4.3 Multivariate model
In univariate analysis, sex, type of AF, late recurrence, LAA flow velocity, and pre-RFCA VENC-MRI was associated with ∆VENC-MRI (Table 2). After including these factors in the multivariate model, late recurrence (unstandardized coefficient = -10.864; p < 0.001) and pre-RFCA VENC-MRI (unstandardized coefficient = -0.628; p < 0.001) were independent factors associated with ∆VENC-MRI (Table 2).
Estimated means calculated by general linear model (analysis of covariance) is depicted in Figure 3. After adjusting sex, AF type, and pre-RFCA VENC-MRI, late recurrence was independently associated with ∆VENC-MRI: patients without late recurrence had significantly higher improvement in LAA hemodynamics after RFCA (∆VENC-MRI = 13.80 [8.38 – 19.21] vs. 26.85 [22.56 – 31.15]; p < 0.001; Figure 3). Low LAA flow volume (pre-RFCA VENC-MRI < 20 cm/sec) was independently associated with higher improvement in VENC-MRI after RFCA after adjusting the influence of sex, AF type, and late recurrence (∆VENC-MRI = 17.17 [13.09 – 21.25] vs. 48.13 [38.29 – 57.98]; p < 0.001; Figure 3).