New approach to facilitate interpretation and presentation of network meta-analysis results




Oral session: Network meta-analysis


Wednesday 23 October 2019 - 11:00 to 12:30


All authors in correct order:

Sadeghirad B1, Brignardello Petersen R1, Florez ID1, Johnston BC2, Busse JW3, Guyatt GH1
1 Department of Health Research Methods, Evidence, and Impact, McMaster University, Canada
2 Department of Community Health and Epidemiology, Dalhousie University, Halifax, Nova Scotia, Canada
3 Department of Anesthesia, McMaster University, Hamilton, Ontario, Canada
Presenting author and contact person

Presenting author:

Behnam Sadeghirad

Contact person:

Abstract text
Background: as the number of treatments being compared increases, interpretation and presentation of results of a network meta-analysis (NMA) becomes more challenging. The challenges increase further when the NMA deals with multiple outcomes. Efforts thus far have failed to produce a fully satisfactory approach to interpretation and presentation of NMA results under these conditions.

Objectives: to provide guidance on how to draw conclusions regarding which treatments are more likely to be superior or inferior to others in terms of effectiveness and harm, considering the estimates of effects, certainty of evidence, and rankings, and to ascertain an intuitive understanding of NMA results.

Methods: we categorized the interventions - from the most effective to the least effective - based on the effect estimates obtained from the NMA and their associated GRADE certainty of evidence. For each outcome, we created groups of interventions as follows:
1) the reference intervention (placebo) and interventions no different from placebo (i.e. 95% confidence interval (CI) includes null value), which we refer to as "among the least effective/no more harmful than placebo";
2) interventions superior to placebo, but not superior to any other of the intervention(s) superior to placebo (which we call category 2 and refer to as "inferior to the most effective, but superior to the least effective/less harmful than some alternatives, but more harmful than placebo"); and
3) Interventions that proved superior to at least one category 2 intervention (which we call “among the most effective/most harmful”).

We then divided all three categories into two groups: those with moderate- or high-certainty evidence relative to placebo, and those with low- or very low-certainty evidence relative to placebo.

Results: we illustrate our new method using two challenging NMAs of randomized trials focused on outcomes of benefit only (Probiotics, prebiotics, and synbiotics for prevention of mortality and morbidity in preterm infants: an NMA of randomized trials (with 9 treatments and 8 outcomes); Table 1) and benefit and harm outcomes (Management of acute, non-back, musculoskeletal pain: an NMA of randomized trials (with 31 treatments and 8 outcomes); Table 2). Furthermore, using our methods, we can rank treatments based on their effectiveness and safety.

Conclusions: our new visualisation approaches can be used by NMAs to gain a more intuitive understanding of results that consider certainty (quality) of evidence, balance of effectiveness and harms, and importance of outcomes.