OE Journal

ORIGINAL ANALYSIS

Grading a Body of Evidence: How certain are we in the results?

The GRADE framework helps determine how confidently a meta-analysis reflects the true treatment effect. Evidence begins at high certainty for randomized trials but may be downgraded across five domains: risk of bias, inconsistency, indirectness, imprecision, and publication bias. Each outcome in a review is assessed separately, meaning quality can vary within the same study. Examples illustrate how high heterogeneity, methodological limitations, small sample sizes, and wide confidence intervals can shift evidence from high to low or very low certainty. By contrast, outcomes with stronger methods and precise estimates may retain moderate certainty. Understanding these ratings is essential when interpreting large effects that rest on uncertain evidence.