Figure 2.

Funnel plot analysis of C. difficile prevalence reports and distribution of studies by the number of food items tested per study. Funnel plots for bias. (A) Standard plot of weighted estimated prevalence vs. standard error of estimated prevalence, however, such strategy is misleading toward suggesting there is publication bias when used on proportion based meta-analyses if the reported effects are close to zero (22). (B) Weighted estimated prevalence vs. sample size (22). (C) Histogram. Study distributions categorized based on total number of food/nonfood item categories tested in each study.