布加综合征住院死亡率预测评分模型的开发与验证
基于美国全国住院患者样本库2008至2017年5306例(加权26110例)成人布加综合征患者数据,采用logistic回归开发住院死亡率预测模型。模型纳入年龄≥50岁、腹水、脓毒症、急性呼吸衰竭、急性肝衰竭、肝肾综合征及肿瘤七个独立危险因素,训练集AUC为0.87(95%CI 0.85-0.95),验证集AUC为0.89(95%CI 0.86-0.92),低、中、高风险组预测住院死亡率分别为4%、30%、66%。
要点
- 布加综合征患者整体住院死亡率为7.14%。
- 七个独立危险因素:年龄≥50岁、腹水、脓毒症、急性呼吸衰竭、急性肝衰竭、肝肾综合征、肿瘤。
- 模型区分度良好,训练集AUC 0.87,验证集AUC 0.89。
核心结论
- 基于全国行政数据库开发的布加综合征住院死亡率预测模型具有良好准确性,可将患者分为低、中、高风险组。
研究设计与样本
- 回顾性队列研究,数据来自美国全国住院患者样本库(National Inpatient Sample),纳入2008至2017年ICD-9或ICD-10编码诊断为布加综合征的成人患者,共5306例出院诊断(加权样本量26110例)。
- 采用logistic回归分析识别独立危险因素并构建预测模型,分别在训练集和验证集中建模与验证,并据此进行风险分层。
主要发现
- 总体住院死亡率为7.14%。
- 预测死亡的独立危险因素包括:年龄50岁或以上、腹水、脓毒症、急性呼吸衰竭、急性肝衰竭、肝肾综合征以及肿瘤。
- 模型在训练数据中的受试者工作特征曲线下面积(AUC)为0.87(95%CI 0.85-0.95),在验证数据中为0.89(95%CI 0.86-0.92)。
- 低、中、高风险评分患者的预测住院死亡率分别为4%、30%和66%。
关键数值与推荐
- 风险评分模型包含7个变量:年龄≥50岁、腹水、脓毒症、急性呼吸衰竭、急性肝衰竭、肝肾综合征、肿瘤。
- 训练集AUC 0.87(95%CI 0.85-0.95),验证集AUC 0.89(95%CI 0.86-0.92)。
- 低风险组预测住院死亡率4%,中风险组30%,高风险组66%。
临床怎么用
- 该模型使用易于获取的临床变量,可用于布加综合征患者住院早期风险分层,辅助识别高危患者并指导临床决策。
局限性
- 研究基于行政数据库(ICD编码),可能存在编码错误或遗漏;原文未报告模型的外部验证情况。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
INTRODUCTION: A model that can predict short-term mortality in patients with the Budd-Chiari syndrome (BCS) with a high degree of accuracy is currently lacking. The primary objective of our study was to develop an easy-to-use in-hospital mortality prediction model in patients with BCS using easily available clinical variables. METHODS: Data were extracted from the National Inpatient Sample to identify all adult patients with a listed diagnosis of BCS from 2008 to 2017 using ICD-9 or ICD-10 codes. After identifying independent risk factors of in-hospital mortality, we developed a prediction model using logistic regression analysis. The model was built and validated in a training and a validation data set, respectively. Using the model, we risk stratified patients into low-, intermediate-, and high-risk groups. RESULTS: Between 2008 and 2017, we identified a total of 5,306 (weighted sample size 26,110) discharge diagnosis of patients with BCS, with an overall in-hospital mortality of 7.14%. The independent risk factors that predicted mortality were age of 50 years or older, ascites, sepsis, acute respiratory failure, acute liver failure, hepatorenal syndrome, and cancers. The mortality prediction model that incorporated these risk factors had an area under the receiver operating characteristic curve of 0.87 (95% CI 0.85-0.95) for the training data and 0.89 (95% CI 0.86-0.92) for the validation data. Patients with low-, intermediate-, and high-risk scores had a predicted in-patient mortality of 4%, 30%, and 66%, respectively. DISCUSSION: Using a national administrative database, we developed a reliable in-patient mortality prediction model with an excellent accuracy. The model was able to risk stratify patients into low-, intermediate-, and high-risk groups.
引用
- Thuluvath PJ, Alukal JJ, Zhang T, et al. A Scoring Model to Predict In-Hospital Mortality in Patients With Budd-Chiari Syndrome. Am J Gastroenterol. 2021. PMID: 33900212. doi:10.14309/ajg.0000000000001273
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。