Publication bias
Publication bias was evaluated graphically by inspecting the asymmetry
of the funnel plots and statistically using Egger’s regression tests
(Sterne & Egger 2005). Significantly asymmetric results were then
augmented using the Trim and Fill method (Duval & Tweedie 2000). The
Trim and Fill method estimates the number of missing studies due to
publication biases, calculates their effect sizes and standard errors
and adds them to the meta-analysis dataset. In addition, the Rosenberg’s
fail-safe number was calculated. Rosenberg’s fail-safe number represents
the number of non-significant, unpublished studies that must be added to
a meta-analysis to change the result from significant to non-significant
(Rosenberg 2005). A fail-safe number that is larger than 5n + 10, where
n is the number of studies used in the meta-analysis, is considered
robust (Rosenthal 1986).