Please click here to obtain permission to reproduce this image.
Click on image to view larger version.
Figure 3 Funnel Plots to Assess Publication Bias
The dashed line is the fixed-effects estimate of change in CRP across interventions, and the dotted lines are pseudo-95% confidence intervals for the standard error of change in CRP, assuming no heterogeneity between studies. Bias is suspected if the plots are asymmetric (i.e., the point estimates tend to fall to one side of the dashed line). (A) Significant bias in all interventions (Egger p < 0.001). However, stratification by the median change in LDL showed no evidence of bias. (B) LDL >40% (Egger p = 0.5). (C) LDL  40% (Egger p = 0.7). Abbreviations as in Figure 2.
|
