How we review research.
From the original paper to the numbers you see. Here is what we check, and who reviews it.
Explore the studiesEvery finding needs a source.
The wiki brings the research question, treatment, comparison, outcomes and limitations into one reading experience. Our summaries are built from accessible full texts; a title or abstract alone does not unlock an analysis.
Read the full paper
Identify the publication and inspect its methods, results and tables. Record the source and where each extracted result appears. An unreadable table stays unresolved.
Keep the comparison intact
Retain the group, denominator, unit and follow-up time. Distinguish a change within one group from a difference between groups. Record missing information explicitly.
Check the draft against the source
AI tools assist drafting, translation and extraction. A separate automated review checks the draft and its extracted data against the source. These checks come before the clinical review.
Clinical review by the editor
Dr. Marcus Yu Bin Pai reviews each study page clinically, and the page shows the date. If the summary, the critique or the source changes, the note is removed until he reviews it again. A reviewed page describes a study; it is not a treatment recommendation for you.
Study design is a starting point
A randomized trial, observational study and systematic review answer questions in different ways. A design label alone is not a certainty rating. We do not present every article as having a formal GRADE assessment.
AI use is disclosed
Automated work can miss context or introduce errors. ICMJE guidance emphasizes transparency and accountability when AI is used in publishing. Our automated source check remains explicitly labelled.
Corrections should be traceable
For an issue in a wiki summary, keep the article link, the affected passage and the original source location together. This makes a correction checkable. Use the Contact & corrections page to prepare an email for the editor.
Reporting is not the same as quality
EQUATOR lists reporting guidelines for different study types. They help identify what a report should describe. Listing a guideline does not mean a study was well conducted or that this wiki has completed a formal appraisal.
The standards behind these distinctions
- Cochrane · Certainty of evidence
- ICMJE · AI in publishing
- ICMJE · Corrections and version control
- EQUATOR · Reporting guidelines
Sources consulted on 22 October 2026.