Hi All
Dan makes a series of good points, all of which I would support.
I would add that there are a number of systems out there that provide a
way to rank levels of evidence/grades of recommendation - which makes
things hard for us sometimes. Although rather complicated and perhaps
loosing something in user friendliness the CEBM at Oxford (Olive & Co)
presented a system which may assist in addressing your question. See
http://www.cebm.net/levels_of_evidence.asp, perhaps level 2a or 3a fit
(depending on the design of the observational studies).
Finally, a comment on MA versus SR. In my book it is often the case
that a good systematic review trumps a meta-analysis. I think we are
sometimes too keen to lump data together, even when we know that it
doesn't really fit. Even a good statistician can't make a silk purse
out of a sow's ear!
Jeff
Jeff Harrison
Senior Lecturer, School of Pharmacy
Faculty of Medical and Health Sciences
Building 504, 85 Park Road, Grafton, Auckland
Phone +64 9 373 7599 ext. 82144
Fax +64 9 367 7192
-----Original Message-----
From: Evidence based health (EBH)
[mailto:[log in to unmask]] On Behalf Of Dan Mayer
Sent: Friday, 1 July 2005 4:52 a.m.
To: [log in to unmask]
Subject: Re: Level of Evidence Assistance
Hi Maria and list.
I think that your question is at the heart of the problems with EBM.
How can you have a meta analysis if there is only one study? By
definition, you need several studies to have one. On the other hand, a
systematic review is simply a review of all the evidence generated on a
clinical question to date and does not depend on the type of studies.
Each study in a systematic review stands on its own merits and then the
author puts them all together and draws a final conclusion on the
validity of them all and how the results ought to be 'spun'. It is now
up to you (the reader) to evaluate the methods of the systematic review
and decide if it is strong enough for the results to be useful.
Sorry if this is a round about way of saying 'we really don't know'.
However, I would look at all the studies and if they all say the same
thing (more or less) and there are no obvious FATAL FLAWS in any of
them, accept the conclusions as strong evidence. The strength of that
evidence (or level if you like) would diminish as the quality and effect
size of the studies decreases.
Hope this helps,
Dan
************************************************************************
****
Dan Mayer, MD
Professor of Emergency Medicine
Albany Medical College
47 New Scotland Ave.
Albany, NY, 12208
Ph; 518-262-6180
FAX; 518-262-5029
E-mail; [log in to unmask]
************************************************************************
****
>>> Olive Goddard <[log in to unmask]> 06/30
8:58 AM >>>
Dear Colleagues,
Would any of you care to respond to Maria.
All good wishes,
Olive
>>> "Maria Shepherd" <[log in to unmask]> 06/29/05 7:21 pm >>>
Greetings-
I am doing a project for my physical therapy schooling, and I wanted to
ask for your input/assistance. I have two articles, both of which are
systematic reviews. However, they each only examine one RCT, and the
remaining articles reviewed (6 in one article and 10 in the other) are
empirical/observational studies which don't have control groups. All of
the findings were positive and concluded the same thing. I'm having
trouble assigning a level of evidence to these two systematic reviews
because they do only have one RCT. Would you mind providing me with
some input as to what level of evidence you might think they would be
and why? I appreciate your assistance.
Thank you.
Maria Shepherd, PT
|