Statistical Analysis

Continuous variables will be described by mean and standard deviation or as median and interquartile range. Categorical variables will be described by the size and frequency of every modality. Means comparison will be performed by analysis of variance or by nonparametric tests if the hypothesis of normality is rejected. The normality of continuous variables will be verified with the Shapiro-Wilk test. The statistical tests are bilateral with a 95% confidence interval.

A chi-square test will be performed for categorical variables. Yates correction or the Fisher exact test will be used if the conditions of validity for the chi-square test are not met.

A multivariate analysis will be performed with anticoagulant treatment (over or undertreated) as dependant factor. The independent variables will be age, gender, body mass index, type of AF, combined therapy. Univariate logistic regression will be carried out with a 10% output threshold. The final model will be performed with the parameters selected by the backward stepwise method of Wald. The selected variables in the final model will be tested at the 5% threshold. The interaction between selected parameters is tested at the 10% threshold.

The TTR will be calculated by the method first described by Rosendaal et al [8], which uses linear interpolation of INR values in each patient under oral anticoagulant treatment to calculate the percentage of days when the INR is in the therapeutic range (2.0-3.0) for nonvalvular AF.