How to rate the certainty of evidence with GRADE
Short answer
Rate certainty separately for each important outcome. Evidence from randomized trials starts at high certainty and from observational studies at low; rate down for risk of bias, inconsistency, indirectness, imprecision or publication bias, and, for observational evidence, consider rating up for a large effect, a dose-response gradient or plausible residual confounding. The result is high, moderate, low or very low certainty, presented in a summary of findings table.
Step by step
- 1
Select the outcomes that matter
Choose the outcomes that are important to patients and decision makers, ideally in the protocol, and rate certainty for each of them separately. GRADE rates a body of evidence for an outcome, not individual studies.
- 2
Set the starting level
A body of randomized trials starts at high certainty and a body of observational studies at low certainty. GRADE guidance also allows non-randomized studies assessed with ROBINS-I to start at high, in which case risk of bias will usually lead to rating down.
- 3
Consider the five reasons to rate down
Risk of bias in the studies that contribute to the outcome; inconsistency of results between studies; indirectness of the population, intervention, comparator or outcome relative to your question; imprecision, judged mainly from the confidence interval; and publication bias. Each can lower certainty by one or two levels.
- 4
Consider the three reasons to rate up
Mainly for observational evidence that has not been rated down: a large magnitude of effect, a dose-response gradient, and situations where all plausible residual confounding would reduce the demonstrated effect, or suggest a spurious effect when none was observed.
- 5
Assign the final level and explain it
Express the result as high, moderate, low or very low certainty. Explain every decision to rate down or up in a footnote, so readers can see how you reached the rating.
- 6
Build the summary of findings table
Present, per outcome, the number of participants and studies, the relative and absolute effects and the certainty of evidence. Tools such as GRADEpro GDT help build the table in the standard format.
What the four levels mean
High: we are very confident that the true effect lies close to the estimate. Moderate: the true effect is likely to be close to the estimate, but may be substantially different. Low: the true effect may be substantially different from the estimate. Very low: the true effect is likely to be substantially different from the estimate.
How Prolytiq does this step
Prolytiq does not produce GRADE ratings today. It gives you the inputs the assessment depends on: risk-of-bias judgments per study and domain, and the pooled estimates, confidence intervals and heterogeneity statistics from the statistics step, which you then carry into your GRADE assessment.
See the workflow →Frequently asked questions
Is GRADE applied to each study?
No. GRADE rates the certainty of a body of evidence for one outcome. Individual studies are assessed for risk of bias, and that assessment feeds into the GRADE rating.
Do I need GRADE in a systematic review?
Cochrane reviews of interventions are expected to use GRADE and include a summary of findings table, and many journals expect certainty of evidence to be reported. PRISMA 2020 asks reviewers to describe any methods used to assess certainty in the body of evidence.
Methodological sources
- GRADE Handbook (Schünemann H, Brożek J, Guyatt G, Oxman A, editors)
- Cochrane Handbook for Systematic Reviews of Interventions, chapter 14: Completing summary of findings tables and grading the certainty of the evidence
- Guyatt GH et al. GRADE guidelines series, Journal of Clinical Epidemiology
Run your next systematic review in Prolytiq
From research question to publishable chart, with every decision logged and exportable for peer review.