| fate |
2013-06-11 17:03 |
There is no biological justification of using all these overlapping genetic models: additive, recessive and dominant models. All these models are significant and therefore, the authors cannot really show the mode of inheritance: is the mode of inheritance additive, recessive or dominant?
-The authors should estimate the mode of inheritance based on the h-index (see Zintzaras and Santos 2011 BMC Med Res Method; Zintzaras and Santos 2012 Stat Appl Genet Mol Biol).
-An additional analysis should be performed using the generalized odds ratio (ORG) metric. The ORG is powerful model-free approach which utilizes the complete genotype distribution and provides an estimate of the magnitude of the association between disease status and genotype (see Zintzaras 2010 Stat Appl Genet Mol Biol; Zintzaras 2012 J Appl Stat)
-A flow-chart indicating the included/excluded studies from each database is not provided.
-The issue of multiplicity is not addressed.
-The existence of differential magnitude of effect in large versus small studies was not examined.
-The authors should examine whether there is enough evidence to deny of claim association.
-HWE should be tested with an exact test.
-In studies not in HWE, an adjustment for HWE deviation should be performed.
-The funnel plot may be misleading in proving publication bias and the Harbord's test should be adopted (see Lau BMJ).
-Meta-regression doe not really offers anything in the meta-analysis of GAS since a subgroup analysis was performed.
|
|