老年非瓣膜性心房颤动患者发生血栓栓塞的列线图预测模型构建

Construction of nomogram model for thromboembolism in elderly patients with non-valvular atrial fibrillation

目的分析老年非瓣膜性心房颤动(NVAF)患者发生血栓栓塞的危险因素,构建列线图预测模型并探讨其预测价值。方法2021年1月—2022年12月新疆维吾尔自治区人民医院诊治老年NVAF患者1267例,根据是否发生血栓栓塞分为血栓栓塞组473例和非血栓栓塞组794例,比较2组临床资料、实验室指标及超声心动图指标。采用多因素logistic回归分析老年NVAF患者发生血栓栓塞的影响因素,根据影响因素构建预测老年NVAF患者发生血栓栓塞的列线图模型,绘制ROC曲线、校准曲线、临床决策曲线评估列线图模型预测老年NVAF患者发生血栓栓塞的效能。结果血栓栓塞组年龄[(76.45±8.28)岁]大于非血栓栓塞组[(73.38±8.58)岁(t=-6.241,P<0.001),CHA_(2)DS_(2)-VASc评分[(5.62±1.38)分]、HAS-BLED评分[(2.22±0.92)分]及合并高血压、冠心病、外周血管疾病、心功能不全、高同型半胱氨酸血症、高脂血症比率(75.05%、61.10%、84.14%、64.06%、32.56%、45.24%)均高于非血栓栓塞组(3.57±1.43)分、(1.22±1.16)分、61.21%、55.16%、70.03%、53.65%、11.96%、33.25%](t=-25.017、t=-16.918、χ^(2)=25.454、χ^(2)=4.270、χ^(2)=31.765、χ^(2)=13.140、χ^(2)=79.610、χ^(2)=18.149,P均<0.05),NVAF病程[3.00(1.00,7.00)年]长于非血栓栓塞组[2.00(1.00,5.00)年](U=-4.160,P<0.001)。血栓栓塞组中性粒细胞计数[4.59×10^(9)/L(3.28×10^(9)/L,6.38×10^(9)/L)]、国际标准化比值(1.27±0.47)、D-二聚体[0.78(0.39,1.83)mg/L]、C反应蛋白[4.63(1.71,17.38)mg/L]、中性粒细胞/淋巴细胞比值[3.52(2.28,6.64)]、脑钠肽[264.69(126.00,529.70)ng/L]、肌钙蛋白T[0.02(0.01,0.03)μg/L]、肌钙蛋白I[0.01(0.01,0.03)μg/L]及左心房内径>40 mm、左室射血分数≤50%比率(64.06%、55.39%)均高于非血栓栓塞组[4.16×10^(9)/L(3.33×10^(9)/L,5.54×10^(9)/L)、1.21±0.36、0.52(0.29,1.06)mg/L、3.70(1.57,11.31)mg/L、2.88(2.02,4.57)、202.25(91.73,462.43)ng/L、0.01(0.01,0.03)μg/L、0.01(0.00,0.02)μg/L、52.52%、39.29%](U=-2.650、t=-2.480、U=-6.580、U=-2.060、U=-4.650、U=-2.820、U=-1.990、U=-3.320、χ^(2)=16.090、χ^(2)=30.990,P均<0.05),凝血酶原时间[(15.55±5.16)s]长于非血栓栓塞组[(14.68±3.62)s](t=-3.510,P<0.001),淋巴细胞计数[1.27×10^(9)/L(0.86×10^(9)/L,1.69×10^(9)/L)]、白蛋白[(37.02±5.80)g/L]均低于非血栓栓塞组[1.40×10^(9)/L(1.00×10^(9)/L,1.80×10^(9)/L)、(37.68±5.38)g/L](U=-2.880,P=0.004;t=2.030,P=0.044)。年龄(OR=1.030,95%CI:1.010~1.050,P<0.001)、NVAF病程(OR=1.030,95%CI:1.010~1.050,P=0.002)、凝血酶原时间(OR=1.030,95%CI:1.010~1.060,P=0.042)、D-二聚体(OR=1.120,95%CI:1.060~1.190,P<0.001)、中性粒细胞/淋巴细胞比值(OR=1.040,95%CI:1.010~1.070,P=0.003)、左心房内径(OR=1.310,95%CI:1.010~1.700,P=0.046)、左室射血分数(OR=1.400,95%CI:1.070~1.840,P=0.015)、外周血管疾病(OR=2.930,95%CI:1.830~4.710,P<0.001)、高同型半胱氨酸血症(OR=3.090,95%CI:2.260~4.220,P<0.001)、高脂血症(OR=1.340,95%CI:1.030~1.750,P<0.001)、高血压(OR=1.350,95%CI:1.020~1.790,P<0.001)是老年NVAF患者发生血栓栓塞的影响因素。列线图模型以0.383为最佳截断值,预测老年NVAF患者发生血栓栓塞的AUC为0.750(95%CI:0.714~0.786,P<0.001),灵敏度为74.5%,特异度为64.4%。列线图模型校准曲线与理想曲线拟合较好,决策曲线显示列线图模型预测老年NVAF患者发生血栓栓塞有较高临床获益。结论年龄较大、NVAF病程及凝血酶原时间较长、D-二聚体和中性粒细胞/淋巴细胞比值较高、左心房内径>40 mm、左室射血分数≤50%及合并高血压、外周血管疾病、高同型半胱氨酸血症、高脂血症的老年NVAF患者发生血栓栓塞的风险较大,根据以上因素建立的列线图模型对老年NVAF患者发生血栓栓塞的预测价值较高。

