At 14:59 01/11/1999 -0500, Mark Ebell wrote:
>Hi,
>
>The Centre for EBM "Levels of Evidence" are widely used and very useful.
Thanks for the vote of confidence!
>However, the recent article in JAMA by Lijmer identifies more objective
>criteria for the evaluation of diagnostic test studies. In particular, the
>study design (case-control vs cohort) is very important,
We currently describe the study design as 'an appropriate spectrum' (i.e. a
cohort of patients suspected to have the target disorder) and an
'inappropriate spectrum' (i.e. some patients with the disease compared to
some without). Case-control studies won't affect the likelihood ratios
reported, but prevent calcuation of pre- and post-test probabilities.
>while the issue of
>consecutive vs non-consecutive is not.
Non-consecutive relates to the potential bias of selecting only certain
types of patients. Some diagnostic studies will miss out patients with
non-diagnostic tests, consequently making the tests appear better than in
reality. Diagnostic studies need to match real clinical situations as
closely as possible otherwise the test characeteristics will be inaccurate.
There are other discrepancies
>between the CEBM levels and Lijmer's data.
We'll check these out!
cheers
chris
>Are the folks who maintain that site planning to update it given this new
>information? As we begin another meta-analysis of diagnostic tests, it is
>tempting to develop a scoring system based on Lijmer's criteria...
>
>Thanks,
>
>Mark Ebell
-----------------------------------------------------------------------
Dr. Christopher Ball mobile: +44 (0)41 041 4003
Senior Editor e-mail: [log in to unmask]
Evidence-based On-call http://eboc.jr2.ox.ac.uk
Cairns Library
Level 3, Academic Street
John Radcliffe Hospital
Oxford OX3 9DU
UK
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|