Hi all,
It seems that the growing volume of raw information is at odds with the
methods of Systematic Reviews to keep up as Dr. Carl Heneghan blogged on
here
http://www.carlheneghan.com/2012/263/how-many-cochrane-reviews-are-needed-t
oo-many/#comment-175 . I wonder if Big Data mining could be programmed to
automate some of the tasks so as to free up those with high skill levels
in systematic reviewing to cover more ground with less effort.
Learning and reaping the value of well executed searches and the
organization of material is important. This has laid the groundwork for
much of the trials reform and knowledge of what is evidence that we have
today. I have on occasion used CASP to show members of the public what to
look for in a paper even when they are lacking subject expertise. I agree
with Nick that no one learns it all in a few modules but those modules can
still frame a way of life that includes applied evidence and at least
alerts us to the possibility that bias exists.
Could social media be helpful in finding which systematic reviews and the
judicious updating of existing will make the most impact?
Best,
Amy
On 3/20/13 6:44 PM, "Bastian, Hilda (NIH/NLM/NCBI) [C]"
<[log in to unmask]> wrote:
>G'day!
>
>The line between fully systematic and all other reviews isn't a precise
>one, but I don't think the issue of just searching trials at registry
>agencies is a viable alternative as a methodology. Firstly, not all
>questions are even relevant here (not every clinical question is about a
>drug or device). Secondly, regulatory agencies do not try to keep their
>dossiers up-to-date for every drug or device - especially for trials that
>aren't sponsored by the manufacturer.
>
>We faced the same kinds of issues when we started to look at how much
>work you have to do for an update. And the problem is, we still don't
>know how to predict which are the ones for which less work will be the
>same as more work. It's a minority of cases where being less systematic
>will make an important difference and so it's not hard to find samples
>where it made no difference. The key question it seems to me is to know
>prospectively when being more systematic will make a difference. At least
>some way of narrowing this down will likely emerge through the work of
>the health technology assessment agencies.
>
>Hilda
>
>________________________________
>From: Jon Brassey [[log in to unmask]]
>Sent: Wednesday, March 20, 2013 6:15 AM
>To: [log in to unmask]
>Subject: Reviews
>
>Hi All,
>
>You may remember I asked about the purpose of systematic reviews the
>other month. I'm still trying to understand the area and one connected
>area is that of language. We say 'systematic reviews', 'rapid reviews'
>etc. But these are just umbrella terms which can be helpful and can also
>mis-lead.
>
>Take the term systematic review (SR) - as we all know there are many
>different methods. A recent big contrast is between those that use
>regulatory data and those that don't - yet they are both SRs. If you
>examine the work of Tom Jefferson on Tamiflu the difference between his
>SR without regulatory data (his initial Cochrane review) and with (the
>follow-up Cochrane SR) the difference was profound. So, I can't help
>feeling the term is unhelpful - other than broadly indicating that the
>review was long/timely etc.
>
>Then, an area I've been looking at the literature on, rapid reviews.
>These are less in-depth than SRs but I dare say the results are broadly
>comparable. Finally, the world I'm actively involved in 'ultra rapid
>reviews' - what are they? To me they are reviews done in a couple of
>days maximum.
>
>So, to my mind, these terms are not hugely helpful - am I alone in that?
>
>I wonder of there is a wider need to codify things? So, you could break
>it down for each component e.g. search. This could range from searching
>a single database through to multiple databases and the 'ironman' being
>using regulatory data? You could do that for all the components.
>
>What would then be helpful would be to compare the results of each
>different component on the subsequent end-point. You could then make a
>reasonable estimate of the cost-benefit of moving up to the next stage of
>search, appraisal, synthesis etc.
>
>Any thoughts?
>
>BW
>
>jon
>
>--
>Jon Brassey
>Trip Database
>http://www.tripdatabase.com
>Find evidence fast
>
|