Objective To analyze the risk factors of thromboembolism in elderly patients with non-valvular atrial fibrillation(NVAF),to construct a nomogram prediction model,and to explore its predictive value.Methods Totally 1267 elderly patients with NVAF were diagnosed and treated in People's Hospital of Xinjiang Uygur Autonomous Region from January 2021 to December 2022,and were divided into thromboembolism group(n=473)and non-thromboembolism group(n=794).The clinical data,laboratory indicators and echocardiographic indexes were compared between two groups.Multivariate logistic regression was used to analyze the influencing factors of thromboembolism in elderly patients with NVAF.According to the influencing factors,a nomogram model was constructed to predict thromboembolism in elderly patients with NVAF.ROC curve,calibration curve and clinical decision curve were drawn to evaluate the efficiency of nomogram model on predicting thromboembolism.Results The age was older,and the CHA_(2)DS_(2)-VASc,HAS-BLED score,and proportions of hypertension,coronary heart disease,peripheral vascular disease,cardiac insufficiency,hyperhomocysteinemia and hyperlipidemia were higher in thromboembolism group[(76.45±8.28)years,5.62±1.38,2.22±0.92,75.05%,61.10%,84.14%,64.06%,32.56%,45.24%]than those in non-thromboembolism group[(73.38±8.58)years,3.57±1.43,1.22±1.16,61.21%,55.16%,70.03%,53.65%,11.96%,33.25%](t=-6.241,t=-25.017,t=-16.918,χ^(2)=25.454,χ^(2)=4.270,χ^(2)=31.765,χ^(2)=13.140,χ^(2)=79.610,χ^(2)=18.149;all P values<0.05),the course of NVAF was longer in thromboembolism group[3.00(1.00,7.00)years]than that in non-thromboembolism group[2.00(1.00,5.00)years](U=-4.160,P<0.001).The neutrophil count,international standardized ratio,D-dimer,C-reactive protein,neutrophil/lymphocyte ratio,brain natriuretic peptide,troponin T,troponin I,and proportions of left atrial diameter>40 mm and left ventricular ejection fraction≤50% were higher in thromboembolism group[4.59×10^(9)/L(3.28×10^(9)/L,6.38×10^(9)/L),1.27±0.47,0.78(0.39,1.83)mg/L,4.63(1.71,17.38)mg/L,3.52(2.28,6.64),264.69(126.00,529.70)ng/L,0.02(0.01,0.03)μg/L,0.01(0.01,0.03)μg/L,64.06%,55.39%]than those in non-thromboembolism group[4.16×10^(9)/L(3.33×10^(9)/L,5.54×10^(9)/L),1.21±0.36,0.52(0.29,1.06)mg/L,3.70(1.57,11.31)mg/L,2.88(2.02,4.57),202.25(91.73,462.43)ng/L,0.01(0.01,0.03)μg/L,0.01(0.00,0.02)μg/L,52.52%,39.29%](U=-2.650,t=-2.480,U=-6.580,U=-2.060,U=-4.650,U=-2.820,U=-1.990,U=-3.320,χ^(2)=16.090,χ^(2)=30.990;all P values<0.05).The prothrombin time was longer in thromboembolism group[(15.55±5.16)s]than that in non-thromboembolism group[(14.68±3.62)s](t=-3.510,P<0.001).The lymphocyte count and albumin level were lower in thromboembolism group[1.27×10^(9)/L(0.86×10^(9)/L,1.69×10^(9)/L),(37.02±5.80)g/L]than those in non-thromboembolism group[1.40×10^(9)/L(1.00×10^(9)/L,1.80×10^(9)/L),(37.68±5.38)g/L](U=-2.880,P=0.004;t=2.030,P=0.044).Age(OR=1.030,95%CI:1.010-1.050,P<0.001),course of NVAF(OR=1.030,95%CI:1.010-1.050,P=0.002),prothrombin time(OR=1.030,95%CI:1.010-1.060,P=0.042),D-dimer(OR=1.120,95%CI:1.060-1.190,P<0.001),neutrophil/lymphocyte ratio(OR=1.040,95%CI:1.010-1.070,P=0.003),left atrial diameter(OR=1.310,95%CI:1.010-1.700,P=0.046),left ventricular ejection fraction(OR=1.400,95%CI:1.070-1.840,P=0.015),peripheral vascular disease(OR=2.930,95%CI:1.830-4.710,P<0.001),hyperhomocysteinemia(OR=3.090,95%CI:2.260-4.220,P<0.001),hyperlipidemia(OR=1.340,95%CI:1.030-1.750,P<0.001)and hypertension(OR=1.350,95%CI:1.020-1.790,P<0.001)were the influencing factors of thromboembolism in elderly patients with NVAF.When the optimal cut-off value of nomogram model was 0.383,the AUC for predicting thromboembolism was 0.750(95%CI:0.714-0.786,P<0.001),with a sensitivity of 74.5%and a specificity of 64.4%.The calibration curve of nomogram model fitted well with the ideal curve,and the decision curve showed that nomogram model had a high clinical benefit in predicting thromboembolism in elderly patients with NVAF.Conclusions The patients with NVAF are at a high risk of thromboembolism when they have an old age,long course of NVAF,long prothrombin time,high D-dimer level,high neutrophil/lymphocyte ratio,left atrial diameter of>40 mm,left ventricular ejection fraction of≤50%,and complications of hypertension,peripheral vascular disease,hyperhomocysteinemia and hyperlipidemia.The nomogram model based on the above factors has a high predictive value.