Study On The Risk Model Of Symptomatic Intracranial Hemorrhage After Intravenous Thrombolysis In Acute Ischemic Stroke

BackgroundSymptomatic intracranial hemorrhage（sICH）is the most serious bleeding complication after ultra-early intravenous thrombolysis in patients with acute ischemic stroke（AIS）,which can cause up to 92%disability and 50%mortality,early evaluation and prevention of sICH are very important.At present,my country still lacks a recognized sICH risk assessment tool for clinical use.Objective1.To investigate the incidence of sICH in patients with AIS in a tertiary hospital in Shenzhen who received recombinant tissue plasminogen activator（rt-PA）intravenous thrombolysis,and to explore the influencing factors of sICH in order to provide reference for the formulation of accurate intervention measures.2.To verify the clinical application effect of Hemorrhage After Thrombolysis（HAT）score,Safe Implementation of Thrombolysis in Stroke-monitoring study（SITS）score,and Totaled Health Risks In Vascular Events（THRIVE）scores in domestic populations,screen the optimal score,explore sensitive baseline indicators for predicting the occurrence of sICH,and jointly screened out the best existing predictive scores,construct a sICH risk model and verifying it,improve the current situation of risk assessment of sICH high-risk patients and promote the development of screening tools.MethodsThis study is a retrospective cohort study.In the first part,462 AIS patients who received rt-PA intravenous thrombolysis in the emergency green channel of a tertiary hospital in Shenzhen from January 2014 to December2020 were continuously collected as the research objects.data to explore the important influencing factors of sICH.In the second part,the predictive value of HAT,SITS and THRIVE three foreign scores for domestic population was verified in 462 AIS patients,based on the area under the receiver operating characteristic curve（ROC）（Area Under ROC Curve,AUC）and Akaike Information Criterion（AIC）to screen the best score.In the third part,308 AIS patients from January 2014 to May 2019 were continuously collected as the modeling group,and 154 AIS patients from June 2019 to December 2020 were collected as the validation group.In the modeling group,univariate analysis was used to screen the baseline indicators with statistically significant differences,and multivariate logistic regression analysis was used to screen the independent influencing factors of sICH.Combining independent influencing factors with the best existing predictive scores to construct and validate the sICH risk model.Use ROC,AUC to compare model discrimination,calibration curve,Hosmer-Lemeshow（H-L）test to evaluate model calibration ability,clinical utility of comparative models for decision curves,sensitivity,and specificity.Results1.Influencing factors of sICH after intravenous thrombolysis in patients with AIS.462 patients were collected,of whom 20 patients（4.33%）developed sICH.Multivariate analysis showed that the National Institute of Health Stroke Scale（NIHSS）（odds ratio[OR]=1.11,95%confidence interval[CI]=1.03-1.19）,onset to treatment time（OTT）（OR=1.02,95%CI=1.01-1.03）and history of atrial fibrillation（OR=5.31,95%CI=1.68-16.75）and hyperdense middle cerebral artery sign（HMCAS）（OR=4.18,95%CI=1.22-14.37）were independent risk factors for sICH（P<0.05）.Triglyceride（TG）level was the protective factor of sICH.TG was divided into Q1（<1.02mmol/L）,Q2（1.03mmol/L-1.44mmol/L）,Q3（1.45mmol/L-2.00mmol/L）and Q4（>2.00mmol/L）by quartile method.After adjusting the confounding factors,it was found that the risk of sICH in Q4 group was 93%lower than that in Q1 group（OR=0.07,95%CI=0.01-0.92,P=0.044）.The trend test was statistically significant（P=0.022）.Neutrophil（NEUT）and neutrophil to lymphocyte ratio（NLR）were significantly positively correlated with the occurrence of sICH.After adjusting the relevant confounding factors by multiple regression equations,it was found that for every increase in NEUT by 1×10~9/L,the risk of sICH increased by 31%（OR=1.31,95%CI=1.04-1.67,P=0.023）;Each increase in NLR increases the risk of sICH by 33%（OR=1.33,95%CI=1.12-1.58,P=0.001）.2.The predictive power of HAT,SITS,and THRIVE scores on the occurrence of sICH.The AUC of the HAT score for predicting the occurrence of sICH was 0.656（95%CI=0.529-0.782）,and the AIC was 158.14;the AUC of the SITS score for predicting the occurrence of sICH was 0.696（95%CI=0.593-0.799）,and the AIC was 159.13;The AUC of THRIVE score for predicting the occurrence of sICH was 0.715（95%CI=0.586-0.844）,and the AIC was 151.51.Based on the comparison of AUC and AIC values,the THRIVE score had the highest predictive power and the best goodness of fit.3.Construction and verification of sICH risk model.Univariate and multivariate regression analysis performed in the modeling group showed that THRIVE score,OTT,and NLR were independent predictors of sICH occurrence.The sICH risk model was constructed according to binary Logistic regression,Logit（sICH）=-9.96355+0.53155×THRIVE score+0.02353×OTT+0.21926×NLR.The discrimination and calibration of the sICH risk model were significantly higher than those of the THRIVE score,in the modeling group:AUC=0.881 vs 0.716,P=0.013,H-L test P=0.503 vs 0.290;and in the validation group:AUC=0.867 vs 0.714,P=0.032,H-L test P=0.674 vs 0.549.The calibration curve showed good agreement between the predicted probabilities of the sICH risk model and the actual observed probabilities.The comparison results of the sensitivity,specificity and decision curve of the two models showed that compared with the THRIVE score,the sICH risk model had higher clinical utility.The metaphase validation showed that the sICH risk model had good stability.Conclusions1.The incidence of clinical sICH was 4.33%.NIHSS score before thrombolysis,OTT,history of atrial fibrillation,HMCAS,NEUT and NLR were independent risk factors for sICH,and TG was a protective factor for sICH.It is suggested that clinical nurses need to strengthen the public health education,improve the patient’s disease identification ability and awareness of seeing a doctor,optimize the construction of specialized nursing teams,and reduce the delay time before thrombolysis;check whether there is atrial fibrillation in the acute stage of AIS attack,and be in the disease prevention stage.Actively treat the primary disease and patients with stable disease,give rational drug use guidance,improve patients’medication compliance,and prevent the recurrence of atrial fibrillation.2.Comparison of HAT,SITS and THRIVE scores,THRIVE score has the highest predictive performance and the best goodness of fit,but AUC<0.8,suggesting that it is necessary to combine other independent influencing factors of sICH to further optimize the score and build a new sICH risk model.3.The sICH risk model included THRIVE score,OTT and NLR,and its discrimination,calibration and clinical utility were better than the original THRIVE score.It had been verified that it had better prediction performance,indicating that the model had good universality.