Publication bias, that is, a lack of small studies without significant results, was addressed by the linear regression test for funnel plot asymmetry described by Egger et al.13.