To add to what Peter has said, I think an expanded review can help to look
at issues of appropriateness e.g. the service may be effective but is it
going to be appropriate for different types of populations?
I was part of a team using systematic review evidence as well as many other
sources to work with different communitites to decide upon the best
configurations for delivering HIV/AIDS prevention services across different
countries (paper to be presented at Cochrane conference). The systematic
review evidence was very much part of an expanded review which looked at
existing service configurations and acceptability and feasibility of using
review evidence in different contexts.
Janet Harris
Academic Director, Health Sciences
University of Oxford
Centre for Continuing Professional Development
16/17 St. Ebbes Street
Oxford OX1 1PT
(44) 1865 286942
> -----Original Message-----
> From: Peter Griffiths [mailto:[log in to unmask]]
> Sent: 23 October 2003 09:13
> To: [log in to unmask]
> Subject: Re: Another view of expanded systematic reviews
>
>
> Badri,
>
> I don't think that you are being naive at all - your point is
> a very real
> one. Essentially we are addressing a blurry boundary between
> primary and
> secondary research.
>
> The introduction of a full blown review process might make it look a
> rather different but there comes a point when what you end up
> with might
> be indistiguishable in principle from a paper / report with a
> structure
> that has a literature review that identifies gaps in the
> exisitng research
> (done in a sytstematic manner)and then presents some research
> to fill the
> gap...Maybe we shouldn't call it a review but I'm not sure it matters.
>
> I do think that there are issues that arise when reviewing in
> muddy waters
> that mean that it is necessary to expand the process but (in
> my opinion)
> keep things firmly on the 'review' side of the primary
> secondary research
> divide.
>
> Firstly it is important to remember that cochrane methods are
> developed
> for questions of effectiveness and while there are generalisable
> princoiples the less your question looks like one of effect the less
> specific guidence there is.
>
> Secondly even where there are clear questions of
> effectiveness reviews
> that examine 'interventions for problem X' always risk making
> a type II
> error that is entirely hidden. A competent reviewer should be able to
> clearly distinguish evidence of no effect from lack of evidence of
> effectiveness where there is evidence to review. However
> where there is no
> evidence or no evidence that comes close to meeting quality
> criteria the
> danger is that practices that occur are totaly omitted from
> mention in the
> review. In these cases extending the review in order to
> identify questions
> that have not even been addressed in the literature can be important.
>
> If you were talking about (say)treatment of MI you could
> argue that it is
> unlikely that any treatment with a valid claim to be considered is
> unlikely to be so hidden (I'm not sure which side of this I'd like to
> argue mind!). However if you are reviewing (as we did) a topic like
> services to facilitate the transition from child to adult
> care services by
> only looking at valid evaluations you would simply fail to
> identify many
> service configurations in your review (and be able to
> conclude that there
> was no evidence of effectiveness.
>
> Similarly if you restricted yourself to publications or
> written sources
> alone alone you would run the risk of failing to identify many of the
> things that are out there and at best get (and give) a very incorrect
> impression of the relative imporatnce of certain services. Again in
> classic review terms this translates into failing to say that
> there is no
> evidence for the effectiveness of an intervention that jmight be quite
> common.
>
> ....too many words!
>
> Peter
>
|