Analysing the analysis
CMAJ 1997;156:765
See response from: D. Moher
The meta-analysis by Sheila Hood, David Moher and Graeme G. Barber of the management of intermittent claudication with pentoxifylline (CMAJ 1996;155:1053-9 [abstract / résumé]) shows a mean improvement in pain-free walking distance of 929 m and an absolute walking distance of 48 m, according to the 12 studies included, but little information about the validity of the results summarized. The article prompted me to scrutinize thoroughly 1 of the studies cited.
There are several problems with the study by Porter and associates.[1] From a statistical standpoint, the F values from the variance analysis are not given, and neither are the standard deviations or the standard errors of the means. More important, there was no analysis by intention to treat, an unacceptable weakness in a trial in which only 82 patients out of 128 completed the study. Had such an analysis been performed, the only convincing finding may have been the positive association between pentoxifylline treatment and nausea (p < 0.001, chi2 = 20.57, my calculation). Finally, the raw data were transformed once for the variance analysis (log of treatment walking distance divided by control walking distance) and a second time in preparing the final figures (geometric mean of treatment walking distance divided by control walking distance, -1 × 100). By choosing such a data transformation for their figures, the authors failed to reveal that, at week 16, the patients in the control group fared better than those in the pentoxifylline group, as could easily be calculated from the table presenting the initial claudication distance (Fig. 1).
These results do not inspire much confidence in their validity. If the other 11 trials summarized by Hood and associates have not been scrutinized for major biases, what is the use of compiling and presenting their results?
Pierre Biron, MD, MS
Professor of Pharmacology
Université de Montréal
Montreal, Que.
Reference
- Porter JM, Cutler BS, Lee BY, et al. Pentoxifylline efficacy in the treatment of intermittent claudication: multicenter controlled double-blind trial with objective assessment of chronic occlusive arterial disease patients. Am Heart J 1982;104:66-72.
 |
Send a letter to the editor responding to this letter
Envoyez une lettre à la rédaction au sujet de cette lettre |
Letters