Risk prediction model for major adverse cardiovascular events (MACE) during hospitalization in patients with coronary heart disease based on myocardial energy metabolic substrate

Background

The early attack of coronary heart disease (CHD) is very hidden, and clinical symptoms generally do not appear until cardiovascular events occur. Therefore, an innovative method is needed to judge the risk of cardiovascular events and guide clinical decision conveniently and sensitively. The purpose of this study is to find out the risk factors related to MACE during hospitalization. In order to develop and verify the prediction model of energy metabolism substrates, and establish a nomogram to predict the incidence of MACE during hospitalization and evaluate their performance.

Methods

The data were collected from the medical record data of Guang'anmen Hospital. This review study was collected the comprehensive clinical data of 5,935 adult patients hospitalized in the cardiovascular department from 2016 to 2021. The outcome index was the MACE during hospitalization. According to the occurrence of MACE during hospitalization, these data were divided into MACE group (n = 2,603) and non-MACE group (n = 425). Logistic regression was used to screen risk factors, and establish the nomogram to predict the risk of MACE during hospitalization. Calibration curve, C index and decision curve were used to evaluate the prediction model, and drawn ROC curve to find the best boundary value of risk factors.

Results

The logistic regression model was used to establish a risk model. Univariate logistic regression model was mainly used to screen the factors significantly related to MACE during hospitalization in the training set (each variable is put into the model in turn). According to the factors with statistical significance in univariate logistic regression, five cardiac energy metabolism risk factors, including age, albumin(ALB), free fatty acid(FFA), glucose(GLU) and apolipoprotein A1(ApoA1), were finally input into the multivariate logistic regression model as the risk model, and their nomogram were drawn. The sample size of the training set was 2,120, the sample size of the validation set was 908. The C index of the training set is 0.655 [0.621,0.689], and the C index of the validation set was 0.674 [0.623,0.724]. The calibration curve and clinical decision curve show that the model performs well. The ROC curve was used to establish the best boundary value of the five risk factors, which could quantitatively present the changes of cardiac energy metabolism substrate, and finally achieved prediction of MACE during hospitalization conveniently and sensitively.

Conclusion

Age, albumin, free fatty acid, glucose and apolipoprotein A1 are independent factors of CHD in MACE during hospitalization. The nomogram based on the above factors of myocardial energy metabolism substrate provides prognosis prediction accurately.