目的 探讨入住重症监护室的脓毒症相关谵妄(sepsis-associated delirium,SAD)患者预后的影响因素,并进一步构建列线图.方法 本研究为回顾性队列研究,收集医学信息市场重症监护数据库 MIMIC-Ⅲ(Medical Information Mart for Intensive Care Ⅲ,MIMIC-Ⅲ)数据库中符合脓毒症相关谵妄的患者作为建模组,以及嘉兴市第一医院2021年1月至2022年9月收治的符合脓毒症相关谵妄的患者作为验证组进行外部验证.纳入标准:(1)年龄≥ 18岁;(2)首次入住ICU;(3)住ICU时间＞24 h;(4)符合脓毒症诊断;(5)采用CAM-ICU工具进行谵妄筛查符合谵妄诊断.收集患者一般信息、生命体征、既往史及实验室检查等,将28 d死亡作为结局指标.采用多因素logistic回归确定独立影响因素,建立列线图模型,通过校准曲线、受试者工作曲线下面积(the area under the receiver operating characteristic curve,AUC)、决策曲线分析以及 Hosmer-Lemeshow检验对预测模型效度进行评价.结果 建模组共250例患者纳入研究,验证组共154例患者纳入研究.在建模组中,多因素Logistic回归显示,年龄(OR=1.057,95％CI:1.030～1.084,P＜0.001),呼吸频率(OR=1.117,95％CI:1.037～1.202,P=0.003),乳酸(OR=1.137,95％CI:1.011～1.279,P=0.032),血红蛋白(OR=0.983,95％CI:0.970～0.997,P=0.020),SOFA 评分(OR=1.184,95％CI:1.070～1.309,P=0.001)是脓毒症相关谵妄患者28 d死亡的独立影响因素.由上述5个因素构建的列线图AUC为0.773(95％CI:0.705～0.841),Hosmer-Lemeshow检验显示模型拟合良好(P=0.875),决策曲线分析显示,该模型的净获益较好.验证组的AUC为0.864(95％CI:0.799～0.928),Hosmer-Lemeshow检验显示模型拟合良好(P=0.488),决策曲线显示模型净获益较好.结论 本研究基于年龄、呼吸频率、乳酸、血红蛋白、SOFA评分构建的预测模型对脓毒症相关谵妄患者预后具有较好的预测价值.能够帮助临床医生早期识别脓毒症相关谵妄患者死亡的相关危险因素以进行早期干预.
Construction and validation of a prediction model for sepsis-associated delirium prognosis
Objective The study aimed to analyze the risk variables influencing the prognosis of patients with sepsis-associated delirium(SAD)in the Intensive Care Unit(ICU)and build a prediction nomogram.Methods This was a retrospective cohort study that includes patients with SAD in the Medical Information Mart for Intensive Care Ⅲ database(MIMIC-Ⅲ)database as training cohort,and patients who were hospitalized in the First Hospital of Jiaxing from January 2021 to September 2022 as validation cohort.Inclusion criteria:(1)age＞18 years old;(2)being admitted to the ICU for the first time;(3)the length of ICU stay＞24 h;(4)consistent with the diagnosis of sepsis;(5)the diagnosis of delirium was identified by CAM-ICU questionnaire.The general information,vital signs,past medical history and laboratory examination results of the patients were collected,and the outcome was 28-day mortality.Multiple logistic regression was used to identify independent influencing factors and the nomogram was constructed.The validity of the prediction model was determined using multiple indicators,including calibration curve,the area under the receiver operating characteristic curve(AUC),decision curve analysis(DCA),and Hosmer-Lemeshow test.Results A total of 250 patients were included in the training cohort and 154 patients were in the validation cohort.The multiple logistic regression demonstrated that age(OR=1.057,95％CI:1.030-1.084,P＜0.001),respiratory frequency(OR=1.117,95％CI:1.037-1.202,P=0.003),lactic acid(OR=1.137,95％CI:1.011-1.279,P=0,032),hemoglobin(OR=0.983,95％CI:0.970-0.997,P=0.020),SOFA score(OR=1.184,95％CI:1.070-1.309,P=0.001)were independent risk factors associated with the 28-day mortality of patients with SAD.The AUC of the nomogram created by the five factors above was 0.773(95％CI:0.705-0.841),and the Hosmer-Lemeshow test showed that the model was a good fit(P=0.875).The DCA curve indicates that the model has potential net benefit.The AUC was 0.864(95％CI:0.799-0.928)in the validation cohort,and the Hosmer-Lemeshow test showed that the model was a good fit(P=0.488).The DCA curve indicates that the model of the validation cohort had potential net benefit.Conclusion The prediction model based on age,respiratory frequency,lactate,hemoglobin,and SOFA scores shows valuable capability of predicting the prognosis of patients with SAD,which could help clinicians identify risk factors at first time and make earlier intervention.
Sepsis associated deliriumMIMIC-ⅢPrognosisPrediction model
万方数据
维普
