Making the GRADE approach for network meta-analysis more efficient




Long oral session 17: Network meta-analysis methods


Friday 15 September 2017 - 11:00 to 12:30


All authors in correct order:

Brignardello-Petersen R1, Bonner A1, Alexander P1, Guyatt G1
1 McMaster University, Canada
Presenting author and contact person

Presenting author:

Romina Brignardello-Petersen

Contact person:

Abstract text
Background: Assessing the certainty of the evidence from a systematic review is a crucial task to draw appropriate conclusions. In systematic reviews that conduct network-meta analysis (NMA), this task can be perceived as onerous and time-consuming.

Objectives: To describe the conceptual advances that could improve efficiency when using the GRADE approach to assess the certainty of the evidence from NMA.

Methods: Members of the GRADE NMA working group have been working on strategies to improve efficiency when assessing the certainty of the evidence from NMA. We have conducted brainstorming sessions, testing of strategies in different networks, and incorporation of feedback obtained at GRADE working group meetings and other research meetings.

Results: We have identified and provided guidance with regards to three main strategies directly related to efficiency: 1) it is not necessary to address imprecision when rating the direct and indirect estimates that inform the rating of a network estimate; 2) it is not necessary to assess the indirect evidence when the direct evidence has high certainty, and contributes to the network estimate as much as the indirect evidence; and, 3) statistical tests of global incoherence are insufficient to address incoherence at the pairwise comparison level.

Conclusions: The uptake of GRADE to assess the certainty of estimates from NMA may be threatened by the effort required. We have developed strategies to improve efficiency that systematic reviewers can apply with due care to not compromise the quality of their assessments.