Establishment And Verification Of Predictive Model For Prognosis Of Aneurysmal Subarachnoid Hemorrhage Treated By Embolization

Objective:Design a novel predictive model that can more accurately predict the prognosis of patients with aneurysmal subarachnoid hemorrhage after interventional embolization at an early stage,and evaluate its validity in comparison with World Federation of Neurological Societies Scale(WFNS)and prognosis on admission of aneurysmal subarachnoid hemorrhage(PAASH).Methods:In this retrospective study,273 patients with aneurysmal subarachnoid hemorrhage were collected and treated by interventional embolization in the Department of Neurosurgery of the affiliated Hospital of Guizhou Medical University from January 1,2019 to June 30,2021 as the modeling group.The patients were divided into good prognosis group and poor prognosis group according to Glasgow Outcome Scale(GOS).The demographic characteristics,past history,imaging data,morphological characteristics of aneurysms,time from onset to operation and laboratory results of admission were collected.The included indexes were analyzed by univariate analysis,and the factors with statistical significance were incorporated into multivariate analysis to determine the independent risk factors affecting prognosis.The nomogram prediction model is drawn by R language.The validity of the prediction model was evaluated in 138 patients who attended Jinyang Hospital affiliated to Guizhou Medical University for aneurysmal subarachnoid hemorrhage and were used to externally verify the prediction model.The effectiveness of the model was verified by calibration curves drawn by Bootstarp method,and compared with WFNS and PAASH,and draw the respective receiver operating characteristic curves(ROC)and calculate the areas under the ROC curve(AUC),and then verify their accuracy.Results:1.Among the 273 patients in the modeling group in this study,62(22.7%)had poor prognosis.Univariate analysis showed that patient age,neutrophil to lymphocyte ratio(NLR),Hunt-Hess Scale,modified CT-Fisher grade,wide-necked aneurysm and C-reactive protein(CRP)were associated with poor prognosis,and the difference was statistically significant(P<0.05).Multivariate analysis revealed that NLR≥7.16 [Odd ratio(OR),15.277;P=0.001],Hunt-Hess III(OR,5.454;P=0.001)and IV-V(OR,19.639;P=0.001),modified CT-Fisher III-IV(OR,8.345;P<0.001)and CRP >5mg/L(OR,3.614;P=0.001)were independent predictors of poor prognosis in patients with aneurysmal subarachnoid space who underwent interventional embolization.2.Nomogram prediction model was drawn using these 4 factors,optimized and validated.The AUC of the modeling group was 0.929(P<0.001,95%CI: 0.893-0.964),the sensitivity was 91.0%,and the specificity was 80.6%.The AUC of the validation group is 0.892(P<0.001,95%CI: 0.821-0.964),indicating excellent predictive is satisfactory.The calibration curve drawn by Bootstarp method showed that the uniformity between the prediction and the actual probability of the prediction model.Compared with WFNS grade and PAASH scale,the AUC of WFNS grade was 0.814(95%CI: 0.741～0.866)and AUC of PAASH scale was 0.808(95%CI: 0.735～0.880).The prediction model was found that have the best predictive efficacy,which was statistically significant.Conclusions:The prediction model established in this study has good discrimination and calibration,it could predict the prognosis of patients with aneurysmal subarachnoid hemorrhage undergoing interventional embolization more accurately in the early stage,and provide decision-making reference for the follow-up treatment of patients.