The BG Study Part 1 (Bergisch Gladbach): Development of a Prototype Coronary Artery Disease Risk Score Incorporating Peripheral Vascular Parameters—Preliminary Insights for Future CAD Risk Prediction Models in Vascular Patients

TL;DR

This study developed a new risk score for coronary artery disease using peripheral vascular data and clinical history, showing high sensitivity but low specificity.

Contribution

A novel CAD risk score integrating peripheral vascular parameters and clinical factors is proposed, with preliminary insights into its performance.

Findings

The risk score achieved high sensitivity (97%) for CAD-positive cases but low specificity (11%) for CAD-negative cases.

Key predictors included carotid artery velocities, ABI, CKD stage, smoking, diabetes, hypertension, and age.

A threshold of ≥7 points maximized the F1 score for CAD risk stratification.

Abstract

Background: Peripheral vascular parameters may provide valuable insights into coronary artery disease (CAD) risk stratification. This study aimed to develop a CAD risk score by integrating carotid duplex velocities, ankle–brachial index (ABI), and clinical history into a point-based model. Methods: We analyzed data from 902 cardiology patients, of whom 592 (65.6%) had confirmed CAD based on coronary angiography. Peripheral vascular assessments included carotid duplex ultrasonography and ABI measurements. Predictors were identified through multivariate logistic regression, addressing multicollinearity and interaction effects. A point-based scoring system was developed using statistically significant variables and evaluated via receiver operating characteristic (ROC) analysis. Results: Key predictors included external carotid artery velocities, ABI, carotid stenosis, chronic kidney disease (CKD) stage, smoking history, diabetes, hypertension, and age. The scoring system demonstrated moderate discriminative ability (AUC: 0.683) and high sensitivity (97%) for detecting CAD-positive cases but lower specificity (11%) for CAD-negative cases. Patients were stratified into risk categories, with an optimal threshold of ≥7 points maximizing the F1 score. Conclusions: This novel scoring system highlights the clinical relevance of integrating peripheral vascular assessments into CAD risk models. While its high sensitivity ensures robust detection of CAD-positive patients, future multicenter studies are needed to improve specificity and validate its broader clinical utility.

Figures