|
| 脓毒症急性胃肠损伤28d死亡风险列线图模型的建立与验证 |
| Development and validation of a nomogram model for risk of 28 -day mortality in septic acute gastrointestinal injury |
| 投稿时间：2022-03-07 |
| DOI：10.3969/j.issn.1000-0399.2022.08.002 |
| 中文关键词: 脓毒症 急性,胃肠损伤 列线图模型 预后(死亡风险) |
| 英文关键词: Sepsis Acute gastrointestinal injury Nomogram model Prognosis(mortality risk) |
| 基金项目:安徽省高校自然科学研究项目(项目编号:KJ2021ZD0023) |
|
| 摘要点击次数: 815 |
| 全文下载次数: 338 |
| 中文摘要: |
| 目的 建立预测脓毒症急性胃肠损伤(AGI)患者28 d死亡风险的列线图(Nomogram)模型并进行验证。方法 回顾性分析2019年10月至2021年3月于安徽医科大学第一附属医院重症医学科收治的169例脓毒症AGI患者的临床资料,依据患者28 d临床结局分为生存组(96例)和死亡组(73例)。应用LASSO回归筛选最优预测变量,纳入多因素logistic回归分析建立预测模型。通过Bootstrap重复抽样法对模型的区分度、校准度、临床有效性等进行内部验证。采用受试者工作特征曲线(ROC)曲线下面积(AUC)评估预测效能。回顾性收集2021年4月至2021年11月于安徽医科大学第一附属医院重症医学科收治的73例脓毒症AGI患者的相关资料进行模型外部验证。应用R软件绘制Nomogram图。结果 ①单因素分析:C-反应蛋白(CRP)、pH值、乳酸(Lac)、平均动脉压(MAP)、血管活性药物使用比率、急性生理与慢性健康状况Ⅱ评分(APACHE Ⅱ)、序贯器官衰竭评估(SOFA)评分、脓毒性休克比率以及AGI分级等资料在生存组和死亡组之间的差异均具有统计学意义(P<0.05)。②LASSO回归筛选出4个最优预测变量,包括Lac、使用血管活性药物、APACHE Ⅱ评分和AGI分级,依此建立logistic预测模型。③预测模型的内部验证和外部验证:一致性指数(C-index)分别为0.917和0.881,校正C-index分别为0.904和0.841,提示区分度良好;Hosmer-Lemeshow检验均显示P>0.05,校准曲线的斜率均为1,Brier值为0.118和0.132,说明校准度较好;决策分析(DCA)曲线阈值概率在>1%和6%～89%之间显示出较好的临床有效性。④预测模型的AUC为0.913(95% CI:0.870～0.955),最佳截断值为0.532,敏感度82.2%,特异度88.5%,预测效能优于传统指标;外部验证的AUC为0.887(95% CI:0.811～0.963),预测效能优于传统指标。⑤应用R软件绘制预测模型的可视化Nomogram图。结论 Lac、使用血管活性药物、APACHE Ⅱ评分和AGI分级是脓毒症AGI患者28死亡风险的独立危险因素。基于上述指标在校正性别和年龄后构建了预测脓毒症AGI患者28 d死亡风险的预后模型,在预测预后方面具有一定的准确度。 |
| 英文摘要: |
| Objective To develop and validate a nomogram model for predicting the risk of 28-day mortality in patients with septic acute gastrointestinal injury (AGI). Methods A retrospective analysis was performed on the clinical data of 169 patients with AGI who were admitted to the Intensive Care Department of the First Affiliated Hospital of Anhui Medical University from October 2019 to March 2021. The patients were divided into survival group (96 cases) and death group (73 cases) according to 28-day clinical outcomes. LASSO regression was used to screen the optimal predictive variables and multivariate logistic regression analysis was used to establish the prediction model. The model's discrimination, calibration and clinical validity were internally validated by bootstrap resampling.The receiver operating characteristic (ROC) curve was used to evaluate the prediction efficiency. Data of 73 AGI patients with sepsis admitted to the Intensive Care Department of the First Affiliated Hospital of Anhui Medical University from April 2021 to November 2021 were retrospectively collected for external validation of the model. Nomogram plots were drawn with R software. Results ①Univariate analysis:the CRP, PH, Lac, MAP, ratio of vasoactive drug use, APACHE Ⅱ score, SOFA score, ratio of septic shock and AGI grade showed statistically significant differences between the survival group and the death group. ②Lasso regression selected four optimal predictor variables including lac, use of vasoactive drugs, APACHE Ⅱ score and AGI grade, following which logistic prediction models were built. ③Internal and external validation of the prediction model.The consistency index (C-index) was 0.917 and 0.881, and the corrected C-index was 0.904 and 0.841, respectively, indicating a good degree of differentiation. Hosmer-lemeshow tests all showed P>0.05, the slope of calibration curve was 1, Brier value was 0.118 and 0.132, respectively, indicating good calibration degree. The threshold probability of decision analysis (DCA) curve showed good clinical effectiveness between more than 1% and 6%~89%, respectively.④The ROC curve for the prediction model achieved an area under the curve (AUC) of 0.913(95%CI:0.870~0.955), and the optimal truncation value was 0.532, with the sensitivity of 82.2% and specificity of 88.5%.The prediction efficiency was better than the traditional indicators. The externally validated AUC was 0.887(95%CI:0.811~0.963), indicating better predictive efficacy than other traditional indicators. ⑤R software could be used to draw a visual nomogram of the prediction model. Conclusions The Lac level, use of vasoactive drugs, APACHE Ⅱ score and AGI grade areindependent risk factors for predicting the risk of death at 28-day for patients with septic AGI. Based on the above indicators, a prognostic model can be constructed to predict the risk of death on 28-day in patients with septic AGI after adjusting for gender and age.This model has certain accuracy in predicting prognosis and can provide a basis for clinicians'diagnosis and treatment measures to improve the prognosis of patients. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
