Analysis Of Risk Factors Of Hemorrhage Transformation After Intravenous Thrombolysis In Patients With Acute Ischemic Stroke And Research On Prediction Model

Background Acute ischemic stroke is a kind of disease which does great harm to human health.It is also one of the main causes of disability and mortality in our country.Intravenous thrombolysis is the most well-documented treatment for AIS and has been widely used worldwide.Hemorrhagic transformation as a serious complication after thrombolysis severely affects patient outcomes,but there are no guidelines that publicly recommend predictors and risk prediction models for hemorrhagic transformation after thrombolysis.Objective To analyze the risk factors of HT after intravenous thrombolysis in patients with AIS,and to establish the risk prediction model and draw the nomogram.Compared with HAT and GRASPS models,we hope to find a better model for predicting HT after thrombolysis in AIS patients.Methods A total of 285 AIS patients treated with RT-PA intravenous thrombolysis from October 2018 to July 2022 in Yichang Gezhouba Central Hospital were collected as a training cohort.A total of 115 AIS patients using RT-PA intravenous thrombolysis at Yichang Yiling District Hospital from October2020 to July 2022 were collected as a validation cohort.All patients should provide the following information:sex,age,drinking history,NIHSS score before thrombolysis,NIHSS score 2 hours after thrombolysis,homocysteine,baseline uric acid,baseline systolic blood pressure,baseline diastolic blood pressure,baseline blood glucose,whether anterior circulation infarction,whether atrial fibrillation,blood lipids,coagulation function,whether the first head CT has low density shadow,whether the thrombolysis time window is within 0-3 hours and whether the thrombolysis time window is within3-4.5hours.The patients were divided into HT group(n=41)and non-HT group(n=244).The independent risk factors of HT were analyzed.According to the selected variables,the risk prediction model was established,and the predictive factors were assigned and the nomograms were drawn.The internal and external validation of the model,draw the ROC and calculate the area under the curve to evaluate the performance of the model,draw the Calibration curve to check the consistency of the model,draw the DCA decision curve to evaluate the practicability of the model in clinic.The external validation of GRASPS and HAT scores was carried out by using the training cohort,which was compared with the modeling of this study.SPSS 26.0was used for single factor analysis and binary Logistic regression equation in this study,and the test level α =0.05.Line chart,ROC curve,Calibration Curve,DCA decision curve evaluation were drawn by R language.Result(1)The presence of HT after intravenous thrombolysis in AIS patients was associated with drinking,NIHSS score before thrombolysis,NIHSS score 2 hours after thrombolysis,baseline blood glucose,baseline diastolic blood pressure,atrial fibrillation and abnormal coagulation,the difference was statistically significant(P<0.05).(2)NIHSS score before thrombolysis(OR=1.214,95CI%:1.068-1.379,P<0.05),baseline blood glucose(OR=1.290,95CI%:1.120-1.485,P<0.001),baseline diastolic blood pressure(OR=1.062,95CI%:1.029-1.097,P<0.001),atrial fibrillation(OR=4.691.95CI%:1.924-11.437,P<0.05)were independent risk factors for HT after thrombolysis in AIS patients,the difference was statistically significant.(3)Risk prediction models were built and plotted on the basis of screened variables,and ROC curves were plotted to assess model discrimination,with an AUROC of 0.887(95% CI: 0.887-0.937)for the training cohort,an optimal cutoff point of 0.181,and a sensitivity of 0.829,the specificity was 0.861.The AUROC for the validation cohort was 0.892(95% CI: 0.816-0.968),and the optimal cutoff point for the validation cohort was 0.237 with a sensitivity of 0.929 and a specificity of 0.782.Using Calibration curve to evaluate model consistency and DCA decision curve to evaluate model practicability,it is concluded that the research model has good discrimination ability,Calibration degree and practicability.(4)The training cohort was scored with GRASPS and HAT and ROC curves were drawn.The results showed that the GRASPS AUROC was 0.779(95% CI:0.709-0.848),the optimal critical point was 81.5,the sensitivity was 0.659,and the specificity was 0.791.The Auroc of the HAT model was 0.776(95% CI:0.688-0.863),the optimal cutoff point was 0.5,the sensitivity was 0.756,and the specificity was 0.689.There was no significant difference in area between the two models(P > 0.05).Compared with this model,the AUROC,sensitivity and specificity of this model are better than GRASPS and HAT.Conclusion(1)Post thrombolytic HT in AIS patients is associated with factors such as alcohol consumption history,pre thrombolytic NIHSS score,2 hours post thrombolytic NIHSS score,baseline blood glucose,baseline diastolic blood pressure,atrial fibrillation,and abnormal coagulation function.NIHSS score before thrombolysis,baseline blood glucose,baseline diastolic blood pressure,and atrial fibrillation are independent risk factors for HT in AIS patients after thrombolysis.(2)Incorporating the above four variables to establish a risk prediction model and drawing a column chart,both internal and external validation have proven that the model has good discriminability,calibration,and clinical practicality.(3)HAT and GRASPS both have good predictive ability,but compared to the new predictive model constructed in this study,the predictive ability of this study model is superior to GRASPS and HAT.