PICC相关性血栓风险预测模型的系统评价

Risk prediction models for peripherally inserted central catheter⁃related thrombosis:a systematic review

目的:系统检索国内外经外周置入中心静脉导管(PICC)相关性血栓风险预测模型的研究并对其基本特征、建模过程、预测性能和验证方法等进行分析和评价。方法:在PubMed、EMbase、Web of Science、万方、维普和知网等数据库检索相关研究,由2名研究者独立进行文献筛选和文献资料提取,并根据PROBAST工具对纳入研究进行质量评价。结果:共纳入13个PICC相关性血栓风险预测模型,建模人群以肿瘤病人为主(n=7);11个模型报道了其预测性能,受试者工作特征(ROC)曲线下面积(AUC)值或C指数为0.64~0.89;5个模型进行了内部验证,4个模型进行了外部验证。质量评价结果为1个低偏倚风险和12个高偏倚风险。预测模型中最常见且赋值较高的预测因子包括化疗史、血栓史、糖尿病、肥胖或体质指数(BMI)≥25 kg/m2、导管尖端位置异位/非最佳位置及白细胞计数升高等。结论:现有研究表明,PICC相关性血栓风险预测模型预测准确度较高,但整体研究质量偏低,仍需从模型构建过程、模型验证和结果报告等方面进一步优化,以便为临床医务工作者提供可早期识别PICC相关性血栓高危人群的可靠的、科学的评估工具。

Objective:To systematically retrieved domestic and foreign researches on PICC-related thrombosis risk prediction models,and to analyze and evaluate their basic characteristics,development process,prediction performance and verification methods.Methods:Relevant studies were retrieved from databases including PubMed,EMbase,Web of Science,Wanfang,VIP,and CNKI.Two reviewers independently screened and extracted information.The quality of included studies was evaluated according to PROBAST tool.Results:A total of 13 studies were included,the modeling population was mainly tumor patients(n=7);11 studies evaluated the predictive performance of the model,and the AUC value or C index ranged from 0.64 to 0.89;only five and four models were internally and externally validated retrospectively.The quality appraisal results were 1 with low risk of bias,12 with high risk of bias.The most frequently and highly valued predictors were history of chemotherapy,personal history of thrombosis,diabetes,obesity or body mass index≥25 kg/m2,malposition of catheter tip,increased platelet count.Conclusion:The existing studies showed that the prediction accuracy of the included PICC⁃related thrombosis risk prediction models is high,but the overall research quality is low.It still needs to be further optimized from the aspects of model construction process,model verification and result report,so as to provide clinical medical staff with a reliable and scientific evaluation tool that can identify the high⁃risk groups of PICC⁃related thrombosis at an early stage.