Separate the kinds of information
Explanatory text is maintained in the site's editorial content. Selected registry and literature records supplement it. A reference count is a count of linked records, not a score for quality or a count of independently reviewed claims. Profiles differ in depth; short registry entries should be read as starting points for investigation.
Identity comes first
PubChem chemistry and IUPHAR/BPS target records can help identify a compound, but names and aliases can refer to a parent molecule, salt, conjugate, fragment or different sequence. Check the record identifier and molecular form. Sequence-based mass calculations omit unencoded modifications, counterions and hydration; they should not replace a verified full-molecule record.
Trial registrations are research footprints
ClinicalTrials.gov describes registered studies, including planned, recruiting and completed trials. A name-based match can include a comparator or related compound. The imported trial footprint is a limited snapshot, not a complete inventory or proof of efficacy, safety or approval. Read the linked record for the intervention, status, population and posted results.
Primary studies and scope
Guide source entries state which sections they concern and their important limits. Laboratory mechanism, animal efficacy, human pharmacokinetics, randomized outcomes and regulator decisions answer different questions. Cross-trial percentages cannot establish a head-to-head ranking. A literature link requires a relevance check beyond matching the title.
Dates describe different events
A registry snapshot date records data collection. A page review date records an individual editorial review when one is known. A build or a new bibliography does not make either event happen again. No universal review date is substituted for pages without a recorded review.
What remains to improve
Thin entries need verified molecular identity, a clear native-versus-modified distinction, the strongest relevant primary studies with populations and endpoints, and explicit missing evidence. They should be expanded only after those checks. The site is not a comprehensive systematic review, and absence of a citation is not proof that a study does not exist.