Tests for Publication Bias

Because of the potential for 'publication bias phenomenon', in which authors/sponsors may differentially or preferentially disclose trials results or publish non-damaging studies in scientific journals, thus leading to potentially biased results, we conducted several tests of publication bias. However, as noted in meta-analysis references (by Egger and by Rothman, see left column), these tests may be underpowered and unable to detect publication or reporting bias even when such bias exists. Thus, results of tests for publication biases are only suggestive, but not conclusive.

Please select the COX-2 inhibitor to see its 'tests for publication bias' :

Rofecoxib

Celecoxib

Valdecoxib and Parecoxib

Etoricoxib


*Lumiracoxib was not examined due to limited number of studies