Establishing and internally validating a predictive model for coronary heart disease incorporating triglyceride-glucose index, monocyte-to-high-density lipoprotein cholesterol ratio and conventional risk factors in patients experiencing chest pain and referred for invasive coronary angiography.

BACKGROUND

Coronary heart disease (CHD) represents a severe form of ischemic cardiac condition that necessitates timely and accurate diagnosis. Although coronary angiography (CAG) remains widely used to detect CHD, healthcare facilities, medical expenses, and equipment technology often limit its availability. Therefore, it is imperative to identify a non-invasive diagnostic approach with high accuracy for CHD.

METHODS

This cross-sectional research included patients with chest pain (≥ 18 years) hospitalized at Chengde Central Hospital between September 2020 and March 2024. Among the participants, 70% were split into the training, and 30% were randomly entered into the validation sets. In the training dataset, univariate and multivariate logistic regression analyses were rigorously employed to ascertain predictors of CHD. A model was formulated by incorporating these predictors in a nomogram, which was evaluated for accuracy using calibration curves. The model's discrimination was quantified by calculating the area under the receiver operating characteristic (ROC) curve, denoted as the area under the curve (AUC), and its clinical application value was determined through decision curve analysis (DCA). Finally, we compare our model against the pretest probability (PTP) calculated by the Update Diamond-Forrester Model (UDFM) as recommended by the ECS guidelines to comprehensively assess its performance.

RESULTS

This study included 1501 patients who presented with chest pain, with a mean age of 60.45 years, 865 males (57.60%). Multivariate logistic regression analysis revealed TyG index, MHR, male, age, diabetes, systolic blood pressure (SBP), regional wall motion abnormality (RWMA), ST-T changes, and low-density lipoprotein cholesterol (LDL-C) as independent predictors of CHD. A novel nomogram incorporating these independent risk factors exhibited high accuracy and perfect consistency, with a training set AUC calculated to be 0.733 (95% CI: 0.698-0.768), and the validation set maintained a strong performance at 0.721 (95% CI: 0.663-0.779). The calibration curves and the Hosmer-Lemeshow test confirmed the well-fitting model (P = 0.576 and P = 0.694). ROC curve analysis and DCA demonstrated that the model has robust forecasting capability.

CONCLUSION

The nomogram model in this study exhibited good discriminative ability, calibration, and a favorable net benefit. Its predictive performance exceeds that of the traditional PTP tool and may serve as a non-invasive and promising approach to aid clinicians in the early identification of CHD risk in patients presenting with chest pain.