Developing A Prediction Model Of Neurological Functional Outcome In Patients With Acute Ischemic Stroke At Discharge

Background and Objective:Acute ischemic stroke（AIS）,an acute cerebrovascular disease with high morbidity,mortality and disability rate,is one of the most important issues concerning global public health,which causes a heavy economic burden for society.So far there is no common consensus on neurological outcome prediction model.Our study aimed to investigate predictive factors associated with poor neurological functional outcome of AIS patients at discharge,and to verify predictive power of the prediction model via external verification.Methods:Clinical data（including gender,age,admission date,length of hospital stay,smoke history,alcoholism,history of chronic diseases,blood test at admission,imaging findings and NIHSS at admission）were collected retrospectively.Patients from January,1^st,2015 to December,31^st,2018 were put into training cohort and patients from January,1^st,2019 to December,31^st,2019 were put into validation cohort.Predictive variables were selected by univariate logistic regression analysis in the training cohort,which were then put into multivariate logistic regression analysis to find the independent risk factors and to build multivariate logistic regression model and nomogram.Discrimination of the prediction model was evaluated by Area under the curve（AUC）of receiver operating characteristic（ROC）,And calibration of the prediction model was evaluated by Hosmer-Lemeshow（H-L）test.The performance of the prediction model was then evaluated by using the validation cohort to calculate its AUC,cutpoint value,sensitivity,and specificty.Results:Clinical data of 2131 AIS patients in Shenzhen second hospital were collected from January,1st,2015 to December,31st,2019.In 1410 AIS patients included in training cohort,the average discharge day was 10.73±6.29（days）,the average age was 64.27±13.09（years）,male 63.97%and female 36.03%.In 721 AIS patients included in training cohort,the average discharge day was 9.48±3.41（days）,the average age was 6 64.92±12.51（years）,male 62.83%and female 37.17%.Poor neurological functional outcome were identified in 349 AIS patients at discharge in training cohort.After monovariate analysis,NIHSS at admission（OR=2.42 95%CI2.17-2.69）,history of high blood pressure（OR=1.41 95%CI 1.09-1.83）,diabetes（OR=1.5 95%CI 1.16-1.94）,atrial fibrillation（OR=2.07 95%CI 1.31-3.26）,neutrophil（OR=1.32 95%CI 1.24-1.39）,D-dimer（OR=1.38 95%CI 1.22-1.57）and together of 25 indicators were confirmed to be the predictive factors for neurological functional outcome at discharge.After multivariable bivariate logistic regression analysis of these factors,the prediction model of neurological functional outcome were established by NIHSS at admission（OR=2.396 95%CI 2.177-2.696）,blood urea nitrogen（OR=1.064,95%CI 1.017-1.114）,uric acid（OR=0.997 95%CI 0.996-0.999）,glycosylated hemoglobin（OR=1.185,95%CI 1.103-1.274）.AUC for the training and validation cohort were 0.947,0.935 respectively.The ROC curve of training and validation cohort were similar,implicating the prediction model still has strong discriminating power in the validation cohort.The p-value of H-L test in training and validation cohort were 0.507（>0.05）and 0.348（>0.05）respectively,indicating good calibration for the prediction model.Conclusions:BUN,glycosylated hemoglobin and NIHSS at admission are risk factors for poor neurological functional outcome at discharge,and uric acid is the protective factor.The prediction model of neurological functional outcome in patients with AIS at discharge is logit（p）=-5.69025+0.06738*BUN-0.00210*uric acid+0.16768*glycosylated hemoglobin+0.88488*NIHSS at admission.It has good predictive value and can be used to aid clinical intervention.