急性心肌梗死患者并发左心室收缩功能不全的危险因素及风险预测模型研究

Risk factors and risk prediction models for left ventricular systolic dysfunction in patients with acute myocardial infarction

目的探讨急性心肌梗死(AMI)患者并发左心室收缩功能不全(LVSD)的危险因素,并基于危险因素构建风险预测模型。方法将连云港市第一人民医院2021年5月~2024年2月收治的279例AMI患者按照7∶3比例随机分为建模组(n=195)和验证组(n=84),另将建模组分为非LVSD组(n=113)和LVSD组(n=82)。多因素logistic回归分析AMI患者并发LVSD的影响因素,并基于筛选的危险因素在R3.4.1中构建列线图预测模型;采用ROC曲线、Hosmer-Lemeshow拟合度检验评估模型预测价值,模型的临床应用价值采用临床决策曲线(DCA)分析。结果279例AMI患者LVSD总发生率为43.01%。建模组和验证组心率、中性粒细胞与淋巴细胞比值(NLR)、血肌酐、吸烟史等比较均无显著差异(P>0.05)。与非LVSD组比较,LVSD组NLR[(6.28±1.19)比(5.03±1.08)]和Gensini积分[(28.02±5.86)分比(23.54±4.35)分]明显较高(t=7.642、6.128,P<0.05),且吸烟史(56.10%比36.28%)、合并糖尿病(31.71%比14.16%)、冠状动脉病变≥3支(45.12%比23.89%)的占比也明显较高(χ^(2)=7.500、8.659、9.712,P<0.05)。多因素logistic回归表明,NLR(OR=2.674)、Gensini积分(OR=1.185)、具有吸烟史(OR=3.009)、合并糖尿病(OR=6.730)、冠状动脉病变≥3支(OR=4.387)为AMI患者并发LVSD的独立危险因素(P<0.05)。ROC曲线显示,建模组曲线下面积(AUC)为0.882(95%CI 0.833~0.932),验证组AUC为0.917(95%CI 0.855~0.979);Hosmer-Lemeshow拟合度检验中,建模组χ^(2)=9.797,P=0.280,验证组χ^(2)=12.164,P=0.144。临床决策曲线(DCA)显示,模型临床应用价值较高。结论高水平NLR、Gensini积分高、具有吸烟史、合并糖尿病以及冠状动脉病变≥3支的AMI患者并发LVSD的风险较高,基于这五个因素构建的模型预测区分度和一致性较佳,有助于临床对AMI患者并发LVSD情况进行评估。

Objective To investigate the risk factors of left ventricular systolic insufficiency(LVSD)in patients with acute myocardial infarction(AMI),and establish a risk prediction model based on the risk factors.Methods A total of 279 AMI patients admitted to our hospital from May 2021 to February 2024 were randomly separated into a modeling group(n=195)and a validation group(n=84)in a 7∶3 ratio.The modeling group was also separated into a non LVSD group(n=113)and an LVSD group(n=82).Multivariate logistic regression analysis was conducted to analyze the influencing factors of AMI patients with LVSD,and a nomogram prediction model was constructed in R3.4.1 based on the screened risk factors.ROC curve and Hosmer-Lemeshow fit test were used to evaluate the predictive value of the model;The clinical application value of the model was analyzed by clinical decision curve(DCA).Results The total incidence of LVSD in 279 AMI patients was 43.01%.There were no obvious differences in heart rate,neutrophil to lymphocyte ratio(NLR),blood creatinine,and smoking history between modeling group and validation group(P>0.05).Compared with the non LVSD group,the level of NLR[(6.28±1.19)vs.(5.03±1.08)]and Gensini scores[(28.02±5.86)scores vs.(23.54±4.35)scores]in the LVSD group were obviously higher(t=7.642,6.128,P<0.05),and the proportions of smoking history(56.10%vs.36.28%),diabetes(31.71%vs 14.16%)and coronary artery lesions≥3 branches(45.12%vs.23.89%)were also obviously higher(χ^(2)=7.500,8.659,9.712,P<0.05).Multivariate logistic regression showed that NLR(OR=2.674),Gensini score(OR=1.185),smoking history(OR=3.009),diabetes(OR=6.730),and coronary artery lesions≥3 branches(OR=4.387)were independent risk factors for LVSD in AMI patients(P<0.05).ROC curve showed that the area under the curve(AUC)of the modeling group was 0.882(95%CI 0.833-0.932),and the AUC of the validation group was 0.917(95%CI 0.855-0.979);In the Hosmer-Lemeshow fit test,modeling groupχ^(2)=9.797,P=0.280,validation groupχ^(2)=12.164,P=0.144.Clinical decision curve(DCA)shows that the model has high clinical application value.Conclusion AMI patients with high level of NLR,high Gensini scores,smoking history,diabetes,and coronary artery lesions≥3 branches had a higher risk of LVSD.The model based on these five factors has better prediction differentiation and consistency,which is helpful for clinical evaluation of LVSD in AMI patients.