In my mind, the beautiful thing about evidence-based anything is that it
should be both evidence-based (with facts that can be checked by others) and
transparent (the way the facts were gathered can be reproduced by others in
the same way the original authors produced them). There are many reasons why
the 'true' systematic review process has become complicated, and highly
undesirable by the masses. Mainly they are built on the evidence that bias,
heterogeneity in interventions, comparators, populations, etc. regularly
occur between experiments, and even evident cases of fraud and data
manipulation have been seen in the literature. At the end of the day, you
have to use the best evidence you can find whether when preparing a
systematic review, a clinical guideline, or even giving advice to a patient.
If we start to believe that cutting corners is the same as doing the best
job that we can then we have a really big problem. We should be cutting
corners out of necessity, not because we believe it will produce the same
results. And even then explicitly state which corners we cut, not just sweep
it under the rug.
Let's compare this with a clinical trial. Dr. X decided to do a randomized
trial to see pain management strategy works best for patients with
rheumatoid arthritis. After getting a team of experts together, he realizes
that there is no way he can do a proper, adequately powered RCT with the
budget at hand or recruit the number of patients needed by the power
calculation. So he decides to do a retrospective chart review... Do we
honestly believe that the results of the chart review should be considered
the same as a well-performed RCT? Dr. X had to cut corners, and so will a
lot of 'systematic reviewers' so that they can finish their review on time
and on budget. But where is the evidence to support those decisions. Here
lies the biggest problem of all... there is a large gap of evidence on what
we should be doing and what we can omit during a systematic review. For
example, most people believe that study identification and inclusion should
be done by two independent reviewers, but what is the evidence behind that.
What if I told you that a randomized trial showed that no difference in
effect whether two individuals worked independently, or if one checked the
work of the other. What about blinding reviewers to journal and author
names, and on and on. Even some of the most respected guidelines on how to
undertake a systematic review are really just consensus agreements among the
top clinical epidemiologists. So before we start to sing the praises of
cutting corners and turning every systematic review into a narrative review
with the word systematic in the title, we need to truly understand the
impact of these decisions.
Ahmed
-----Original Message-----
From: Evidence based health (EBH)
[mailto:[log in to unmask]] On Behalf Of Amy Price
Sent: Friday, October 26, 2012 9:07 AM
To: [log in to unmask]
Subject: Re: Medline based systematic review
Jacob and Kev,
I must confess that for me this top down approach remains quite overwhelming
and intimidating from a personal labor perspective especially over the
multiple topics that would be needed by a general Physician. I have stayed
silent on this until now because I felt perhaps it was my personal concern
and that I just needed more expertise. This is likely still true but
nevertheless Jacob raises important issues. If we want those that practice
to be active as well as passive contributors how are we making that a
reality?
There is the aspect where we want the best quality and I understand the need
for quality. That being said all the search quality in the world will not
fix some of the bias problems we face because they are not initiated by the
doctor trying to find the best intervention for the patient but are
triggered by industry data games. This is indirectly evident in multiple
Cochrane reviews as well where researchers unknowingly quote research where
data reporting has not been transparent. In my field I know this because I
can spot the players and know the latest work but in areas where I most need
clinical expertise quickly this is not the case.
There is the drawback that if all this is beyond the reach of those with a
day job due to time and expertise constraints there will be few
contributions from them, the learning will be passive and their voice, the
voice of those in the trenches will be lost. Are their ways we could
simplify the process and be more inclusive without compromising quality?
Are those in daily practice and our junior Drs just meant to be passive
consumers of Up 2 Date and other commercially prepared solutions. If this is
the case how does it differ from the old textbook days and how will active
interest in EBM and consumption be sustained and integrity maintained over
time if the users role is reduced to that of only a passive consumer without
input?
Best,
Amy
On 10/26/12 1:09 AM, "Jacob Puliyel" <[log in to unmask]> wrote:
>Dear Kev
>You write
>"The conclusion seems to be that at least most of the time, a less than
>comprehensive search will give us the same outcome as a comprehensive
>search. Unfortunately, there may be occasions that this is not so.
>The trouble is, we cannot tell when those occasions will occur."
>
>Please let me suggest that by letting perfect become the enemy of the
>good, we may be doing more harm than good to the cause of EBM.
>
>EBM began as a ³bottom-up² paradigm that taught residents to ask
>answerable and focused questions, search the literature in a
>transparent and reproducible way to find the best evidence and to
>critically appraise it in an explicit and structured manner, often
>using mathematical analyses to give a clear idea of the strength,
>statistical significance and possible clinical significance of the
>results.
>
>Unfortunately it is no longer an amateursı enterprise.
>Journals now insist on numerous boxes being ticked before they will
>even consider an review article for publication.
>Multiple data bases have to be explored, the references in the papers
>need to be hand-searched for new references, clinical trials registers
>and conference proceedings scrutinised and pharmaceutical companies and
>individual researchers must have been contacted for unpublished data
>and ongoing trials.
>
>Only organisations with very deep pockets can afford this anymore.
>Vested interests and jumped into the void.
>Now we have even Cochrane Meta analysis, written up by persons with
>direct conflicts of interests declared.
>
>Als-Nielsen and colleagues have shown that association with for profit
>organisations had little impact on treatment effect but the conclusions
>were more positive due to biased interpretation of trial results
>(Als-Nielsen et al 2003). Lundh and colleagues have shown that
>publication of industry-supported trials was associated with an
>increase in journal impact factors and revenue (Lundh et al 2010).
>Smith (2010), the former editor of the BMJ, has suggested that
>publishing the RCT sponsored by one drug company could yield a million
>dollars in the sales of reprints alone.
>
>We can counteract this form of biased interpretation only if allow less
>comprehensive searches done be independent researchers to challenge
>these 'comprehensive' reports.
>
>I must declare that most of this stuff I had researched for paper on
>EBM published elsewhere (Evidence Based Medicine: Making It Better)
>
>Jacob Puliyel MD MRCP M Phil
>Head of Pediatrics
>St Stephens Hospital
>Delhi
>
|