急性ST段抬高型心肌梗死患者直接PCI术后院内死亡风险预测模型的建立和验证

Establishment and validation of prediction model of in-hospital mortality risk in patients with acute ST-segment elevation myocardial infarction after primary percutaneous coronary intervention

目的 探讨急性ST段抬高型心肌梗死(STEMI)患者直接经皮冠状动脉介入治疗(PCI)术后院内死亡危险因素,建立风险预测模型并进行验证。方法 将2015年1月至2018年4月于柳州市人民医院心血管内科诊断STEMI并行直接PCI的480例患者进行回顾性分析(建模组),记录相关临床资料,包括年龄、性别、吸烟史、高血压病史、糖尿病史、血脂异常史、冠心病史、陈旧性心肌梗死(OMI)病史、PCI史、脑血管病史、肾功能不全病史、入院收缩压、舒张压、脉压差、心率、梗死部位、恶性室性心动过速或心室颤动发生情况、入院Killip功能分级、再灌注时间、严重冠状动脉(冠脉)病变支数、慢复流/无复流发生情况、支架内血栓发生情况、心肌钙蛋白Ⅰ(cTnI)、肌酸激酶同工酶(CK-MB)、入院血糖、空腹血糖、总胆固醇(TC)、低密度脂蛋白胆固醇(LDL-C)、高密度脂蛋白胆固醇(HDL-C)、三酰甘油(TG)、超敏C反应蛋白(hs-CRP)、N末端脑钠肽前体(NT-proBNP)、血肌酐(Cr)、白细胞计数(WBC)、血红蛋白(Hb),以及是否发生院内死亡。通过单因素和多因素回归分析筛选出与发生死亡相关的独立危险因素,建立风险预测模型,采用Hosmer-Lemeshow(H-L)拟合优度检验和受试者工作特征(ROC)曲线对该模型进行评估。另收集2018年4月至2019年6月于柳州市人民医院心血管内科诊断STEMI并行直接PCI的108例患者临床资料(验模组),验证本模型的预测价值。结果 建模组患者院内死亡16例(3.3%),生存464例(96.7%),将建模组患者分为院内死亡组和存活组,单因素分析结果显示,院内死亡组患者年龄大于存活组,恶性室性心动过速或心室颤动发生率、入院Killip泵功能分级≥3级发生率、3支严重冠脉病变发生率、支架内血栓发生率、入院cTnI、CK-MB、入院血糖、空腹血糖、hsCRP、NT-pro BNP、Cr高于存活组,脉压差、TG、Hb低于存活组,差异有统计学意义(P<0.05)。多因素Logistic回归分析结果显示,年龄、前壁梗死、入院Killip功能分级≥3级、3支严重冠脉病变、hs-CRP是院内死亡的独立危险因素。基于独立危险因素建立的Logistic回归模型为:Logit P=-10.015+0.056×(年龄)+1.896×(前壁梗死)+4.296×(入院Killip功能分级≥3级)+1.231×(3支冠脉严重病变)+0.018×(hs-CRP)。H-L检验显示本模型拟合优度较高(P=0.898)。建模组患者院内死亡的ROC曲线下面积为0.938(95%CI:0.885~0.991,P<0.01),提示模型预测效度显著。根据模型计算的院内死亡风险分值最佳界点为-3.28,敏感度为0.875,特异度为0.897,阳性预测值为22.22%,阴性预测值为99.52%。验模组患者院内死亡的ROC曲线下面积为0.937(95%CI:0.828~1.000,P<0.01)。建模组和验模组患者院内死亡的ROC曲线下面积的比较,差异无统计学意义(P=0.9938)。结论 年龄、前壁梗死、入院Killip功能分级≥3级、3支严重冠脉病变、hs-CRP是STEMI患者直接PCI术后发生院内死亡的独立危险因素。基于危险因素建立的风险预测模型,预测效度显著,敏感度和特异度高,可用于指导临床实践。

