脓毒症急性肾损伤短期预后个体化预测模型列线图的建立及验证

Establishment and analysis of the prognostic nomogram for sepsis-associated acute kidney injury in short-term

目的分析脓毒症并发急性肾损伤患者(SA-AKI)短期不良预后的影响因素,建立个体化预测SA-AKI患者短期不良预后风险的列线图模型。方法选取2017年1月1日至2019年6月1日重庆市梁平区人民医院急诊重症监护室收治的SA-AKI患者共363例为研究对象,通过电话或门诊复查等方式随访,并根据患者30 d内预后分为存活组(251例)和死亡组(112例)。采用单因素和多因素Logistic回归分析SA-AKI患者30 d死亡的影响因素,利用R软件建立预测SA-AKI短期不良预后的列线图模型。结果单因素分析显示年龄(Z=-3.268、P=0.005)、真菌感染(χ^2=12.318、P=0.002)、APACHEⅡ得分(t=9.587、P=0.001)、SOFA得分(t=6.328、P=0.001)、未及时诊断AKI(χ^2=12.587、P=0.001)、抗凝血酶Ⅲ(Z=-8.332、P=0.001)和AKI分期(Z=-7.009、P=0.001)7个因素可能是脓毒症急并发性肾损伤患者在30 d内死亡的相关危险因素。多因素Logistic回归分析显示:年龄(65~79岁:P=0.041、OR=3.573、95%CI:1.054~12.112;>79岁:P=0.028、OR=3.681、95%CI:1.215~11.984)、真菌感染(P=0.018、OR=4.684、95%CI:1.356~15.367)、APACHEⅡ评分(P=0.001、OR=1.265、95%CI:1.164~1.532)、SOFA评分(P=0.001、OR=1.291、95%CI:1.117~1.532)、抗凝血酶-Ⅲ(60%~80%:P=0.001、OR=1.291、95%CI:1.117~1.532;<60%:P=0.001、OR=15.329、95%CI:4.387~55.321)和AKI分期(2期:P=0.022、OR=2.998、95%CI:1.039~8.325;3期:P=0.001、OR=9.367、95%CI:5.235~22.327)均为SA-AKI患者30 d内死亡的独立影响因素。对列线图模型进行验证,其初始一致性指数(C-index)为0.943,经1000次的模型内部验证后一致性指数(C-index)为0.945,校正曲线显示该列线图模型具有良好的区分度及一致性,可为SA-AKI患者的预后评估提供针对性指导。结论基于年龄、真菌感染、APACHEⅡ评分、SOFA评分、抗凝血酶-Ⅲ和AKI分期6个因素为SA-AKI的独立影响因素,构建个体化预测SA患者并发KAI风险的列线图模型,可较为准确地预测SA-AKI患者短期不良预后发生的风险。

Objective To investigate the influencing factors of short-term adverse prognosis in patients with sepsis complicated with acute kidney injury(SA-AKI),and to establish a personalized line chart model to predict the short-term adverse prognosis risk of patients with SA-AKI.Methods Total of 363 patients with sepsis who were admitted to the Emergency Intensive Care Unit of Chongqing Liangping District People’s Hospital from January 1 st,2017 to June 1 st,2019 were collected;Follow-ups were conducted by telephone or outpatient review,and patients were divided into survival group(251 cases)and death group(112 cases)based on their prognosis within 30 days.Single-factor and multi-factor Logistic regression were used to analyze the influencing factors of SA-AKI patients’death within 30-day,and R software was used to establish a line chart model to predict the short-term adverse prognosis of SA-AKI.Results Univariate analysis showed that age(Z=-3.268,P=0.005),fungal infection(χ^2=12.318,P=0.002),APACHEⅡscore(t=9.587,P=0.001),SOFA score(t=6.328,P=0.001),AKI failure to diagnose in time(χ^2=12.587,P=0.001),antithrombinⅢ(Z=-8.332,P=0.001)and AKI(Z=-7.009,P=0.00)stage may be associated risk factors for death in patients with acute concurrent renal injury in sepsis within 30 d.Multivariate Logistic regression analysis showed that age(65-79 years old:P=0.041,OR=3.573,95%CI:1.054-12.112;>79 years old:P=0.028,OR=3.681,95%CI:1.215-11.984),fungal infection(P=0.018,OR=4.684,95%CI:1.356-15.367),APACHEⅡscore(P=0.001,OR=1.265,95%CI:1.164-1.532),SOFA score(P=0.001,OR=1.291,95%CI:1.117-1.532),antithrombin-Ⅲ(60%~80%:P=0.001,OR=1.291,95%CI:1.117-1.532;<60%:P=0.001,OR=15.329,95%CI:4.387-55.321),and AKI stage(2 period:P=0.022,OR=2.998,95%CI:1.039-8.325;3 period:P=0.001,OR=9.367,95%CI:5.235-22.327)were all independent related factors for death within 30 days of patients with SA-AKI.The collinear map model validation showed that its initial consistency index(C-index)was 0.943;After 1000 internal verifications of the model,the consistency index(C-index)was 0.945.The calibration curve showed that the collinear map model had a good differentiation and consistency,which could provide targeted guidance for prognosis evaluation of patients with SA-AKI.Conclusions This study was based on 6 independent influence factors including age,fungal infection,APACHEⅡscore,SOFA score,antithrombin-Ⅲand AKI stage,to construct an individualized line map model to predict the concurrent KAI risk of patients with SA,which could accurately predict the risk of short-term adverse prognosis of patients with SA-AKI.