Reply to 'Cardiac remodeling after reduction of high-flow arteriovenous fistulas in end-stage renal disease: methodological issues'
2017
Abstrakt
In the present study, cardiac index cutoff value was based on accepted values. Constructing an receiver operating characteristic curve, as suggested by the authors of the comment, would require dichotomization of the left ventricular end-diastolic diarneter change, which would bring another source of bias.
While cardiac index is a subject of intra-individual variability, so is the AVF flow. In the present study, a mixed linear model with cardiac index group, time and their interaction as fixed variables was used; thus, these models were not overpararneterized to the sample size.