| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
LETTER TO THE EDITOR
Recently, we advocated the need to account for shear stress in determining whether differences in FMD between groups are due to endothelial function and/or simply a difference in reactive hyperemia-induced shear stress (6). In the study addressed by Dr.'s Harris and Padilla (7), our main objective was to determine whether it is the peak shear stimulus on release of forearm ischemia, or the continued, progressively decaying shear stimulus that is responsible for the magnitude of FMD. The salient finding from this study was that the area under the curve (AUC) of a reactive hyperemia-induced shear stimulus, not the peak shear, determines the magnitude of FMD. When we divided FMD by AUC of shear stimulus, differences in FMD with different AUC were eliminated.
Dr's Harris and Padilla proposed that ANCOVA may be a more appropriate method for "normalization" of FMD. In this approach, the contribution of a covariate (shear stress) to differences in the dependent variable (FMD) between levels of an independent variable (i.e. control vs. patient group) is accounted for statistically, providing adjusted means for the dependent variable. We agree with the potential benefits of such an approach, which include lessening the effects of the covariate, preservation of FMD units as percentages and improved power. However, the authors have not considered two critical assumptions of the ANCOVA that must be met for this analysis approach to be valid. These are 1) normal distribution of the covariate and, more importantly, 2) the assumption of parallelism. Mathematically, the precise definition of the latter is that the regression weight (slope) of the regression line for the dependent variable and covariate within each group is not different across groups. If parallelism is not found, then the ANCOVA approach should be avoided, as any adjusted mean from ANCOVA would be misleading [see p. 365–366, "Assumption of parallelism: a potential drawback" in (3)]. Thus a test for parallelism is required before proceeding with the ANCOVA (3) approach for assessment of FMD differences between groups of interest.
The concept and application of normalization of FMD for shear is in its infancy, and there is a clear need to determine the most effective and valid approach. We encourage Dr.'s Harris and Padilla to 1) test their proposed hypothesis that normalization via ANCOVA vs. dividing FMD by shear stress will result in different interpretations of the same data, and 2) assess the validity of the assumption of parallelism when comparing FMD responses in groups of interest.
FOOTNOTES
Address for reprint requests and other correspondence: M. E. Tschakovsky, School of Kinesiology and Health Studies and Dept. of Physiology, Queen's Univ., Kingston, ON, Canada K7L 3N6 (e-mail: mt29{at}post.queensu.ca)
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  G. Atkinson, A. M. Batterham, M. A. Black, N. T. Cable, N. D. Hopkins, E. A. Dawson, D. H. J. Thijssen, H. Jones, T. M. Tinken, and D. J. Green Is the ratio of flow-mediated dilation and shear rate a statistically sound approach to normalization in cross-sectional studies on endothelial function? J Appl Physiol, December 1, 2009; 107(6): 1893 - 1899. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
