Determinants of postoperative atrial fibrillation: A retrospective evaluation of postoperative atrial fibrillation in cardiac surgery

Abstract

Introduction: Atrial fibrillation is the most common postoperative arrhythmia and is associated with increased length of stay, cost, morbidity and mortality. The incidence of postoperative atrial fibrillation for noncardiac, nonthoracic surgeries ranges from 0.4% to 26%. The incidence increases to 20%–50% in cardiac surgery, occurring in approximately 30% of isolated coronary artery bypass grafting (CABG), approximately 40% of isolated valve surgeries and up to 50% of CABG plus valve surgeries. Our aim was to identify risk factors that may predispose patients to postoperative atrial fibrillation and compare the efficacy of previously developed prediction tools to a new bedside prediction tool. We sought to develop a bedside screening tool using 4 easily identifiable variables: body mass index, age, congestive heart failure and hypertension (BACH). We predicted that our model would compare similarly to previously developed and validated prediction models but would be easier to use.

Methods: We retrospectively identified 672 patients without a history of atrial fibrillation who had undergone cardiac surgery from July 2011 to December 2018. The risk factors for atrial fibrillation were evaluated alongside previously developed prediction tools. Using logistic regression, t tests and receiver operator characteristic (ROC) analysis, we compared previously used risk stratification scores of CHA₂DS₂-VASc, CHARGE-AF and age. We also compared our proposed BACH risk prediction tool to our population and compared it against CHA₂DS₂-VASc, CHARGE-AF and age. In a subpopulation analysis of 259 people, we evaluated if left atrial size was an independent risk factor for the development of postoperative atrial fibrillation.

Results: A total of 131 patients—approximately 20%—developed postoperative atrial fibrillation. CHA₂DS₂-VASc had the lowest area under the curve (AUC) and did not perform as well at classifying patients with postoperative atrial fibrillation as the other 3 predictors. CHARGE-AF, age by itself and age per 5 years performed relatively similarly to one another. ROC was greatest for age alone (ROC area .634, 95% CI: .581–.688), followed by CHARGE-AF (ROC area .631, 95% CI: .577–.684), and finally CHA₂DS₂-VASc (ROC area .564, 95% CI: .509–.619). A logistic model was fit for the BACH variables (continuous versions of body mass index, age, congestive heart failure and hypertension). The model achieved good fit, χ2(671, N=672)=633.029, P=.816, Nagelkerke R2=.070. However, only the predictors of age and prior heart failure were found to be significant. For BACH, the C-statistic (and AUC) for the model was .645 (95% CI: .601, .707), which was marginally better than age alone. All the models that were fit using ROC analyses were not statistically different from one another in terms of performance. No statistical significance was found between the 2 groups for preoperative left atrial size.

Conclusion: These findings suggest that age may be the highest risk factor for postoperative atrial fibrillation. The bedside prediction tool BACH compared slightly better than age alone but was not statistically different from the other prediction tools’ performance. The BACH prediction tool is easy to use, includes only 4 factors that are readily available at the bedside and improves prediction over age alone.