PubMed
Bibliographic discovery, corrections, retractions, and linked publications.
Evidence process
Methodology
Documented searches identify candidate evidence. Qualified reviewers decide what it means and whether it is ready to publish.
Scope
A structured educational review
Each entry begins with a versioned review protocol: canonical name, synonyms and identifiers, research questions, source set, inclusion and exclusion rules, outcomes, study types, and jurisdictional scope.
Unless an entry follows a registered systematic-review protocol, it is described as a structured literature review and evidence synthesis—not a comprehensive systematic review. The service is for informational and educational use only; it does not evaluate patients, provide medical advice, or function as clinical decision support.
Source hierarchy
Authoritative discovery
ClinicalTrials.gov
Registration and results context; listing does not mean government review or endorsement.
Drugs@FDA
United States approval records, regulatory reviews, and official product information.
DailyMed
Current and historical labeling context, verified against approval sources where status matters.
FDA safety communications
Regulator-reviewed safety notices and identified potential signals.
Primary literature
Full reports used for material extraction whenever access and rights permit.
From record to publication
Six controlled stages
- 01
Define
Version the question, concept identifiers, synonyms, jurisdictions, study types, outcomes, and exclusions.
- 02
Search
Search declared authoritative sources and retain the query, access date, identifiers, and records reviewed.
- 03
Screen
Deduplicate records, link registrations and publications, and assess relevance against the documented scope.
- 04
Extract
Record study design, population, intervention, comparator, outcomes, harms, limitations, and exact source locations.
- 05
Appraise
Evaluate certainty by outcome and keep applicability separate from internal validity.
- 06
Approve
Require independent verification, clinical-editor review, and publication approval for material changes.
Outcome-specific appraisal
Certainty is not a compound score
Human-outcome findings are assigned high, moderate, low, or insufficient certainty using documented domains. Preclinical findings are kept separate, and the number of studies alone never determines certainty.
Risk of bias
How study conduct and reporting may distort the finding.
Directness
How directly the population, intervention, comparator, and outcome answer the question.
Consistency
Whether findings agree across sufficiently comparable studies.
Precision
How much uncertainty surrounds the estimated effect.
Reporting bias
Whether selective publication or outcome reporting may shape the record.
Applicability
How context may limit transfer to another setting; shown separately from certainty.
Study-reported regimens
Reported, not recommended
Regimen fields preserve the source population, formulation, route, amount, unit, frequency, duration, comparator, indication studied, jurisdiction, approval context, and exact source location. Protocol-planned, study-reported, and regulator-labeled information remain distinct.
We do not silently convert routes, salts, esters, units, or formulations. Ambiguity blocks publication until reviewed. Descriptions avoid terms such as “optimal,” “typical,” “safe,” “equivalent,” or “recommended” unless accurately characterizing a cited source—and never as our advice.
Related entries
Connections require a reason
A relationship may reflect a shared class, studied target, registered comparison, reported combination, or regulatory category. Every connection is human-approved and labeled; it is never presented as a substitute or recommended alternative.
Review dates
Maintenance with visible dates
The compendium is intended to be continuously maintained through documented editorial review, but source checks are not represented as automatic or real-time. Review priority is based on potential significance, and published pages distinguish the source-search date, last human review, and entry revision date.
Editorial readiness
Methods need named owners.
Before any medical synthesis is represented as published, this methodology requires clinical-editor approval, named reviewer qualifications, versioned protocols, quality-control checks, escalation owners, and legal review.