脓毒症患者血流感染列线图预测模型的建立与验证

Establishment and validation of a nomogram model for predicting bloodstream infection in sepsis patients

目的构建急诊脓毒症患者血流感染的列线图预测模型。方法收集2017年至2020年宁夏医科大学总医院急诊科就诊的313例脓毒症患者的资料,随机按7∶3的比例分为建模组(217例)和验证组(96例),比较两组患者的临床资料。建模组中根据是否存在血流感染分为血流感染组(91例)与非血流感染组(126例),纳入单因素及多因素Logistic回归分析得出的影响因素建立脓毒症患者血流感染的预测模型。预测模型的区分度、校准度和临床有效性通过受试者工作特征(ROC)曲线下面积(AUC)、校正曲线和决策曲线分析(DCA)进行验证。结果建模组和验证组患者的临床资料具有可比性(P>0.05),进一步回归分析结果显示,降钙素原(PCT)(OR=3.217,95%CI 1.895~5.460)、SOFA评分(OR=1.531,95%CI 1.290~1.817)是急诊脓毒症患者血流感染的独立影响因素(P<0.05)。建模组列线图预测模型的AUC为0.873(95%CI 0.826~0.920),验证组列线图预测模型的AUC为0.920(95%CI 0.870~0.971)。两组患者拟合优度检验较好(P>0.05),DCA证实,列线图预测模型的临床实用性良好。结论本研究基于PCT和SOFA评分构建的列线图预测模型能够早期预测急诊脓毒症血流感染的高风险患者。

Objective To construct a nomogram model for predicting bloodstream infection in sepsis patients in the emergency room.Methods The data of 313 sepsis patients who visited the emergency room of the General Hospital of Ningxia Medical University from 2017 to 2020 were collected.313 sepsis patients were randomly divided into a modeling group(217 cases)and a verification group(96 cases)according to the ratio of 7∶3,and the clinical data of the two groups were compared.The modeling group was divided into a bloodstream infection group(91 cases)and a non-bloodstream infection group(126 cases)according to the presence or absence of bloodstream infection,and the influencing factors obtained by univariate and multivariate Logistic regression analysis were included and a prediction model of sepsis patients with bloodstream infection was established.The differentiation,calibration,and clinical validity of the prediction model were validated by the area under the receiver operating characteristic(ROC)curve(AUC),correction curve,and decision curve analysis(DCA).Results The clinical data of the two groups were comparable(P>0.05),and further regression analysis showed that procalcitonin(PCT)(OR=3.217,95%CI 1.895-5.460)and SOFA score(OR=1.531,95%CI 1.290-1.817)were independent influencing factors in sepsis patients with bloodstream infection in the emergency room(P<0.05).The AUC of nomogram model in the modeling group was 0.873(95%CI 0.826-0.920),and the AUC of nomogram model in the verification group was 0.920(95%CI 0.870-0.971).The goodness-of-fit test of the two groups was good(P>0.05),and the DCA confirmed that the clinical usability of the nomogram model was good.Conclusions The nomogram model constructed in this study can predict high-risk patients of sepsis with bloodstream infection in the emergency room early.