Re: Question re: Meta-Analysis
- From: jason@xxxxxxxxxx (Jason)
- Date: Sat, 15 Oct 2005 07:18:08 -0700
In article <dipu8p$3cm$1@xxxxxxxxxxxxxxxxx>, David Rind
<drind@xxxxxxxxxxxxxxxxxxxxx> wrote:
> Jason wrote:
> > I have a question about a term used in a recent post:
> > META-ANALYSIS OF DATA--
> >
> > If a statin company wanted to prove that statins do not cause
> > cancer, would it be possible to do the following:
> >
> > Step 1
> > Secretly do a Meta-Analysis of Data of 50 studies.
> >
> > Step 2
> > Eliminate any of the studies that showed that
> > statins do cause cancer.
> >
> > Step 3
> > Do a Meta-Analysis of Data of the 20 studies that
> > showed that statins do NOT cause cancer.
> >
> > Step 4
> > Release to the news media the results of the
> > Meta-Anylysis of Data of the 20 studies that
> > that showed that statins do NOT cause cancer.
> >
> > They could say in the press release:
> > The results of previous radomised trials show that
> > statins do not cause cancer.
>
> No, not in the way you described. The studies that could potentially be
> used in the meta-analysis are published, and the meta-analysis lists the
> included studies, so there is no "secret" involved.
>
> However, sometimes it's not clear what studies should or should not be
> included in the analysis, and it is possible that decisions about this
> could alter the results. This has come up with meta-analyses of
> screening for breast cancer, for instance.
>
> In the case of the statin studies, most are well known and well
> designed. But, implicit in my earlier post, I was arguing that ALLHAT
> should potentially not have been included with the other studies because
> I don't like the way it was carried out. Of course, if I wanted to slant
> the results in favor of statins I would make just such an argument
> because ALLHAT showed little or no benefit with statins. And so, if I
> published a meta-analysis that excluded ALLHAT, people would have to
> decide whether I was doing this out of a prostatin bias or because I had
> legitimate impartial concerns about ALLHAT.
>
> Personally, I'd be quite concerned about a meta-analysis backed by a
> drug company that excluded a large negative study like ALLHAT, but more
> comfortable if the same meta-analysis were performed by someone without
> such ties.
Thanks for your post. I will only comment on your last point. Statin
companies donate or contribute money to various colleges and agencies that
conduct research studies related to statins. The people employed by those
colleges and agencies know that those statin companies will STOP donating
or contributing money to them if they produce studies that show that
statins have side effects. As a result, I don't trust studies that are
done by those sorts of agencies and colleges. Do you have this same
attitude about those types of colleges and agenicies?
Jason
--
NEWSGROUP SUBSCRIBERS MOTTO
We respect those subscribers that ask for advice or provide advice.
We do NOT respect the subscribers that enjoy criticizing people.
.
- Follow-Ups:
- Re: Question re: Meta-Analysis
- From: Robert
- Re: Question re: Meta-Analysis
- References:
- Question re: Meta-Analysis
- From: Jason
- Re: Question re: Meta-Analysis
- From: David Rind
- Question re: Meta-Analysis
- Prev by Date: For MEM,MD and others
- Next by Date: Re: Question re: Meta-Analysis
- Previous by thread: Re: Question re: Meta-Analysis
- Next by thread: Re: Question re: Meta-Analysis
- Index(es):
Relevant Pages
|
