Clinical Prediction Rule for the Development of New Onset Postoperative Atrial Fibrillation After Cardiac Surgery

Abstract

This project set out to derive a prediction rule based on preoperative clinical variables to identify patients with high risk of developing atrial fibrillation following cardiac surgery. Methods: Prospectively collected data from a perioperative database was corroborated with chart review to identify eligible patients who had non-emergent surgery in 2010. Details on 28 preoperative variables were collected and significant predictors (p<0.2) were inserted into multivariable logistic regression and recursive partitioning. Results: 305 (30.5%) of 999 patients developed new onset postoperative atrial fibrillation. Eleven variables were significantly associated with atrial fibrillation, however, both final models included only three: left atrial dilatation, mitral valve disease and age. Bootstrapping with 5000 samples confirmed that both final models provide consistent predictions. Coefficients from the logistic regression model were converted into a simple seven point predictive score. Conclusions: This simple risk score can identify patients at higher risk of developing atrial fibrillation after cardiac surgery.