妊娠期糖尿病产妇发生产后下肢深静脉血栓形成的危险因素及列线图模型构建

Influencing factors of postpartum lower limb deep vein thrombosis in puerpera with gestational diabetes mellitus and construction of nomogram model

目的分析妊娠期糖尿病(GDM)产妇发生产后下肢深静脉血栓形成(DVT)的危险因素并构建列线图模型。方法回顾性分析滕州市中心人民医院2021年1月至2023年1月163例GDM产妇的临床资料。产妇随访2个月,其中发生产后下肢DVT 78例(DVT组),未发生产后下肢DVT 85例(非DVT组)。记录产妇的基本资料、围生期指标,采用免疫比浊法检测血清D-二聚体水平,酶联免疫吸附法检测血清C反应蛋白(CRP)水平。采用多因素Logistic回归分析影响GDM产妇发生产后下肢DVT的独立危险因素。采用R3.6.3软件及"rms"包构建预测GDM产妇发生产后下肢DVT的列线图模型;利用Hosmer-Lemeshow拟合优度检验和校准曲线评估列线图模型的一致性,受试者工作特征(ROC)曲线评估模型的区分度。结果DVT组孕前体质量指数≥25 kg/m2、高血压史、剖宫产、产后卧床时间≥2 d、产后血糖控制不良、下肢静脉曲张史比例及D-二聚体明显高于非DVT组[44.87%(35/78)比27.06%(23/85)、39.74%(31/78)比18.82%(16/85)、43.59%(34/78)比25.88%(22/85)、34.62%(27/78)比14.12%(12/85)、44.87%(35/78)比20.00%(17/85)、35.90%(28/78)比16.47%(14/85)和(582.74±72.42)μg/L比(462.39±57.65)μg/L],差异有统计学意义(P<0.05或<0.01);两组年龄、吸烟史、饮酒史、生育史、下肢患侧、下肢疼痛、下肢肿胀和CRP比较差异无统计学意义(P>0.05)。多因素Logistic回归分析结果显示,孕前体质量指数≥25 kg/m2、高血压史、剖宫产、产后卧床时间≥2 d、产后血糖控制不良、下肢静脉曲张史和D-二聚体高水平是影响GDM产妇发生产后下肢DVT的独立危险因素(OR=3.267、3.464、5.078、3.346、3.174、6.111和1.027,95%CI 1.171~9.273、1.183~10.146、1.737~14.843、1.064~10.523、1.737~14.843、1.727~21.629和1.017~1.036,P<0.05或<0.01)。以孕前体质量指数、高血压史、剖宫产、产后卧床时间、产后血糖控制、下肢静脉曲张史和D-二聚体构建了列线图模型。ROC曲线分析结果显示,此列线图模型区分度较优(曲线下面积为0.930,95%CI 0.880~0.964);校准曲线预测值与实际值基本一致,且Hosmer-Lemeshow拟合优度检验一致性较好(χ^(2)=10.00,P=0.265)。结论基于GDM产妇发生产后下肢DVT危险因素构建的列线图模型一致性和区分度较好,可有效预测GDM产妇产后下肢DVT的发生。

Objective To analyze the risk factors of postpartum lower limb deep vein thrombosis(DVT)in puerpera with gestational diabetes mellitus(GDM),and to construct a nomogram model.Methods The clinical data of 163 puerpera with GDM in Tengzhou Central People′s Hospital from January 2021 to January 2023 were analyzed retrospectively.The puerpera were followed up for 2 months,in which 78 cases had postpartum lower limb DVT(DVT group)and 85 cases did not have postpartum lower limb DVT(non-DVT group).The basic data and perinatal indexes were recorded.The serum D-dimer level was detected by immunoturbidimetry,and serum C-reactive protein(CRP)level was detected by enzyme-linked immunosorbent assay.Multivariate Logistic regression was used to analyze the independent risk factors of lower limb DVT in puerpera with GDM.R3.6.3 software and"rms"package were used to construct a nomogram model for predicting lower limb DVT in puerpera with GDM,Hosmer-Lemeshow goodness of fit test and calibration curve were applied to evaluate the consistency of nomogram model,and receiver operating characteristic(ROC)curve was used to evaluate the discrimination of model.Results The rates of body mass index before pregnancy≥25 kg/m 2,history of hypertension,caesarean section,postpartum bed rest time≥2 d,postpartum poor blood glucose control,history of lower limb varicose veins and D-dimer in DVT group were significantly higher than those in non-DVT group:44.87%(35/78)vs.27.06%(23/85),39.74%(31/78)vs.18.82%(16/85),43.59%(34/78)vs.25.88%(22/85),34.62%(27/78)vs.14.12%(12/85),44.87%(35/78)vs.20.00%(17/85),35.90%(28/78)vs.16.47%(14/85)and(582.74±72.42)μg/L vs.(462.39±57.65)μg/L,and there were statistical differences(P<0.05 or<0.01);there were no statistical differences in age,smoking history,drinking history,birth history,lower limb lesional side,lower limb pain,lower limb swelling and CRP between the two groups(P>0.05).Multivariate Logistic regression analysis result showed that body mass index before pregnancy≥25 kg/m2,history of hypertension,caesarean section,postpartum bed rest time≥2 d,poor postpartum blood glucose control,history of lower limb varicose veins and high level of D-dimer were the independent risk factors of lower limb DVT in puerperal with GDM(OR=3.267,3.464,5.078,3.346,3.174,6.111 and 1.027;95%CI 1.171 to 9.273,1.183 to 10.146,1.737 to 14.843,1.064 to 10.523,1.737 to 14.843,1.727 to 21.629 and 1.017 to 1.036;P<0.05 or<0.01).A nomogram model was constructed based on body mass index before pregnancy,history of hypertension,caesarean section,postpartum bed rest time,postpartum blood glucose control,history of lower limb varicose veins and D-dimer.The ROC curve analysis result show that the nomogram model had good discrimination(the area under the curve was 0.930,95%CI 0.880 to 0.964).The predicted value of calibration curve was basically consistent with the actual value,and the goodness of fit test of Hosmer-Lemeshow was in good agreement(χ^(2)=10.00,P=0.265).Conclusions The nomogram model based on the risk factors of postpartum lower limb DVT in puerperal with GDM has good consistency and discrimination,and it can effectively predict the occurrence of postpartum lower libm DVT in puerpera with GDM.