IJCS | Volume 33, Nº2, March / April 2020

164 Figure 3 - Dispersion of points representing predicted postoperative atrial fibrillation (POAF) (by logistic model) and observed POAF among the patients (n = 1,054, events = 272 POAF). The Pearson coefficient was r = 0.99 m with x² (Hosmer-Lemeshow) = 0.98 (p = 0.986), indicating good performance of the final risk score model. Observed postoperative atrial fibrillation Predicted postoperative atrial fibrillation (logistic model) associated with age and consequently associated with a reduction of atrial electric conduction velocity, generating an arrhythmogenic substrate. 28 The presence of mitral valve disease was a predictor of POAF in our sample, adding one point to the score, and increasing by 2.3-fold the risk for POAF. This factor alone had not been described by other authors in studies on POAF and differs from the classic associationwithAF. As a pathophysiological mechanism, stenosis andmitral insufficiency cause atrial dysfunction due to left atrial pressure and volume overload with consequent atrial dysfunction and arrhythmic substrate. We emphasize that pathophysiological changes at the cellular level in valve disease have been poorly described in the literature. In addition, surgical manipulation of the atria may be associated with the development of POAF. The diameter of the left atrium alone was not a predictor of POAF. 14,29 The absence of beta blockers or their discontinuation in the postoperative period had an important contribution (possibly four points) to the score developed in our study. Previous studies, including meta-analyses, have described this strong association. In a meta-analysis including 28 studies and 4,074 patients, Crystal et al., 30 reported an OR of 0.35 (95% CI 0.26-0.49) associated with this finding. Andrews et al., 31 analyzing 24 studies, reported that patients with ejection fractions greater than 30% were associated with an OR of 0.28 (95% CI 0.21- 0.36). The worst clinical scenario would be the non-use of beta-blokers during the pre- and postoperative periods. The only postoperative factor per se, namely, the presence of a positive water balance greater 1,500 mL in 48 hours after surgery, was a predictor of POAF in our cohort, contributing one point to the score. The mechanism was likely related to atrial dilatation during this critical in ammatory period, which has been described by Kalus et al. 32 High risk scores for the other variables in the preoperative period could guide a restrictive strategy in the management of postoperative hydration if there is no clinical contraindication. The area under the ROC curve was 0.76 (95% CI 0.72 - 0.79), reflecting the discriminating power of the model. Regarding the calibration, an HL test of r = 0.99, x² = 0.98 Ronsoni et al. POAF risk score Int J Cardiovasc Sci. 2020; 33(2):158-166 Original Article