Fig. 3. Total blood loss funnel plot assessing publication bias. Plotted is the standard error (SE) versus the log treatment effect of each study. Vertical line is summary log treatment effect for blood loss ( i.e. , log [0.84]=−0.17). Diagonal lines are expected 95% confidence interval (±1.96 SE) from the summary log treatment effect. Studies outside the funnel indicate heterogeneity. The fairly symmetric plot indicates low publication bias.