脑卒中病人深静脉血栓风险预测模型的系统评价OA

Objective:To systematically evaluate the risk prediction models for deep vein thrombosis in stroke patients and provide reference for clinical practice.Methods:Comprehensively searched literatures on deep vein thrombosis risk prediction models in stroke patients from PubMed,the Cochrane Library,Embase,CINAHL,CNKI,VIP database,SinoMed,Wanfang Data Knowledge Service Platform and other databases.The retrieval period was from the establishment of the database to October 2023.Two researchers independently screened literature and extracted data.The bias risk assessment tool(PROBAST)for predictive model studies was used to analyze the risk of bias and applicability for literature.Results:A total of 18 models for predicting the risk of deep vein thrombosis in stroke patients were included.The AUC of the modeling cohort ranged from 0.610 to 0.912,and that of the validation cohort ranged from 0.550 to 0.840.Older age,D-dimer,female,malignant tumor,hemorrhagic stroke were the predictors of higher frequency.Conclusions:The development and application of deep vein thrombosis risk prediction model for stroke patients are still in the development stage.In the future,researchers can use different modeling methods and strictly follow the predictive model bias risk assessment tool to design the development and application of the model.The clinical staff should carefully select the corresponding prediction model for clinical application,identify the risk of deep vein thrombosis in stroke patients as early as possible and take corresponding preventive measures.