Definitions
| Publication bias | The publication or non-publication of research findings, depending on the nature and direction of the results.22 |
| Certainty in the evidence | The certainty that a true effect lies on one side of a specified threshold or within a chosen range supporting a decision.4 |
| Selection model | A weight function of effect size or p value is used to model the probability of publication. This method highly depends on this weight and is usually recommended as a sensitivity analysis. |
| Begg test | A method that uses the rank test to examine the association between the observed effect sizes and their variances. However, it suffers from low statistical power. |
| Egger test | A method in which we regress the standardised effect size against the precision. The intercept is close to zero if no publication bias is present. This method may have inflated false-positive rates for ORs. |
| Trim and fill method | A method in which the missing studies are imputed to provide a bias-adjusted effect estimate. However, it requires the strong assumption that the missing studies have the most negative (or positive) effect sizes. |
| Skewness | A method that examines the asymmetry of residuals of the regression test. It has more statistical power than other tests. However, it may lose power if the available studies have a distribution that tends to have multiple modes. |