Fig. 7. Late pain (24 h) at rest funnel plot assessing publication bias. Plotted is the SE versus standard difference in mean (Effects). Vertical line is the combined effect for early pain with diagonal lines representing the expected 95% CI from the combined effect. Studies outside the funnel indicate heterogeneity. The funnel demonstrated moderate asymmetry (P = 0.01, one-sided) with 5 of the 25 studies outside the 95% CI, with 2 favoring placebo and 3 favoring dexamethasone. There was no asymmetry in funnel plots for low-, moderate-, or high-dose dexamethasone subgroups.