GRADE & Certainty of Evidence

GRADE (Grading of Recommendations Assessment, Development and Evaluation) rates the certainty of evidence per outcome as high, moderate, low or very low. You start from the study design (RCTs high, observational low), then rate DOWN for risk of bias, inconsistency, indirectness, imprecision and publication bias, or UP for a large effect, a dose–response gradient, or plausible confounding that would only have diluted a real effect.

Design fixes only the starting rung. GRADE then moves an outcome up or down the certainty ladder: even RCTs can be downgraded to low certainty (e.g. for serious risk of bias plus imprecision), and a body of observational evidence can be upgraded by a very large, consistent effect. The output is a certainty rating attached to each individual outcome, not to the study or the guideline as a whole.

Read certainty as “how confident are we that the estimate is close to the truth” for that outcome. Then read the recommendation separately: GRADE labels recommendations strong or weak (conditional), and that label also weighs patients’ values and preferences, the balance of benefits and harms, and cost — not certainty alone.

It forces transparency: instead of a single “Level A” badge, a guideline must say which outcome, how certain, and why it rated up or down. It also explains why a guideline can make a strong recommendation on low-certainty evidence (e.g. an obviously life-saving intervention) and only a weak one despite high-certainty evidence.

• Certainty: high / moderate / low / very low (per outcome)
• Rate DOWN: bias, inconsistency, indirectness, imprecision, publication bias
• Rate UP: large effect, dose–response, opposing confounding
• Certainty of evidence ≠ strength of recommendation

A GRADE-rated guideline — modern bodies such as NICE, the WHO and the American College of Physicians build recommendations with GRADE. A typical evidence profile might rate mortality as moderate certainty but quality of life as low certainty for the same intervention, then issue a weak/conditional recommendation “for” it. The guideline shows you the rating and the reasons it was downgraded. Notice two things: certainty is reported outcome-by-outcome, and the strength of the recommendation is stated separately from that certainty.

• Confusing certainty of evidence with strength of recommendation.
• Forgetting that RCTs can be downgraded (high certainty is not automatic).
• Treating one certainty rating as covering all outcomes — it is outcome-specific.