Abstract: Background　Venous thromboembolism（VTE） is a common underrated vascular disease，but clinical detection methods and risk assessment for it are used insufficiently.Mean platelet volume （MPV） may be one of the important biomarkers for deep venous thrombosis （DVT）.Both Padua prediction score and MPV are easily measured parameters with low cost in primary hospitals，and it is of clinical significance to study their combination predictive value for VTE in non-surgical inpatients with D-dimer >500μg/L.Objective　To evaluate the combination predictive value of Padua prediction score and MPV for VTE in non-surgical inpatients with D-dimer >500 μg/L.Methods　A retrospective analysis was performed on the non-surgical inpatients with highly suspected VTE and D-dimer >500 μg/L selected from the Second People's Hospital of Hefei from February 2017 to April 2019.All of them underwent CT pulmonary angiography or compression ultrasound or venography of the veins in both upper （lower） limbs.Patients were divided into VTE group and non VTE group according to the study criteria.Baseline data were collected，including age，gender，heart rate，D-dimer，Padua prediction score，platelet，MPV，platelet distribution width，platelet thrombocytocrit and mean red blood cell volume.For identifying the risk factors for VTE，baseline parameters were analyzed with univariate Logistic regression initially，then the identified variables with P value less than 0.1 were further analyzed with multivariate Logistic regression.ROC curve analysis was carried out to assess the predictive value of Padua prediction score，MPV，and the combination of the two for VTE.Results　There were statistically significant differences in mean Padua prediction score and MPV between VTE and non-VTE groups （P<0.01）.Multivariate Logistic regression analysis showed that Padua prediction score〔OR=1.930，95%CI（1.326，2.809），P=0.001〕and MPV 〔OR=2.566，95%CI（1.482，4.444），P=0.001〕were the influencing factors for VTE.For predicting VTE，the AUC of Padua prediction score was 0.723 〔95%CI （0.607，0.839） 〕，with 4.5 for optimal cut-off value，0.600 for sensitivity，0.815 for specificity，and 0.41 for Youden index.The AUC predicted by MPV was 0.771 〔95%CI （0.661，0.881） 〕，with 10.35 fl for optimal cut-off value，0.655 for sensitivity，0.852 for specificity，and 0.51 for Youden index.When Padua prediction score ≥4.5 and MPV ≥10.35 fl，the sensitivity was 0.350，the specificity was 1.000，and the Youden index was 0.35.When Padua prediction score ≥4.5 or MPV ≥10.35 fl，the prediction sensitivity was 0.910，with 0.700 for specificity，and 0.61 for Youden index.Analysis with Z scoring model developed based on the multivariate Logistic regression model using the algorithm of Z=-22+ Padua prediction score+ 2×MPV showed that the AUC of the Z score was 0.851〔95%CI （0.754，0.947） 〕，and when Z ≥1.90，the Youden index was 0.63.Conclusion　 Both Padua prediction score and MPV are associated with VTE in non-surgical inpatients with D-dimer >500μg/L，but each has unsatisfactory predictive value.The Z scoring model developed based on the multivariate Logistic regression model proved that the predictive value of the combination of the two may be higher.Padua prediction score ≥4.5 in combination with MPV ≥10.35 fl has high specificity，which may be used as a clinical method for the diagnosis of VTE.

Key words: Venous thromboembolism, Venous thrombosis, Thromboembolism, Risk assessment, Mean platelet volume, D-dimer

摘要： 背景　静脉血栓栓塞症（VTE）是常见的易被低估的血管性疾病，临床实际工作中发现现有的检测方式以及风险评估效果有限。平均血小板体积（MPV）可能是深静脉血栓（DVT）重要的生物标志物之一。Padua评分和MPV均为基层医院廉价、易获得的指标，研究两项指标对D-二聚体>500 μg/L的非手术住院患者发生VTE的预测价值具有重要临床意义。目的　研究Padua评分联合MPV对D-二聚体>500 μg/L的非手术住院患者发生VTE的预测价值。方法　回顾性分析合肥市第二人民医院2017年2月—2019年4月收治的临床高度疑似VTE且D-二聚体>500 μg/L的非手术住院患者，完善CT肺动脉造影（CTPA）或双下（上）肢加压深静脉彩超或深静脉造影检查。按照研究标准将患者分为VTE组和非VTE组。收集两组患者的基线资料，包括年龄、性别、心率（HR）、D-二聚体、Padua评分、血小板计数（PLT）、MPV、血小板分布宽度（PDW）、血小板压积（PCT）、平均红细胞体积（MCV）。比较两组患者的基线资料。将基线资料纳入单因素Logistic回归分析，筛选D-二聚体>500 μg/L的非手术住院患者发生VTE的影响因素，将Logistic回归分析结果中P<0.1的变量纳入多因素Logistc回归分析进一步筛选其影响因素。采用ROC曲线评估各指标对D-二聚体>500 μg/L的非手术住院患者发生VTE的预测价值。结果　VTE组和非VTE组基线资料中Padua评分、MPV比较，差异有统计学意义（P<0.01）。多因素Logistic回归分析结果显示，Padua评分〔OR=1.930，95%CI（1.326，2.809），P=0.001〕、MPV〔OR=2.566，95%CI（1.482，4.444），P=0.001〕是D-二聚体>500 μg/L的非手术住院患者发生VTE的影响因素。对于D-二聚体>500 μg/L的非手术住院患者发生VTE的情况，Padua评分对其预测的ROC曲线下面积（AUC）为0.723〔95%CI（0.607，0.839）〕，最佳截断值≥4.5分，灵敏度为0.600，特异度为0.815，约登指数为0.41；MPV对其预测的AUC为0.771〔95%CI（0.661，0.881）〕，最佳截断值≥10.35 fl，灵敏度为0.655，特异度为0.852，约登指数为0.51；Padua评分≥4.5分并且MPV≥10.35 fl对其预测的灵敏度为0.350，特异度为1.000，约登指数为0.35；当Padua评分≥4.5分或MPV≥10.35 fl对其预测的灵敏度为0.910，特异度为0.700，约登指数为0.61。利用多因素Logistic回归模型构建了预测D-二聚体>500 μg/L的非手术住院患者发生VTE的评分模型（Z评分模型），Z=-22+Padua评分+2×MPV。结果显示，对于D-二聚体>500 μg/L的非手术住院患者发生VTE的情况，Z评分对其预测的AUC=0.851，95%CI（0.754，0.947），Z≥1.90时，约登指数为0.63。结论　Padua评分、MPV是D-二聚体>500 μg/L的非手术住院患者发生VTE的危险因素。Padua评分、MPV单独指标对发生病症的临床预测价值有限。利用多因素Logistic回归模型构建的Z评分模型可以有效提高对该类病症的预测价值。Padua评分≥4.5分且MPV≥10.35 fl联合诊断，特异度高，可能成为一种诊断非VTE患者的临床手段。

关键词: 静脉血栓栓塞, 静脉血栓形成, 危险性评估, 平均血小板体积, D-二聚体