IJCS | Volume 33, Nº2, March / April 2020

162 Figure 1 - Area under the ROC curve for the occurrence of postoperative atrial fibrillation: c = 0.76 (95% confidence interval 0.72 - 0.79) in the final model (n = 1,054). Sensitivity Specificity Table 3 - Logistic regression and multivariable risk scores of the total sample (n = 1,054) Variable B P RR 95% CI Points Age ≥ 70 years 0.93 < 0.001 2.55 1.84-3.53 2 Mitral valve disease 0.53 0.01 1.70 1.09-2.65 1 Absence of beta-blocker 1.61 < 0.001 5.04 3.67-6.90 4 Water balance > 1,500 ml 0.43 0.01 1.53 1.1-2.15 1 Constant -2.56 < 0.001 0.07 Logistic equation: Prob(POAF) = 1 / (1 + exp (- (-2.56 + [0.93 * age ≥ 70] + [0.53 * mitral valve disease] + [1.61 * non-use and/or discontinuation of beta-blockers] + [0.43 * Water balance > 1500 mL]))). Discussion Despite advances in surgical techniques and postoperative management, POAF continues to be a very frequent complication. Alhoughmany factors associated with the occurrence of POAF have been reported, there are few prediction models available. 16-19 Our study identified four predictors for POAF that comprised: age ≥ 70 years, mitral valve disease, the non-use of beta-blockers in the preoperative period or their discontinuation in the postoperative period and a positive water balance greater than 1,500 mL within 48 hours after the surgery. Thus, an easy-to-apply and clinically useful tool was used to calculate the POAF risk. The selection of the variables was made based on the experience of the department of cardiac surgery of the ICIFUC and available literature. 14,20 When using predictive risk models, we evaluate the possibility of Ronsoni et al. POAF risk score Int J Cardiovasc Sci. 2020; 33(2):158-166 Original Article