Correcting for outcome reporting bias in a meta-analysis: A meta-regression approach