中国冠状动脉旁路移植术住院死亡风险模型
基于中国心脏外科注册数据库87家医院2013-2016年56776例冠状动脉旁路移植术患者,开发并验证了包含16个独特变量(21个风险因素)的住院死亡风险预测模型。模型在训练、测试和验证样本中C统计量分别为0.79、0.79和0.78,校准良好,且判别力和校准均优于EuroSCORE II和SinoSCORE。
要点
- 新模型含16个变量,预测CABG住院死亡风险,判别力与校准良好。
- 模型在训练、测试和验证样本中C统计量分别为0.79、0.79和0.78。
- 与EuroSCORE II和SinoSCORE相比,新模型判别力和校准更优。
核心结论
- 开发并评估了包含16个风险因素的模型,可预测中国CABG患者住院死亡风险,有助于识别高危患者。
研究设计与样本
- 基于中国心脏外科注册数据库,纳入2013年1月至2016年12月87家医院行CABG的患者56776例,平均年龄61.8岁(SD 8.8),女性占24.6%。
- 2013-2015年患者随机分为训练样本(n=31297,75%)和测试样本(n=10432,25%),2016年患者(n=15047)为验证样本。
主要发现
- 总体住院死亡率为2.1%。
- 最终模型包含21个风险因素,由16个独特变量表示。
- 训练样本C统计量0.79(95%CI 0.77-0.80),测试样本0.79(95%CI 0.76-0.82),验证样本0.78(95%CI 0.76-0.81)。
- Hosmer-Lemeshow检验显示3个样本校准良好(P > .05)。
- 与EuroSCORE II和SinoSCORE相比,该模型判别力和校准更优。
关键数值与推荐
- 模型包含21个风险因素(16个独特变量),具体变量列表原文未提供。
- 训练样本C统计量0.79(95%CI 0.77-0.80),测试样本0.79(95%CI 0.76-0.82),验证样本0.78(95%CI 0.76-0.81)。
- Hosmer-Lemeshow检验P > .05(3个样本)。
临床怎么用
- 可用于中国CABG患者术前住院死亡风险评估,帮助外科医生和医院识别高危患者。
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
BACKGROUND: To meet the demand of increasing surgical volume and changing of patient's risk profiles of coronary artery bypass grafting in China, we developed a new risk model that predicts in-hospital mortality. METHODS: The analysis included patients who underwent coronary artery bypass grafting between January 2013 and December 2016 at 87 hospitals in the Chinese Cardiac Surgery Registry. Patients in years 2013 to 2015 were randomly divided into training (n = 31,297 [75%]) and test (n = 10,432 [25%]) samples; 2016 patients (n = 15047) comprised the validation sample. Demographic and clinical risk factors were identified. The Harrell C statistic was used to evaluate model discrimination, and the Hosmer-Lemeshow goodness-of-fit test was used to assess calibration. RESULTS: The 56,776 patients were a mean age of 61.8 (SD, 8.8) years, and 24.6% were women. Overall, in-hospital mortality was 2.1%. The final model included 21 risk factors represented by 16 unique variables. The model achieved good discrimination, with a C statistic of 0.79 (95% confidence interval [CI], 0.77-0.80) in the training sample, 0.79 (95% CI, 0.76-0.82) in the test sample, and 0.78 (95% CI, 0.76-0.81) in the validation sample. Model calibration was good according to the Hosmer-Lemeshow test (P > .05 in the 3 samples). Compared with the European System for Cardiac Operative Risk Evaluation 2011 revision (EuroSCORE II) and the Sino(Chinese) System for Coronary artery bypass grafting Operative Risk Evaluation (SinoSCORE), the model had better discrimination and calibration. CONCLUSIONS: We developed and evaluated a model with 16 risk factors that predicted in-hospital mortality risk after coronary artery bypass grafting in China. This updated model may help surgeons and hospitals better identify high-risk patient.
引用
- Hu Z, Chen S, Du J, et al. An In-hospital Mortality Risk Model for Patients Undergoing Coronary Artery Bypass Grafting in China. Ann Thorac Surg. 2019. PMID: 31541633. doi:10.1016/j.athoracsur.2019.08.020
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。