2. Explain: -What we meant by a meta-analysis?
-Why we should do a meta-analysis?
-When we can do a meta-analysis?
-How we can do a meta-analysis?
3. Meta-analysis did not begin to
appear regularly in the medical
literature until the late 1970s
but since then a plethora of
meta-analyses have emerged
and the growth is exponential
over time (Figure 2).
Moreover, it has been shown
that meta-analyses are the most
frequently cited form of clinical
research.
4.
5. The Greek root ‘meta’ means ‘with’, ‘along’
‘after’ or ‘later’, so here we have an analysis after
the original analysis has been done.
Meta-analysis is a statistical technique, or set of
statistical techniques, for summarizing the
results of several studies into a single estimate.
7. Meta-analysis in another word is the process of
combining a number of separate studies to
produce one ‘superstudy’.
So, for example, we might have three separate
studies, with sample sizes of 40, 80 and 150.
When combined, we get a super-study with a
sample size of 270.
8. If we follow the assumption of the meta-analysis is
that this super-study will provide a more reliable and
precise overall result for the output variable in
question, than do any of the smaller individual studies.
9. To do a meta-analysis, we must have more than one
study which has estimated something, and they must
satisfy the homogeneity criterion.
The participants, interventions, or risk factors, and
settings in the studies carried out need to be
sufficiently similar for us to say there is something in
common for us to investigate.
10. Estimate summary effect estimate, and doing a forest
plot.
Determine the extent to which the articles are
heterogeneous.
Assess Publication Bias.
11. To identify statistical heterogeneity, we can test the
null hypothesis that: the studies all have the same
treatment (or other) effect in the population.
This statistic, usually denoted by Q, gives a chi-
squared test with degree of freedom = number of
studies – 1.
12. If there is significant heterogeneity, then we have
evidence that there are differences between the
studies. It may therefore be invalid to pool the results
and generate a single summary result.
We should try to describe the variation between the
studies and investigate possible sources of
heterogeneity. We should not just ignore it, but try to
account for the heterogeneity in some way.
13. If we can explain the heterogeneity, we may be able to
produce a final estimate of the effect which adjusts for
it. If not, we can also carry out meta-analysis which
allows for heterogeneity, called random effects
analyses.
If the heterogeneity not significant, we have little or no
statistical evidence for differences between studies.
14. However, the test for heterogeneity has low power. The
number of studies is usually low and the test may fail
to detect heterogeneity as statistically significant when
it exists. As with any significance test, we cannot
interpret a not significant result as evidence of
homogeneity.
To compensate for the low power of the test some
authors accept a larger P value as being significant,
often using P < 0.1 rather than P < 0.05.
15. Quantification of heterogeneity is only one
component of a wider investigation of variability
across studies, the most important being diversity in
clinical and methodological aspects.
Meta-analysts must also consider the clinical
implications of the observed degree of inconsistency
across studies. For example, interpretation of a given
degree of heterogeneity across several studies will
differ according to whether the estimates show the
same direction of effect.
16. I square categorized as low, moderate, and high to I
square values of 25%, 50%, and 75%.
An alternative quantification of heterogeneity in a
meta-analysis is the among-study variance (often
called tau squared ), calculated as part of a random
effects meta-analysis.
17.
18. Although the intent of a meta-analysis is to find and
assess all studies meeting the inclusion criteria, it is
not always possible to obtain these.
A critical concern is the papers that may have been
missed. There is good reason to be concerned about
this potential loss because studies with significant,
positive results (positive studies) are more likely to be
published and, in the case of interventions with a
commercial value, to be promoted, than studies with
non-significant or “negative” results (negative studies).
19. To assess the Publication Bias , we can use graphs
(Funnel Plot), or statistical tests (Begg’s &
Mazumadar’s test, Egger’s test).
There are some ways can be used to correct the effect
of publication bias:
- trim and fill method
- selection model
- meta-regression
20. Meta-analysis can be performed in various general
statistical and numerical analysis environments (e.g.,
Stata, R/ Splus, Octave/MATLAB), or in dedicated
programs (e.g., the Microsoft DOS version of Meta-
Analyst, Comprehensive Meta-Analysis, RevMan,
MIX).
22. Please see the handout, try to solve the questions and
see the model answers in the last papers.
23. 1. Bland JM. An Introduction to Medical Statistics, 4th edition. Oxford Oxford
Univ Press. 2015;
2. Bowers D. Medical statistics from scratch : an introduction for health
professionals. John Wiley & Sons; 2008. 284 p.
3. Basu A. How to conduct meta-analysis: A Basic Tutorial. 2017 [cited 2017 Nov
27]; Available from: https://peerj.com/preprints/2978v1.pdf
4. Basu A. Introduction to Meta Analysis. 2014 [cited 2017 Nov 27]; Available
from: https://peerj.com/preprints/665v1.pdf
5. Ab H, Haidich A-B. Meta-analysis in medical research. HIPPOKRATIA
PASCHOS KA HIPPOKRATIA [Internet]. 2010 [cited 2017 Nov 27];14(14).
Available from:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3049418/pdf/hippokratia-
14-29.pdf
6. Higgins JPT, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency
in meta-analyses Need for consistency. [cited 2017 Nov 27]; Available from:
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC192859/pdf/3270557.pdf