Related Experiment Videos

A simple model to predict coronary disease in patients undergoing operation for mitral regurgitation

Eric Lim1, Ziad A Ali, Clifford W Barlow

  • 1Department of Cardiothoracic Surgery, Papworth Hospital, Cambridge, United Kingdom. ericlim2@hotmail.com

Insights

A simple additive model accurately predicts coronary artery disease in patients with mitral valve disease, offering a bedside-applicable tool. This method improves upon traditional risk factor identification for better patient outcomes.

Area of Science:

  • Cardiology
  • Cardiac Surgery
  • Medical Informatics

Background:

  • A significant portion of patients with mitral valve disease also have coexistent coronary artery disease.
  • Accurate identification of coronary artery disease is crucial for guiding treatment decisions in these patients.

Purpose of the Study:

  • To evaluate a candidate selection strategy using risk factor identification and logistic regression.
  • To develop a simple additive model for predicting coexistent coronary artery disease in patients undergoing mitral valve repair.

Main Methods:

  • Analysis of a consecutive series of patients who underwent mitral repair between 1987 and 1999.
  • Calculation of sensitivities and specificities for individual risk factors.
  • Development of a predictive score using univariate and stepwise multivariate logistic regression, followed by derivation of a logistic regression-derived additive model.
  • Comparison of model discrimination and precision using receiver operating characteristic curves and the Hosmer-Lemeshow statistic.

Main Results:

  • The American Heart Association and American College of Cardiology risk factor identification criteria achieved 100% sensitivity but only 1% specificity.
  • The developed logistic regression model demonstrated strong predictive ability (Area Under Curve [AUC] = 0.91, Hosmer-Lemeshow p = 0.9).
  • The five-item additive model showed comparable discrimination (AUC = 0.91, Hosmer-Lemeshow p = 0.80), outperforming the Cleveland Clinic model (AUC = 0.79).

Conclusions:

  • While simple risk factor identification is sensitive, its low specificity limits its clinical utility.
  • Logistic regression modeling provides accurate risk prediction but can be complex for bedside application.
  • The logistic regression-derived additive model offers a balance of simplicity and accuracy for predicting coexistent coronary artery disease in this patient population.
Abstract

Related Concept Videos