study design, characteristics of participants, intervention (such
as disease or healthy control, exposure or not) and outcome
measures.- The profiling studies included in this example employ micro-
arrays or PCR. For quality assessment, MIAME and MIQE
would be appropriate. The choice of the quality assessment
tools or scales is based on the design of included studies in
specific fields. For example, randomized controlled trials
should be evaluated according to the Cochrane Collaboration
tool for risk of bias assessment tool while primary diagnostic
accuracy studies should be evaluated according to the
QUADAS-2 in medicine. - Meta-analyses in the health/medical sciences are often based
on the data in 22 tables (Fig.9):
where ai, bi, ci, and di denote the cell frequencies and n1i and
n2i the row totals in the ith study. In our example, they are
based on the numbers of dysregulation events in both T2D and
nondiabetic control samples. log odds ratio is equal to the log
of (aidi)/(bici). Suppose an miRNA is reported upregulation
in a study, where the sample size is 6 both in type 2 diabetes and
nondiabetic control groups, then the logOR is equal to the log
of (66)/(00). A cell entry with zero value can be problematic
especially for odds ratio. Adding a small (negligible) constant
to the cells of the 22 tables is a common solution to this
problem. For example, adding 0.1 to the cell entries with
zero values, then the logOR will equal to 3.11. It should be
noted that different software packages might deal with zero
values differently.
7.Pvalue correction methods include the Bonferroni (bonfer-
roniin R) method in whichPvalues are simply multiplied by
the number of comparisons. Less conservative corrections
include Holm (1979) (holm), Hochberg (1988) (hochberg),
Hommel (1988) (hommel), Benjamini and Hochberg (1995)
(BH or its alias fdr), and Benjamini and Yekutieli (2001)
(BY) methods. Researchers could choose a correction method
according to their study designs and purposes.
Fig. 9A22 table482 Hongmei Zhu and Siu-wai Leung