| |
Why
the BMG was established
It was recognized that a deeper understanding of bias was
required so that previously published systematic reviews and
meta-analyses could be evaluated, and evidence-based guidelines for the
conduct of future reviews and meta-analyses be developed. Such
guidelines would be important aids for the reviewer in deciding, at the
outset of a review, how to frame the review question, how to search the
literature and how to interpret the findings of each report identified
in the search.
Moreover, there is a need for a group within
the Cochrane Collaboration, which will promote the prospective
registration of primary studies as a prophylactic measure to prevent or
reduce the effects of bias.
In 1998, an initial meeting was held in
Baltimore, MD USA at the 6th Cochrane Colloquium. A further exploratory
meeting was held in Rome at the 7th Cochrane Colloquium in 1999. At the
8th Cochrane Colloquium in Cape Town in 2000 a pre-registration meeting
was held and the RBMG was formally represented at other relevant
sessions, including "meet the entities" and the annual general meeting.
Now formally registered as a methods group, the RBMG met at the 9th
Cochrane Colloquium in Lyon in 2001 and again at the systematic review
symposium in July 2000, Oxford, U.K. In September 2003, the RBMG
officially relocated from Bristol, UK to Ottawa, Canada. Recently, the
RBMG gathered at the 11th Cochrane colloquium in Barcelona 2003.
What
is the BMG?
The Bias Methods Group (BMG) focuses on the "epidemiology of
results and outcomes", that is, the characteristics of primary study
results and their distribution within and outside the literature. There
is a risk of bias being introduced into a systematic review and/or
meta-analysis if the characteristics and distribution of results
(concerning a particular treatment) are associated with the nature and
direction of the results (degree of harm or benefit, statistical
significance, precision) themselves, and if the reviewer does not take
account of these associations. There is now evidence at a general level
that such associations do exist.
The BMG investigates a range of
different forms of bias, such as "publication bias", "language bias",
and "outcome variable bias", and examines whether, and in which
circumstances, these biases may have a substantial impact on systematic
reviews and meta-analyses. The BMG will do this with regard to
the broad categories of quantitative and non-quantitative syntheses,
and with regard to experimental and observational research. Special
regard will be given to individual topic areas and Cochrane Review
Group interests. The BMG would like to conduct methodological
systematic reviews, provide advice to Cochrane entities, hold workshops
and publicize its findings through appropriate means.
Liaison
with Cochrane Groups, Networks & Centers
Bias is now the focus of an existing Methods Group, and is a topic of
interest to many Cochrane reviewers.
The BMG includes a large
proportion of Cochrane Collaborators with an active interest in this
area. The BMG also views links with the Cochrane Consumer
Network as important. Members of the Network should be encouraged to
join the BMG or be included on its mailing list. Biases such as outcome
variable selection bias, developed country biases and reviewer bias,
(as described above), are likely to be of particular interest to
consumers, as the nature of these biases is determined by the behaviour
and mindset of health professionals, researchers and editors.
The BMG, in turn, will benefit greatly from the perspectives
of consumers on these biases.
BMG Module
To access the current BMG module please visit
The Cochrane Library.
|
|
