From the authors:
We thank J.D. Blakey for his interest in our article [1]. In the paper he cited, Bland and Altman [2] criticise randomised studies in which baseline is compared with a final measurement separately for each group through paired t-tests. By using simulations, they concluded that the use of paired tests against baseline separately for each group is conceptually wrong and statistically invalid. We agree that this is not a proper approach. However, this was not the approach adopted in our study. As explained in the Methods section, we applied a repeated-measures ANOVA that allows integration of the between- and within-group analyses into the same model. This methodology starts from the hypothesis of interaction between group and time; a significant p-value for this hypothesis indicates that groups behave differently across time, and, in this situation, the within-group comparisons should be performed directly in this model as contrasts. Therefore, the methodology adopted in our study was, indeed, different from the one criticised by Bland and Altman [2], and we assure you that we used a statistically valid method [3, 4].
Acknowledgments
The authors thank C.S. Redigolo (Dept of Public Politics and Public Health, Federal University of São Paulo, Santos, Brazil) and R.A. de Oliveira Figueiredo (Faculty of Public Health, University of São Paulo, São Paulo, Brazil) for assistance with the statistical analysis.
Footnotes
Conflict of interest: Disclosures can be found alongside the online version of this article at erj.ersjournals.com
- Received June 26, 2014.
- Accepted June 30, 2014.
- ©ERS 2014