How we assess evidence.
One claim. A traceable judgment.
Draft method v0.1 · Methodological review not yet completed
AHigh certaintyHigh confidence in this specific claim.
The main limitations identified in the appraisal are unlikely to change the judgment. High certainty does not mean the claim applies to everyone or that the effect is large.
BModerate certaintySome confidence, with limitations that still matter.
Further research or more direct evidence may change the judgment. Read the scope and grading rationale alongside the grade.
CLow certaintyLimited confidence; the conclusion may change substantially.
Study limitations, inconsistency, indirectness or imprecision affect the judgment. Preliminary findings should not be presented as assured effects.
DVery low certaintySubstantial uncertainty; a reliable judgment remains difficult.
Grade D does not mean an intervention has been shown to be ineffective. Evidence of no important effect and a lack of reliable evidence are different situations.
UnassessedThe assessment is not complete. This does not mean low certainty.
Not applicableOutside the scope of this grading process. A reason must be provided.
What we assess.
Record each judgment, and the reason behind it.
Risk of bias
Could study design, conduct or selective reporting affect credibility?
Consistency
Do studies reach similar results? Can differences be explained?
Directness
Do the population, intervention, comparison and outcome directly address the question?
Precision
Are the sample and uncertainty intervals sufficient to support this claim?
Publication bias
Could unpublished, negative or unfavourable results be missing?
These domains draw on common questions in evidence appraisal. Our A–D framework is an editorial communication tool, not a formal GRADE assessment. Four letters cannot replace a complete methods record. Where a domain does not apply, explain why.Methods reference: Cochrane Handbook, Chapter 14 ↗
Content type and grade, explained separately.
Scientific concept
Explains concepts, definitions and boundaries used in research. A concept does not receive a single overarching evidence grade.
Research explanation
Organises and appraises research around a specific question, explaining the evidence and certainty for each important claim.
Project model
An explanatory framework, metaphor or educational simulation developed by this project, with explicit assumptions and limits.
For example, the Allostatic Load page is a “Scientific concept”. A specific claim about measurement lists its sources separately and awaits assessment. The recovery curve is a “Project model”: a precise-looking graphic should not be mistaken for measured data.
How a grade is reached.
Keep a record of the process before displaying a formal grade.
See it in an article- Define the question
State the claim, population or scope, comparison and outcome. First determine whether grading is appropriate.
- Document the search
Record databases, search terms, dates and inclusion criteria. Explain how negative findings and unpublished evidence are considered. Selecting a few supportive papers is not enough.
- Assess each domain
The assessor records the evidence for each judgment, key limitations and the final grading rationale. Use methods appropriate to the question.
- Have another person verify it
Record the actual assessor, verifier, date and method version. Incomplete records remain “Unassessed”.
- Updates and corrections
Reassess when new evidence emerges or an error is found. Keep a record of important changes and their reasons.
What is the status of this method?
This is a draft method. Claims in the articles have not yet been formally graded. The framework, example and editorial process are in place; assessors, scientific verifiers and methodological review are still to be arranged.
Grades help readers understand uncertainty. They are not diagnostic tools and cannot be converted directly into product efficacy ratings or purchasing recommendations.