METHODS: A systematic search was conducted across ten databases from inception to October 20, 2023. Studies were eligible if they compared the accuracy of a RAM to that of another RAM for predicting the risk of PICC-RVT in adult patients with cancer. Two reviewers independently performed the study selection, data extraction and risk of bias assessments. A Bayesian network meta-analysis (NMA) was used to evaluate the performance of the RAMs.
RESULTS: A total of 1931 studies were screened, and 7 studies with 10 RAMs were included in the review. The most widely used RAMs were the Caprini (4 studies), Padua prediction score (3 studies), Autar (3 studies), Michigan risk score (2 studies) and Seeley score (2 studies). The sensitivity, specificity and accuracy varied markedly between the models. Notably, the Caprini score achieved higher sensitivity than 4 RAMs (Wells, Revised Geneva, modified MRS, MRS). The Michigan risk score had greater specificity than did the other 6 RAMs (Caprini, Autar, Padua, Seeley, the novel RAM, Wells). The predictive accuracy of the MRS is significantly greater than that of the Caprini and Autar RAM.
CONCLUSIONS: The MRS could be the most accurate RAM for identifying patients at high risk of PICC-RVT. However, as limited studies are available, more rigorous studies should be conducted to examine the accuracy of the Michigan risk score for PICC-RVT in different contexts.
方法:从开始到2023年10月20日,对10个数据库进行了系统搜索。如果将RAM的准确性与另一个RAM的准确性进行比较,以预测成年癌症患者的PICC-RVT风险,则研究合格。两名评审员独立进行研究选择,数据提取和偏见风险评估。使用贝叶斯网络荟萃分析(NMA)来评估RAM的性能。
结果:总共筛选了1931项研究,和7项10RAM的研究被纳入审查。最广泛使用的RAM是Caprini(4项研究),帕多瓦预测评分(3项研究),Autar(3项研究),密歇根风险评分(2项研究)和西利评分(2项研究)。敏感性,模型间的特异性和准确性差异显著。值得注意的是,Caprini评分比4RAM(Wells,订正日内瓦,修改后的MRS,MRS)。密歇根风险评分比其他6个RAM具有更大的特异性(Caprini,Autar,帕多瓦,Seeley,新颖的RAM,威尔斯)。MRS的预测准确性明显高于Caprini和AutarRAM。
结论:MRS可能是识别PICC-RVT高危患者最准确的RAM。然而,由于有限的研究可用,应进行更严格的研究,以检查不同情况下密歇根州PICC-RVT风险评分的准确性.