Approaches for setting thresholds or ranges for certainty of evidence ratings (adapted from Hultcrantz et al, 2017)38
| Setting | Contextualisation | Threshold or range | How to set | What certainty rating represents |
| Primarily for systematic reviews and health technology assessment | Non-contextualised | Range: 95% CI | Using existing limits of the 95% CI, which implies that precision is not routinely part of the rating | Certainty that the effect lies within the CI |
| OR≠1; RR≠1; HR≠1; RD≠0 | Using the threshold of null effect | Certainty that the effect of one treatment differs from another | ||
| Primarily for systematic reviews and health technology assessment | Partly contextualised | Specified magnitude of effect | For example, small effect is the effect small enough to not use the intervention if adverse effects/costs are appreciable | Certainty in a specified magnitude of effect for one outcome (eg, trivial, small, moderate or large) |
| Primarily for practice guidelines | Fully contextualised | Threshold determined with consideration of all critical outcomes | Considering the range of effects on all critical outcomes, and the values & preferences for those ranges | Confidence that the direction of the net effect will not differ from one end of the certainty range to the other |
RD, risk difference; RR, risk ratio.