Abstract

This article describes GRADE (grading of recommendations assessment, development and evaluation) guidance on how to make conclusions from a network meta-analysis of interventions that includes individual randomised controlled trials for one outcome at a time. The guidance is based on a partially contextualised approach in which review authors must establish ranges of magnitudes of effect that represent a trivial to no effect, small but important effect, moderate effect, and large effect. The principles guiding this framework are that interventions should be grouped in categories, based on the magnitude of the effect; and that the judgments that place interventions in such categories should consider the estimates of effect, the certainty of the evidence, and the rankings. We describe and illustrate the four steps of this framework using an example.