PMC

Methods of vascular occlusion:blood transfusion (red blood cells)
Forest plot of the comparisons in which direct and indirect estimates were available. There do not appear to be any discrepancies between direct and indirect estimates, although the credible intervals are different (the direct evidence had narrower credible intervals in four of the five comparisons above) resulting in the differences in the comparisons in which there was evidence for difference. Direct evidence appears to be preferable over indirect evidence and network meta‐analysis based on the quality of evidence for the comparison 'continuous selective portal triad clamping versus continuous portal triad clamping'. Indirect evidence and network meta‐analysis appear to be preferable over direct evidence for the comparison 'continuous portal triad clamping versus control'. Direct evidence and network meta‐analysis appear to be preferable over indirect evidence for the comparison 'intermittent portal triad clamping versus control'. There was little apparent difference in the quality of evidence between direct, indirect estimates, and network meta‐analysis; so, we could not choose one estimate over the others based on the quality of evidence.
¹Risk of bias was unclear or high in the trial(s) (downgraded by 1 point).
 ²Sample size was low (downgraded by 1 point).
 ³Confidence intervals spanned no effect and clinically significant effect (downgraded by 1 point).

Direct and indirect evidence.
In the literature we can found RCT directly comparing interventions (e.g. A versus B in green; and B versus C in orange). Each circle represents an intervention and lines represent direct comparisons. Dashed lines are for indirect comparison. An global effect value is generated for each comparison (direct or indirect). Indirect evidence is generated by using B as common comparator for the comparison of A versus C (model proposed by Bucher). Network meta-analysis combining both direct and indirect evidence may be built.

Rows correspond to network meta-analysis ORs (separated for mixed and indirect evidence) and columns correspond to direct meta-analysis ORs. The contribution of each direct comparison to the total network evidence that provides the ranking of the treatments is presented separately (row named Entire network). The sizes of the boxes are proportional to the percentage contribution of each direct estimate to the network meta-analysis estimates (rows 1–6) and to the entire network (row 7). The last row shows the number of included direct comparisons. The names of the treatments are given in .

Graphs showing how the relative treatment effect (eg, log odds ratio) for Treatment 3 vs Treatment 2 could change with a covariate value with separate lines representing direct evidence (from trials that allocated Treatments 2 and 3), indirect evidence (from the remaining trials), and all evidence in various scenarios: A, there is no treatment by covariate interaction based on all evidence, and the relative treatment effects at 0 covariate are consistent, and the regression coefficients for the treatment by covariate interaction are consistent; B, there is an interaction based on all evidence, and the relative treatment effects at 0 covariate are consistent, and the coefficients are consistent; C, there is no interaction based on all evidence, and the relative treatment effects at 0 covariate are consistent, and the coefficients are inconsistent; D, there is an interaction based on all evidence, and the relative treatment effects at 0 covariate are consistent, and the coefficients are inconsistent; E, there is no interaction based on all evidence, and the relative treatment effects at 0 covariate are inconsistent, and the coefficients are consistent; F, there is an interaction based on all evidence, and the relative treatment effects at 0 covariate are inconsistent, and the coefficients are consistent; G, there is no interaction based on all evidence, and the relative treatment effects at 0 covariate are inconsistent, and the coefficients are inconsistent; and H, there is an interaction based on all evidence, and the relative treatment effects at 0 covariate are inconsistent, and the coefficients are inconsistent. Direct, indirect, and all evidence is overlapping in plots (A) and (B)