Public resource · definitions
Evidence glossary: plain-language definitions for source types, endpoints, uncertainty, and study status
Definitions reduce ambiguity. They do not replace the surrounding study, document, or regulatory context.
Study and source terms
A primary study reports original research. A systematic review describes a planned method for locating and synthesizing multiple studies. A registry record describes a study protocol, status, and results where posted. A preprint is generally an author-posted manuscript before formal journal peer review.
These labels identify document types, not a ranking of personal usefulness. The reader still needs to inspect the question, population, methods, and limitations.
Outcome and uncertainty terms
A primary outcome is the main outcome specified for a study. A surrogate outcome is a measured marker intended to stand in for another outcome. A confidence interval expresses uncertainty around an estimate. An adverse event is an event recorded during a study and does not by itself establish causation.
Definitions should not be pulled out of context. The same term may be used differently across protocol, statistical, and regulatory documents, which is why the original source remains important.
Status and correction terms
Recruiting, completed, withdrawn, and terminated are registry-status terms. Correction, expression of concern, and retraction are publication-record terms. Verified, unavailable, blocked, and manual review required are checkout-log categories on this site.
Keeping these categories separate makes the editorial and verification record easier to audit.
Definitions need an evidence context to be useful
A definition can reduce confusion, but it can also create a false sense that the reader has evaluated the underlying document. Terms such as “randomized,” “significant,” “adverse event,” “completed,” or “verified” only become meaningful when tied to a particular study design, endpoint, registry record, or cart observation. The glossary therefore gives plain-language distinctions while directing readers back to the source context.
The same word may also be used differently across scientific, regulatory, and commercial records. “Status,” for example, can refer to a trial’s recruitment stage, a regulatory record’s category, or a checkout archive’s evidence classification. Keeping those uses distinct prevents a familiar label from doing more work than its surrounding document allows.
Operational definitions make the glossary testable
A useful glossary definition tells the reader how a term would be recognized in a real record. “Primary outcome,” for example, should direct the reader to the specified main measure in a protocol or report; “cart subtotal” should direct the reader to the amount shown before the code effect; “accepted” should direct the reader to the event fields required for that archive category. Definitions become more valuable when readers can apply them.
This is also why the glossary avoids using familiar words as universal conclusions. “Verified,” “significant,” and “available” each depend on defined evidence contexts. A reader who knows the operational boundary can distinguish an observed fact from an assumption that happens to use the same label.