HealthWatchlist separates three questions: what did the research find, how much confidence does it deserve, and what safety concerns apply? Each assessment concerns a particular preparation, route and outcome.
What the finding means
- Promising benefit: a signal worth investigating, not proof of effectiveness or a treatment recommendation.
- Mixed results: results differ between studies, preparations or outcomes.
- No demonstrated benefit: the relevant human studies did not establish the claimed benefit. How much that tells us depends on their quality and size.
- Not adequately tested: direct human testing is absent or insufficient. This is different from a reliable negative trial.
How strong is the evidence?
Strong, moderate and limited describe editorial confidence in direct human evidence, considering study design, consistency, size and relevance. Strong means consistent, convincing direct human findings. Moderate means credible findings with important limitations. Limited means the direct human evidence is small, inconsistent or otherwise uncertain.
Indirect evidence concerns a different preparation, route, population or outcome. Laboratory research only describes experiments that do not establish effects in people. These are not formal GRADE ratings.
Safety is a separate question
A promising finding does not establish safe use. The assessment considers irritation, allergy, poisoning, interactions and the risk of delaying appropriate care where relevant. An essential oil, extract, food, isolated molecule and finished medicine are not interchangeable.
How a claim is assessed
- Define the question. Identify the exact preparation, route, population and outcome.
- Find relevant evidence. Use authoritative guidance, systematic reviews and primary human studies, checking whether they address that question.
- Compare the results. Include negative and conflicting findings, not only favourable results.
- Explain the limitations. Consider sample size, blinding, comparators, follow-up and whether a measured change matters to patients.
- Publish the reasoning. Provide a finding, limitations, sources and safety context so readers can inspect the judgement.
What the Matrix covers
The Evidence Matrix lists selected structured assessments. It is not a complete list of every comparison discussed in the condition guides. An unlisted pairing is not a negative verdict. “Classification pending” means a record still needs editorial classification.
Why some pages still show a numerical score
Earlier assessments used John’s Evidence Score out of ten. Where retained, this is a secondary editorial summary, not a percentage chance of benefit, a safety score or a validated clinical measurement. It does not determine the Matrix labels or colours.
Search and review limitations
These are editorial evidence assessments, not registered systematic reviews. Searches may miss studies, and an abstract may not provide enough detail to assess a full trial. Individual reviews explain their search scope and access limitations. Laboratory and animal findings cannot by themselves establish a benefit in people.
Understanding dates and changes
A publication date records when a page first appeared. An updated date may reflect an editorial or link change, not a comprehensive new evidence review. A stated evidence-review date should identify a substantive check of the evidence supporting a claim.
A larger trial, a better review or an important safety finding may change an assessment. A targeted correction should explain its scope rather than imply that every claim has been reassessed.
Editorial responsibility and corrections
John Hamlen is responsible for publication and editorial decisions. HealthWatchlist uses AI-assisted research and publishing tools, as explained in the Editorial Policy. Articles are not independently clinician-reviewed unless a named reviewer and the scope of review are stated.
Reasonable readers may weigh imperfect evidence differently. If a source has been misread or important evidence is missing, please use the corrections process. The aim is to make each judgement inspectable and correct material errors.