Development and validation of a nomogram prediction model for adverse postoperative outcomes in hypertensive cerebral hemorrhage based on a multicenter retrospective study

ABUDURUSULI Rexiti; CHEN Julin; QIN Hu; FAN Guofeng; JIA Zheyong; JIANG Yanwen; WANG Yongxin

Abstract：Objective To explore the predictive value of various demographic characteristics and clinical risk factors for patients with hypertensive intracerebral hemorrhage undergoing surgical intervention,and to develop and validate a predictive model for adverse postoperative outcomes.Methods Based on a multicenter retrospective study and the International Classification of Diseases(ICD-10),the general data of patients with hypertensive cerebral hemorrhage who underwent surgical treatment in 24 representative hospitals in 11 prefectures(prefectures and prefecture-level cities)of Xinjiang Uygur Autonomous Region from January to December 2022 were collected.Factors such as gender,age,mean arterial pressure(MAP)at admission,GCS score,bleeding site,bleeding volume,ventricle invasion and midline shift were used as predictors.The outcome indicators were divided into poor prognosis(GOS=1-3 points)and good prognosis(GOS=4-5 points)according to the GOS score at discharge.The data of the First Affiliated Hospital of Xinjiang Medical University were used as the external validation set,and all the remaining data were randomly divided into the training set and the internal validation set in a ratio of 7:3.LASSO Logistic regression analysis was made to screen independent risk factors and establish a nomogram for predicting the poor prognosis of surgical treatment.The receiver operating characteristic(ROC)curve and the calibration curve were plotted using the predictive value of single factor Logistic regression.The area under ROC curve(AUC)was used to evaluate the predictive performance of the nomogram.The clinical Decision Curve Analysis(DCA)was drawn to determine the predicted net benefit threshold.Results A total of 1163 patients were included,with 699 in the training set,299 in the internal validation set,and 165 in the external validation set.In the overall cohort,60.5%were male,with a male to female ratio of 1.5:1.The median age was 60 years(IQR:52,69),with the majority of patients(76.0%)aging<70 years.Seven potential predictors(age,MAP,GCS score at admission,bleeding site,bleeding volume,ventricular rupture and midline displacement)were screened out by LASSO regression analysis for further model establishment.One-way Logistic regression and ROC curves were performed separately for each predictor,and AUC values for each predictor were greater than 0.5.Multivariate logistic regression analysis was then performed to screen statistically significant independent predictors(MAP at admission,GCS score,bleeding site,bleeding volume,and midline shift),and a visual and simplified nomogram was established.Based on the nomogram established by the training set,the prediction model was verified in the internal validation set and the external validation set.The results showed that the AUC values of the training set,internal validation set and external validation set were 0.886,0.900 and 0.826 respectively.The performance of the model was verified by plotting the Calibration curve and the clinical Decision Curve Analysis(DCA)of the validation set.The results showed that the Calibration curve of the model was close to the reference line,and the DCA curve was within the threshold probability range,indicating that the prediction results of the model were consistent with the actual results,which provided substantial net benefits for clinical application and had considerable application value.Conclusion The MAP at admission,GCS score,hemorrhage location and volume,and midline shift emerge as key predictors of adverse postoperative outcomes in hypertensive cerebral hemorrhage patients.The validated model offers significant predictive accuracy,enabling clinicians to refine clinical decision-making and enhance patient outcomes.