Systematic evaluation of predictive models for deep vein thrombosis risk in patients undergoing hip surgery

TAN Zhou; LI Chunmei; TENG Hui; CHEN Chen; LUO Xingyu; LIU Pingfang

doi:10.12173/j.issn.1004-5511.202501030

Abstract：Objective To systematically search and evaluate studies on deep vein thrombosis(DVT)risk prediction models for postoperative hip patients,to provide scientific references for the development and optimization of models and their clinical application.Methods Chinese and English databases were searched for studies on predictive models of DVT risk in postoperative hip patients,with a timeframe from the inception to 10 July 2025.Two investigators independently screened the literature and extracted the data,and the included studies were evaluated for risk of bias and applicability using the predictive models PRO BAST risk of bias evaluation tool.The predictive factors and performance,construction and validation of the prediction model were descriptively analyzed.Results A total of 17 studies involving 27 risk prediction models were included,with the number of predictors ranging from 4 to 10.Predictors that appeared more frequently included age,D-dimer level,bed rest time,comorbid diabetes mellitus,hypertension,time from injury to surgery,BMI,fibrinogen,plasminogen time,and activated partial thromboplastin time.The AUC of the included models ranged from 0.579 to 0.982;most of the models were internally validated,but none were externally validated.The 17 included studies had good overall applicability,but were generally at high risk of bias.Conclusion The predictive performance of existing models for predicting the risk of DVT in patients undergoing hip surgery is good,but many shortcomings remain.Future studies should focus on study design,enriching the way of variable screening,exploring and optimizing the modelling method,and strengthening the internal and external validation of the models.