Objective To investigate the risk factors of in-hospital mortality in patients with acute STsegment elevation myocardial infarction(STEMI) after primary percutaneous coronary intervention(pPCI), establish and validate the risk prediction model. Methods STEMI patients undergone pPCI(n=480) were chosen into modeling group from Department of Cardiovascular Medicine in People’s Hospital of Liuzhou City from Jan. 2015to Apr. 2018. The clinical materials were recorded, including age, gender, histories of smoking, hypertension,diabetes, dyslipidemia, coronary heart disease(CHD), old myocardial infarction(OMI), PCI, cerebrovascular diseases and renal insufficiency, hospitalized systolic blood pressure(SBP), diastolic blood pressure(DBP), pulse pressure(PP), heart rate(HR), infarction sites, malignant ventricular tachycardia(VT) or ventricular fibrillation(VF), Killip classification, reperfusion time, number of severe coronary lesion vessels, slow/no reflow, in-stent thrombosis, cardiac troponin I(cTnI), creatine kinase isoenzyme(CK-MB), blood sugar, fasting plasma glucose(FPG), total cholesterol(TC), low-density lipoprotein-cholesterol(LDL-C), high-density lipoprotein-cholesterol(HDL-C), triglyceride(TG), high-sensitivity C-reactive protein(hs-CRP), N-terminal pro-brain natriuretic peptide(NT-proBNP), serum creatinine(SCr), white blood cell count(WBC), hemoglobin(Hb), and whether in-hospital mortality occurred or not. The independent risk factors related to death were screened by using single-factor and multi-factor regression analysis, and the risk prediction model was established. The model was evaluated by using Hosmer-Lemeshow(H-L) goodness-of-fit test and receiver operating characteristic(ROC) curve. In addition, the clinical materials were collected from STEMI patients undergone pPCI(validation group, n=108) from Department of Cardiovascular Medicine in People’s Hospital of Liuzhou City from Apr. 2018 to June 2019. Results There were 16 patients(3.3%) died in hospital and 464 patients(96.7%) survived in modeling group. The modeling group was divided into death group and survival group. The results of single-factor regression analysis showed that age,incidence rates of malignant VT or VF, Killip grade≥3 and severe coronary lesion in 3 vessels, and levels of cTnI,CK-MB, blood sugar, FPG, hs-CRP, NT-proBNP and SCr were higher, and PP, TG and Hb were lower in death group than those in survival group(P<0.05). The results of multi-factor Logistic regression analysis showed that age,anterior myocardial infarction, Killip grade≥3, severe coronary lesion in 3 vessel and hs-CRP were independent risk factors of in-hospital mortality. The Logistic regression model established based on independent risk factors was: logit P=-10.015+0.056(age)+1.896(anterior myocardial infarction)+4.296(Killip grade≥3)+1.231(severe coronary lesion in 3 vessel)+0.018(hs-CRP). The results of H-L test showed that the goodness of fit of the model was higher(P=0.898). The area under ROC curve of the model in predicting in-hospital mortality was 0.938(95%CI:0.885~0.991, P<0.01) in modeling group. The best cutoff point of in-hospital mortality risk scores calculated by the model was-3.28, sensitivity was 0.875 and specificity was 0.897, positive predictive value(PPV) was 22.22%,and negative predictive value(NPV) was 99.52%. The area under ROC curve of the model in predicting in-hospital mortality was 0.937(95%CI: 0.828~1.000, P<0.01) in validation group. The comparison in the area under ROC curve had no statistical significance between modeling group and validation group(P=0.9938). Conclusion Age, anterior myocardial infarction, Killip grade≥3, severe coronary lesion in 3vessels and hs-CRP are independent risk factors for in-hospital mortality in STEMI patients after pPCI. The risk prediction model established based on risk factors has significant predictive efficacy and higher sensitivity and specificity, which can be used to guide clinical practice.