83.87％vs.30.53％),comorbid underlying diseases(29.03％vs.10.53％),combined intracranial hemorrhage(38.71％vs.15.79％),time from injury to emergency＞4 h(64.52％vs.33.68％),injury severity score(ISS,points:33.54±4.52 vs.29.89±3.42),6 h lactate value(mmol/L:5.21±0.22 vs.3.32±0.87),activated partial thromboplastin time(APTT,s:39.90±3.45 vs.36.42±2.94),thrombin time(TT,s:17.21±2.87 vs.15.45±1.76),prothrombin time(PT,s:16.98±2.19 vs.14.23±1.98)were higher in death group than in the survival group,and Glasgow coma scale(GCS,points:4.53±0.98 vs.10.23±2.42),fibrinogen(Fib,g/L:2.34±0.32 vs.3.87±0.33)were lower in death group than those in the survival group(P＜0.05).Multivariate Logistic regression analysis showed that the time from injury to emergency(OR=3.898,95％CI 1.287-8.275),GCS score(OR=3.978,95％CI 1.814-7.989),ISS score(OR=2.342,95％CI 1.191-4.375),6 h lactate value(OR=2.881,95％CI 1.239-5.689)and Fib(OR=2.543,95％CI 1.198-5.389)were independent risk factors for the death in the patients with traumatic hemorrhagic shock caused by multiple fractures(P＜0.05).A nomogram prediction model for mortality risk in the patients with traumatic hemorrhagic shock caused by multiple fractures was constructed based on the time from injury to emergency(X,),GCS score(X2),ISS score(X3),6 h lactate value(X4)and Fib(X5).The AUC of the prediction model in the modeling set was 0.897(95％CI 0.723-0.923),and the AUC of the prediction model in the validation set was 0.901(95％CI O.786-0.955).The calibration curve showed that the predicted probability of mortality risk in the patients with traumatic hemorrhagic shock caused by multiple fractures in both datasets was close to the actual incidence rate.DCA found that the curve of the nomogram prediction model was far away from the reference line with a net benefit of 0 and the reference line with all the samples being positive,indicating that the nomogram prediction prediction model has good predictive value for the risk of death in the patients with traumatic hemorrhagic shock caused by multiple fractures.Conclusions The time from injury to emergency,GCS score,ISS score,6 h lactate value and Fib are independent risk factors for the death in the patients with traumatic hemorrhagic shock caused by multiple fractures.The nomogram prediction model constructed based on these factors has high discrimination and calibration,and can provide a theoretical basis for early clinical evaluation of patients'disease outcomes and interventions.