Figure 1

Schematic illustration of the potential impact of unpublished studies on a meta-analytic summary effect. The vertical line demarcates an effect size showing no difference between treatment groups for the outcome being studied. In this case, effect sizes to the right of the vertical line correspond to the intervention treatment group having more favorable outcomes than the control treatment group; all four published studies found the intervention treatment to be significantly better than the control treatment with respect to the study outcome, as indicated by having confidence intervals completely to the right of the vertical line. The first diamond shows the pooled estimate across the four published studies. If we then include results from unpublished studies in the meta-analysis, how might the conclusion of the meta-analysis change? The three diamonds at the bottom of the figure each represent a possible outcome of the pooled estimate across all studies (published and unpublished).