Effect of Rosiglitazone (Avandia) on the Risk of Myocardial
Infarction
and Death from Cardiovascular Disease:
A
Critical Review and Appraisal
Another poorly constructed paper. We need to stop
publishing (and reading) meta-analyses. They have no constructive value. There
is too much heterogeneity of data, and studies that have a multiplicity of
defined endpoints do not allow for comparisons in any meaningful fashion.
These are then misinterpreted by the writers, reviewers and
further distorted by the popular press. The general public is left with widely
out-of-proportion conclusions. It only continues a culture feeding on mass
hysteria and precludes more rational debate and understanding.
This meta-analysis fails to prove the author’s assertions
for these reasons:
- The absolute occurrence of events is
exceedingly
small in all groups, a fact that is largely ignored – to your detriment.
The real rate of occurrence of all events varied
between 0.14% and 1.65%. The absolute
difference between control and drug groups never exceeds 0.44% --
less than 1%.
The difference between relative and absolutely outcomes continues to be
ignored by the scientific community and the popular press. See the
attached
table as illustration. We consistently and repeatedly advise reading
Calculated Risks by Gerd Giggerenzer.
- Once again, the excluded data are just as important
as the included data. Studies where no deaths or myocardial infarctions were
reported were excluded. The very absence of described endpoints thus skew the
data adversely. The control is defined as any other group not using Avandia. But this is poor variable control. Certainly not a true blind
control, where one group receives a true placebo and the other group receives
the tested drug (scientific variable).
- The meta-analysis includes three large groups. One
is a very large group of pooled data from a variety of short term trials.
This has limited meaning because the data is not pure and transparent to the
reviewers. The author admits this.
- The DREAM group used true control Rosiglitazone (Avandia) vs.
placebo.
- The ADOPT group used Rosiglitazone (Avandia) vs.
Glyburide or Metformin. Not a true comparison – but found the drug group
actually had more favorable outcomes. Not at all mentioned in the paper.
- Event confusion and data contamination:
"Because only summary data were
available, it was not possible to discern whether the same patient had both
events. Therefore, an outcome measure based on the composite of death or
myocardial infarction could not be constructed. Accordingly, these two
outcomes are reported separately."
- This results in Synthetic and arbitrary categorizations
and data distortions.
- All these agents are not alike:
“The patterns of gene activation
or suppression differ substantially among various PPAR agonists, even within
closely related compounds. The biologic effects of the protein targets for
most of the genes influenced by PPAR agonists remain largely unknown.” Actose
seems to have a more favorable effect on lipids, than Avandia.
- The ultimate goal of diabetic therapy [to which we agree]:
“…the ultimate value of anti-diabetic therapy is the
reduction of the complications of diabetes, not improvement in a laboratory
measure of glycemic control.”
- The effect on macro vascular complications has proved
to be unpredictable.
- Meta-analyses are un-convincing and ill advised [we
agree]:
“A meta-analysis is always considered less convincing
than a large prospective trial destined to assess the outcome of interest.”
- This paper contributes nothing of constructive value
to our understanding of this class of agent. It only further swells a
battalion of legal opportunists.
Philip Lee Miller, MD
Los Gatos Longevity Institute
May 26, 2007
Los Gatos, CA
author, the Life Extension Revolution
