Development And Validation Of A Prediction Model For Cerebral Cardiac Syndrome

Research Purpose:To explore the risk factors related to Cerebral Cardiac Syndrome(CCS).To develop a risk prediction model for CCS.And to provide an evaluation tool for clinical screening of high-risk groups of CCS.Research Methods:(1)Retrospective modeling part: The medical records for cases of 318 stroke patients in 4tertiary A hospitals from June 2018 to April 2019 were reviewed.Patients were divided into CCS group and non-CCS group according to whether the CCS occurred or not.Univariate and multivariate logistic regression were performed to find associated factors with CCS.Furthermore,we established a multi-index prediction model and a simple risk assessment scale based on predicting risk of CCS.(2)Prospective verification part: 118 stroke patients enrolled onto prospective study served as independent validation cohorts.We further validated the accuracy of our model prediction and the score scale using validation cohort in the real clinical nursing work Research Results:(1)The incidence of CCS in this derivation cohort of 318 patients was 53.5%.There are 9independent risk factors for CCS: Age,NIHSS score,B-type Natriuretic Peptide(BNP),Neutrophils,and Prothrombin time(PT),Activated Partial Thromboplastin Time(APTT),Lactate Dehydrogenase(LDH),Blood Glucose and Carotid Stenosis Rate.(2)The H-L test of the risk prediction model was p=0.257,the area under the ROC curve was0.920,the sensitivity was 79.5%,the specificity was 91.3%,the Youden index was 0.871,and the clinically verified accuracy was 72.88%.(3)The clinical validation accuracy of the simple risk score scale was 71.19%.The scale and model prediction results were tested for consistency Kappa=0.928.Research Conclusion:The risk prediction model and simple risk score scale had advantages for CCS occurrence prediction,and as a tool to provide support for screening high-risk groups of CCS for clinical nursing staff.