老年脓毒性休克患者进展为慢重症的列线图预测模型的构建及验证

目的 探讨老年脓毒性休克患者进展为慢重症的相关危险因素，并在此基础上构建与验证预测慢重症发生风险的列线图模型。方法 将纳入研究的252例年龄≥65岁的脓毒性休克患者作为训练集，并根据是否进展为慢重症将其分为慢重症组（86例）和非慢重症组（166例）。统计2组患者入EICU 24 h内一般资料、查尔森合并症指数（CCI）评分、序贯器官衰竭评估（SOFA）评分、腹内压（IAP）、机械通气（MV）和连续肾脏替代治疗（CRRT）比例以及血清乳酸（Lac）、降钙素原（PCT）水平的差异。通过多因素Logistic回归确定老年脓毒性休克患者进展为慢重症的独立危险因素，并以此构建预测慢重症发生风险的列线图模型。分别通过校准曲线和受试者工作特征（ROC）曲线验证模型的校准度和区分度，并采用决策曲线分析法（DCA）确定模型的临床实用性。另外选取74例老年脓毒性休克患者作为验证集对预测模型进行外部验证。结果 训练集老年脓毒性休克患者慢重症发生率为34.13%。与非慢重症组相比，慢重症组年龄≥75岁，CCI评分≥3分，CRRT比例、MV比例、SOFA评分、IAP水平较高（P<0.05）。多因素Logistic回归分析显示，CCI评分≥3分、SOFA评分升高、IAP升高、MV以及CRRT是老年脓毒性休克患者进展为慢重症的独立危险因素（P<0.05）。校准曲线显示，以上述5个因素构建的列线图预测模型在训练集和验证集中的预测概率与实际概率均接近，具有较好的校准度；ROC曲线显示，该模型在训练集和验证集中预测慢重症发生风险的曲线下面积分别为0.806（95%CI：0.750~0.862）和0.802（95%CI：0.697~0.908），具有较好的区分度；DCA曲线表明，该模型具有较好的临床实用性。结论 基于CCI评分、SOFA评分、IAP、MV和CRRT构建的列线图模型对老年脓毒性休克患者进展为慢重症风险具有良好的预测性能。

Construction and validation of a nomogram prediction model for the progression to chronic critical illness in elderly patients with septic shock

Objective To explore the related risk factors of the progression to chronic critical illness in elderly patients with septic shock, and to construct and verify a nomogram model to predict the risk of chronic critical illness based on the results. Methods A total of 252 patients with septic shock aged ≥ 65 years were enrolled in this study as the training set, and patients were divided into the chronic critical illness group (n=86) and the non-chronic critical illness group (n=166) according to whether they progressed to chronic critical illness. The data of general information, Charlson comorbidity index (CCI) score, sequential organ failure assessment (SOFA) score, intra-abdominal pressure (IAP), proportion of continuous renal replacement therapy (CRRT), mechanical ventilation (MV), serum levels of lactate (Lac) and procalcitonin (PCT) within 24 hours of entering EICU were analyzed in the two groups of patients. The independent risk factors for the progression of chronic critical illness in elderly patients with septic shock were identified by multivariate Logistic regression analysis, so as to establish a nomogram model based on the results to predict the risk of chronic critical illness. The calibration and discrimination of the model were evaluated by calibration curve and receiver operating characteristic (ROC) curve, respectively. The clinical practicability of the model was determined by decision curve analysis (DCA). In addition, 74 elderly patients with septic shock were selected as the verification set for external verification of the prediction model. Results The incidence of chronic critical illness in elderly patients with septic shock was 34.13% in the training set. Compared with the non-chronic critical illness group, the proportion of age≥75 years, CCI score≥3 points, MV, CRRT, SOFA score and IAP levels were higher in the chronic critical illness group (P<0.05). Multivariate Logistic regression analysis showed that CCI score≥3 points, elevated SOFA score, elevated IAP, MV and CRRT were independent risk factors for the progression to chronic critical illness in elderly patients with septic shock (P<0.05). The calibration curve showed that the nomogram prediction model constructed by the above five factors displayed good calibration in the training set and the verification set with the predicted probability closely to the actual probability. The ROC curve showed that the model displayed good discrimination in the training set and the verification set. The areas under the curve of predicting risks of chronic critical illness were 0.806 (95%CI: 0.750-0.862) and 0.802 (95%CI: 0.697-0.908), respectively. And the DCA curve demonstrated that the model had good clinical practicability. Conclusion The nomogram model based on CCI score, SOFA score, IAP, MV and CRRT shows good predicting performance in predicting the risk of progression to chronic critical illness in elderly patients with septic shock.