Linear regression was used. Only factors are shown that were univariately associated (P < 0.1) with MVCs either in males or females. AF, atrial fibrillation; n/a, factor did not enter the multivariate model due to univariate P > 0.1; n/s, factor was not significant in multivariate model (stepwise forward method); SEM, standard error of the mean; SLOPE, change of MVC rate (%) per unit change of factor; R, regression coefficient; US, ultrasound.