急性心肌梗死患者心力衰竭风险预测模型研究

Study on risk prediction model of heart failure in patients with acute myocardial infarction

目的建立基于单中心数据的急性心肌梗死(AMI)患者术后心力衰竭发生风险预测模型并进行外部验证。方法收集我院2015年1月至2017年12月收治行经皮冠状动脉介入术(PCI)治疗的572例AMI患者的临床资料,按照入组先后时间分为建模组(438例)和验证组(134例),以术后发生心力衰竭为终点事件,建立心力衰竭风险预测模型,基于Cox多因素回归分析检验模型的判别能力,采用拟合优度检验对预测模型进行验证与评价。结果建模组438例患者根据PCI术后是否发生心力衰竭分为心力衰竭组(113例)和非心力衰竭组(325例)。经单因素Cox比例风险回归分析显示,年龄、糖尿病史、心脑血管病史、Gensini等级、运动幅度改变、病理性Q波、心律失常、心脏结构改变、高敏C反应蛋白、血肌酐、心肌肌钙蛋白T等与患者发生心力衰竭明显相关(均为P<0.05);进一步行多因素Cox比例风险回归分析显示,年龄、Gensini评分等级、心律失常、心脏结构改变、血肌酐、心肌肌钙蛋白T、室壁运动幅度属于发生心力衰竭的独立危险因素(均为P<0.05)。结合多因素Cox回归分析建立列线图模型,内部验证显示列线图对预测AMI患者发生心力衰竭的C-index指数为0.832(95%CI:0.792~0.886)。将验证组134例的研究数据作为测试集,对心力衰竭风险预测模型进行外部验证,发现校准曲线与标准曲线有良好的拟合度(χ;=8.962,P=0.288),证实该模型有较高的预测准确率。结论基于单中心数据所构建的AMI患者心力衰竭风险预测模型具有较好的准确性,能直观、个体化地预测心力衰竭的发生风险。

Objective To establish a postoperative heart failure risk prediction model for patients with acute myocardial infarction(AMI) based on single-center data and conduct external verification. Methods A total of 572 AMI patients treated with percutaneous coronary intervention(PCI) in our hospital from January 2015 to December 2017 were enrolled and divided into modeling group(438 cases) and verification group(134 cases) according to the time of enrollment. The end-point event was heart failure, and a heart failure risk prediction model was established. The discriminative ability of the model was tested based on Cox multivariate regression analysis, and a goodness of fit test was used to validate and evaluate the risk prediction model for heart failure. Results According to whether heart failure occurred after PCI, 438 patients in the modeling group were divided into the heart failure group(113 cases) and the non-heart failure group(325 cases). Univariate Cox proportional hazard regression analysis showed that age, history of diabetes, history of cardiovascular and cerebrovascular diseases, Gensini grade, changes in exercise amplitude, pathological Q waves, arrhythmia, changes in cardiac structure, high-sensitivity C-reactive protein, blood creatinine and troponin T were significantly related to the rate of heart failure(all P<0.05). Multivariate Cox proportional hazard regression analysis showed that age, Gensini score, arrhythmia, changes in cardiac structure, blood creatinine, troponin T and the amplitude of ventricular wall motion were independent risk factors for heart failure(all P<0.05). Combining multivariate Cox regression analyses results, the nomogram model was established, and internal verification showed that the C-index for predicting heart failure in AMI patients was 0.832(95%CI: 0.792-0.886). The research data of 134 cases in the verification group were taken as in the test set, the heart failure risk prediction model is externally verified. The calibration curve and the standard curve have a good fit(χ^(2)=8.962, P=0.288), which showed that the model had a higher prediction accuracy. Conclusions The heart failure risk prediction model for AMI patients based on single-center data has good accuracy, and can intuitively and individually predict the risk of heart failure.