This text is
based on p 28 of Hemilä
(2006)
These documents have up to date links to documents that are available
via
the net.
Harri Hemilä
Department of Public Health
University of Helsinki,
Helsinki, Finland
harri.hemila@helsinki.fi
In general, ‘meta-analysis’ denotes systematic and thorough
investigation of scientific literature on a specific topic, and
combining the results of ‘close enough’ studies by statistical
formulae, but there is substantial difference of opinion as to how
people see the coverage of the term ‘meta-analysis.’ The term was
coined by Gene Glass to describe the process of synthesizing results
from separate but similar studies (Mann 1990).
Interestingly, the origin of the meta-analytic approach was connected
to vitamin C (Hampton 2002; Milne & Chalmers 2004), since in his
treatise on vitamin C deficiency, James Lind carried out a systematic
search of all the older literature and wrote that "As it is no easy
matter to root out old prejudices, or to overturn opinions which have
acquired an establishment by time, custom and great authorities; it
became therefore requisite for this purpose, to exhibit a full and
impartial view of what has hitherto been published on the scurvy; and
that in a chronological order, by which the sources of those mistakes
may be detected. Indeed, before this subject could be set in a clear
and proper light, it was necessary to remove a great deal of rubbish"
(1753 p 7; see Appendix 1).
Typically, meta-analysis is used (1) to increase statistical power for
primary end points and for subgroups, or (2) to improve estimates of
the size of the effect, or/and to (3) resolve uncertainty when reports
disagree (Sacks et al. 1987). The purpose of increasing statistical
power emerges from the problem that a large proportion of controlled
trials are so small that they cannot provide meaningful evidence for
the effectiveness or otherwise of therapy, simply because the
confidence intervals (CI) are very wide. This problem of low
statistical power was illustrated by Freiman, Chalmers, et al. (1978)
who analyzed 71 ‘negative’ trials (P > 0.05 for the difference of
interest), showing that 50 of them could have missed a 50% benefit
because the trials were simply too small. Thus, meta-analysis can be
used to enhance the use of data from small studies with ambiguous
results by combining the results of several to test whether there is
any overall evidence of effect, and to estimate its magnitude. This is
the most common use of meta-analysis.
When the optimism on the fruitful opportunities provided by
meta-analysis was high, Chalmers’ group believed that "A quantitative
synthesis of the data in similar randomized controlled trials can
potentially be more useful to the practicing physician than a
traditional narrative review article, but such a synthesis must be
properly performed to warrant serious attention" (Sacks et al. 1987).
Thomas Chalmers also commented that "Meta-analysis is the wave of the
future. The days of the expert supposedly putting the state of the
field into a review article are numbered" (Mann 1990).
References
Freiman JA, Chalmers TC, Smith H, Kuebler RR (1978) The importance of
beta, the type II error and sample size in the design and
interpretation of the randomized control trial: survey of 71 “negative”
trials. N Engl J Med 299:690-4 PubMed
Hampton JR (2002) Evidence-based medicine, opinion-based medicine, and
real-world medicine. Persp
Biol Med 45:549-68
Lind J (1753) A Treatise of the Scurvy, in Three Parts, Containing an
Inquiry into the Nature, Causes, and Cure of That Disease. Together
with a Critical and Chronological View of What Has Been Published on
the Subject. Edinburgh: Sands, Murray and Cochran. 456 pp. Republished
in: (1953) Lind’s Treatise on Scurvy [Stewart CP, Gutrie D, eds].
Edinburgh, UK: Edinburgh University Press [pages in the current thesis
refer to the 1953 reprint]
Mann C (1990) Meta-analysis in the breech. Science 249:476-80 * see
also: garfield.library
Milne I, Chalmers I (2004) Documenting the evidence: the case of
scurvy. Bull
WHO 82:791-2 * see also: (2004);82:793-6
Sacks HS, Berrier J, Reitman D, Ancona-Berk VA, Chalmers TC (1987)
Meta-analyses of randomized controlled trials. N Engl J Med 316:450-5
PubMed