Editorial policy

We work from primary sources, in roughly this order of preference:

• Clinical practice guidelines from recognized professional bodies.
• Government health agencies such as the NIH, CDC, NHS and WHO.
• Systematic reviews and meta-analyses.
• Individual peer-reviewed studies, where they are the origin of a specific equation.

Every formula, threshold and category is traceable to one of these, cited on the page with the publisher, the publication date where available, and the date we last accessed it. We do not cite content farms, and we do not cite a secondary source for a claim when the primary one is available.

We do not invent consensus. Where guidelines disagree — as they do on blood pressure categories — we say so and name both, rather than picking one and presenting it as settled.

Health content can mislead through phrasing alone. We hold ourselves to specific rules, which are checked automatically before anything is published:

• We never tell a reader they have a condition, or that they are healthy or safe.
• We never recommend starting, stopping or changing a medication or dose.
• We never present an estimate as a measurement, or a screening range as a diagnosis.
• We use hedged language where hedging is accurate — “may”, “is commonly interpreted as”, “is an estimate” — and avoid absolutes such as “guaranteed” or “definitely”.
• We prefer “within the reference range” to “normal”, because “normal” reads as a verdict about the person rather than a description of a range.

Content moves through defined states before publication: draft, research review, medical review, editorial review, then approved. Publication requires the approved state. Automated tooling can prepare a draft and run checks; it cannot approve or publish health content, and that restriction is enforced in the database rather than by convention.

Before anything is published, automated checks verify that required sections are present, every cited source resolves over HTTPS, every internal link points at something real, and no flagged language pattern appears.

A clinician-review credit appears only where a real, identity-verified clinician has actually reviewed that page. When they have, they are named with their credentials and the date of review.

Where no reviewer is named, no clinician review has taken place. Our launch content carries an editorial byline and no review claim, because implying review that did not happen would be a fabricated credential — the most damaging thing a health publisher can do.

Each page records when it was created, published, last substantively updated, and last reviewed. The “updated” date moves only when the content itself changes — a technical change that touches a page does not make it look fresher than it is.

Content older than twelve months is flagged for review, as is any page whose cited sources have stopped resolving or been superseded.

If we get something wrong we correct it and update the date. Please tell us if you find an error.

• Content written because a keyword exists rather than because a reader needs it.
• A calculator without a documented equation, a cited source and stated limitations.
• Fabricated statistics, invented references, or a threshold we cannot attribute.
• Fear-based framing, or content designed to make a healthy reader anxious.
• Anything that gamifies a health outcome rather than learning.