脓毒症患者早期发生急性肾损伤的预测模型构建与内部验证

Construction and internal validation of a predictive model for early acute kidney injury in patients with sepsis

目的构建预测重症监护室(intensive care unit,ICU)脓毒症患者发生急性肾损伤(acute kidney injury,AKI)的列线图模型,并验证其早期预测的有效性。方法回顾性纳入2015年4月至2021年12月入住济宁医学院附属医院ICU的脓毒症患者,将符合纳入标准的患者以7∶3的比例随机分为训练集和验证集。采用单因素和多因素Logistic回归分析脓毒症患者发生AKI的影响因素并建立列线图模型,通过校正曲线、受试者工作曲线(receiver operating characteristic curve,ROC)以及决策曲线分析(decision curve analysis,DCA)对模型进行评价。结果741例脓毒症患者纳入研究,其中335例入ICU 7 d内发生AKI,AKI发生率为45.1%。随机分为训练集(n=519)和内部验证集(n=222)。多因素分析发现急性生理学与慢性健康状况评分Ⅱ、序贯器官衰竭评分、血乳酸、降钙素原、去甲肾上腺素使用剂量、尿素氮、中性粒细胞百分比是发生AKI的独立影响因素,结合以上变量构建列线图绘制列线图。在训练集中,列线图AUC为0.875(95%CI:0.767~0.835),校正曲线显示其预测概率和实际概率具有一致性,DCA显示其具有良好的临床净获益。在内部验证集中,列线图对AKI具有相似的预测价值(AUC=0.871,95%CI:0.734~0.854)。结论基于入室的危重症评分联合炎性标志物构建的列线图可用于ICU脓毒症患者发生AKI的早期预测,帮助临床医师早期识别脓毒症患者发生AKI。

Objective To construct a nomogram model predicting the occurrence of acute kidney injury(AKI)in patients with sepsis in the intensive care unit(ICU),and to verify its validity for early prediction.Methods Sepsis patients admitted to the ICU of the Affiliated Hospital of Jining Medical University from April 2015 to December 2021 were retrospectively included,and those who met the inclusion criteria were randomly divided into training and validation sets at a ratio of 7:3.Univariate and multivariate logistic regression models were used to identify independent risk factors for AKI in patients with sepsis,and a nomogram was constructed based on the independent risk factors.Calibration curve,receiver operating characteristic(ROC)curve,and decision curve analysis(DCA)were used to evaluate the nomogram model.Results 741 patients with sepsis were included in the study,335 patients developed AKI within 7 d of ICU admission,with an AKI incidence of 45.1%.Randomization was performed in the training set(n=519)and internal validation set(n=222).Multivariate logistic analysis revealed that acute physiology and chronic health status scoreⅡ,sequential organ failure score,serum lactate,calcitoninogen,norepinephrine dose,urea nitrogen,and neutrophil percentage were independent factors influencing the occurrence of AKI,and a nomogram model was constructed by combining these variables.In the training set,the AUC of the nomogram model ROC was 0.875(95%CI:0.767-0.835),the calibration curve showed consistency between the predicted and actual probabilities,and the DCA showed a good net clinical benefit.In the internal validation set,the nomogram model had a similar predictive value for AKI(AUC=0.871,95%CI:0.734-0.854).Conclusions A nomogram model constructed based on the critical care score at admission combined with inflammatory markers can be used for the early prediction of AKI in sepsis patients in the ICU.The model is helpful for clinicians early identify AKI in sepsis patients.