Understanding Evidence
How do we rate evidence?
TL;DR — What the Evidence Shows
We rate evidence using GRADE framework, the same system used by the WHO and Cochrane. Each rating tells you how confident we are in a finding. We also tag every claim with its source type so you can see whether it comes from human or animal research.
This site rates evidence using the GRADE framework (Grading of Recommendations, Assessment, Development, and Evaluations). GRADE is the same system used by Cochrane, the World Health Organization, and most major clinical guideline bodies.
Each rating attaches to a specific finding, not to CBD overall. "CBD works" is not a rating we can assign. "Oral CBD may not result in meaningful pain reduction compared to placebo" is.
Ratings tell you how confident researchers are in the finding. They do not tell you how large the effect is.
What do the four certainty levels mean?
High certainty
Standard language: "[Treatment] results in [effect]."
We are very confident the true effect lies close to the estimate. Further research is very unlikely to change confidence. This typically requires multiple large, well-designed randomized controlled trials with consistent findings.
No CBD claim for arthritis currently meets this level.
Moderate certainty
Standard language: "[Treatment] likely results in [effect]."
We are fairly confident the true effect is near the estimate, but it could differ. Further research may change the picture. On this site, some safety and drug interaction findings reach this level.
Low certainty
Standard language: "[Treatment] may result in [effect]."
Our confidence is limited. The true effect may be quite different from the estimate. Further research is likely to have a big impact. The pooled evidence on CBD for pain across all types falls here.
Very low certainty
Standard language: "The evidence is very uncertain about whether [treatment] results in [effect]."
We have very little confidence in the estimate. The true effect is likely to be very different. Most CBD-for-arthritis findings on this site are rated very low due to small study sizes and mostly negative results.
What do the source type tags mean?
Every claim on this site also carries a tag that tells you what kind of research it comes from. This matters because not all research is equally close to answering the question "will this help a person?"
Human Trial
A randomized controlled trial or clinical study conducted in people. Participants are randomly assigned to receive the treatment or a placebo. This is the strongest type of evidence for whether a treatment works in humans.
Human Survey
Self-reported data from people who use CBD, collected through questionnaires. Surveys can tell us what people experience, but they cannot tell us whether CBD caused the benefit. Without a placebo comparison, survey results may reflect expectation effects.
Animal Study
Research conducted in animal models, usually rats or mice. These studies help scientists understand biology, but only about 1 out of 20 treatments that work in animals end up working in humans. (Ineichen 2024) Animal findings must be confirmed in human trials before they can guide treatment decisions.
Lab Study
Test-tube or cell culture research (also called "in vitro"). Scientists study cells in a dish to understand how a compound affects them. These findings are the furthest from real-world treatment effects. A lab study showing CBD affects a cell does not mean CBD will have the same effect inside a living person.
Why does source type matter so much for CBD?
The CBD evidence base has an unusual structure. Most of the positive findings about CBD and inflammation come from animal and lab studies. When researchers tested CBD for arthritis pain in human trials, the results were mostly negative.
This gap between what animal studies suggest and what human trials find is the central challenge when evaluating CBD for arthritis. It means that many claims you see in media or on product labels are based on animal or lab research that has not held up in human testing.
The preclinical-clinical gap in numbers
- Animal/lab studies showing CBD reduces inflammation: multiple positive results
- Human trials testing CBD for arthritis pain: 3 completed, all negative
- Overall translation rate from animal success to human approval: about 1 out of 20
Source: (Ineichen 2024) (translation success rate estimate).
This is why we tag every claim with its source type. When you see Animal Study on a finding, it means that result has not been confirmed in people. When you see Human Trial , the finding comes from a controlled study in humans. Learn more about animal vs. human studies.
How are ratings assigned?
GRADE ratings are assigned per outcome and per comparison. Researchers start with a baseline level based on the type of studies available. Then they adjust up or down based on the quality of that evidence.
Factors that lower certainty:
- Risk of bias — problems in how studies were designed (for example, participants knowing which treatment they received)
- Inconsistency — different studies finding very different results
- Indirectness — the evidence answers a different question than the one being asked (for example, animal data used to support human treatment claims)
- Imprecision — too few participants or results that are too spread out to draw firm conclusions
- Publication bias — evidence that studies with negative results were less likely to be published
For CBD and arthritis, two factors drive most of the downgrades:
- Indirectness: Many CBD claims rest on animal or lab studies that have not been confirmed in human trials. Presenting animal data with a clinical certainty badge would overstate the evidence. Claims supported only by animal or lab research receive a "not rateable" designation instead of a GRADE level.
- Imprecision: The total number of people studied in arthritis-specific CBD trials is about 270 across three studies. This is small enough that the results could shift substantially with new data.
What does a rating NOT tell you?
- It is not a recommendation for or against CBD
- It does not measure how much benefit occurs — only how certain we are about it
- It does not rate CBD overall — only a specific finding about a specific comparison
- It is not final — ratings change if new major studies are published
- "Very uncertain" does not mean "proven not to work" — it means we lack enough evidence to know
Key sources cited on this page
Page last reviewed: March 2026 · Authored by Claude (Anthropic AI) · Research methodology