SEARCH

SEARCH BY CITATION

Abstract

Background:

We sought to develop and validate a logistic model and a simple score system for prediction of significant coronary artery disease (CAD) in patients undergoing operations for rheumatic aortic valve disease.

Hypothesis:

The simple score model we established based on the logistic model was efficient and practical.

Methods:

A total of 669 rheumatic patients (mean age 51 ± 9 years), who underwent routine coronary angiography (CAG) before aortic valve surgery between 1998 and 2010, were analyzed. A bootstrap-validated logistic regression model on the basis of clinical risk factors was developed to identify low-risk (≤5%) patients, from which an additive model was derived. Receiver operating characteristic (ROC) curves were used to compare discrimination, and precision was quantified by the Hosmer-Lemeshow statistic. Significant coronary atherosclerosis was defined as 50% or more luminal narrowing in 1 or more major epicardial vessels determined by means of coronary angiography.

Results:

Eighty-eight (13.2%) patients had significant coronary atherosclerosis. Independent predictors of CAD include age, angina, diabetes mellitus, and hypertension. A total of 325 patients were designated as low risk according to the bootstrap logistic regression and additive models. Of these patients, only 4 (1.2%) had single-vessel disease, and none had high-risk CAD (ie, left main trunk, proximal left anterior descending, or multivessel disease). The bootstrap logistic regression and additive models show good discrimination, with an area under the ROC curve of 0.948 and 0.942, respectively.

Conclusions:

Our logistic regression model can reliably estimate the prevalence of significant CAD in rheumatic patients undergoing aortic valve operation, while the additive simple score system could reliably identify the low-risk patients in whom routine preoperative angiography might be safely avoided. Clin. Cardiol. 2012 doi: 10.1002/clc.22033

The authors have no funding, financial relationships, or conflicts of interest to disclose.

Dr. Guan-xin Zhang and Dr. Bai-ling Li have contributed equally to the work. Dr. Lin-han is co-corresponding author (sh_hanlin@hotmail.com).