Platform standard
Evidence standards
A visible grading system so that a source being mentioned is never mistaken for a fact being established.
Every factual claim on this site is graded by the kind of source behind it, not by how many links sit at the foot of the page. Grade A is a regulator or primary source used within its stated scope. Grade B is a qualified interpretation, shown as an interpretation with its limits visible. Grade C is a lead worth researching and never supports a published claim. Grade D is never published as fact. When credible sources disagree, the page explains the disagreement or narrows the claim to what is undisputed. A record is treated as unverified once its review interval passes, so a page that has not been checked recently says so instead of quietly continuing to assert itself. The sections below set out each grade and the rules that apply to numbers.
Why grade evidence at all
Citation is not verification. A page can carry an impressive list of links at the foot and still make claims that none of those links support: this is the single most common failure in consumer health writing, and it is almost impossible for a reader to detect. Grading forces the question at the claim level: what kind of source supports this specific sentence, and is that kind of source capable of supporting it? A grade is about the relationship between a claim and its evidence, not about the prestige of the publisher.
Grade A: controlling or primary
Regulators, official standards bodies, government health agencies, licensed product records and direct research may support a defined factual claim, strictly within their stated scope. Scope is the operative constraint: a device clearance says what was cleared and for what indication, and it does not say that the device is appropriate for a particular patient. A Grade A source used beyond its scope is not Grade A evidence for that sentence.
Grade B, qualified interpretation
Where a question requires professional judgment instead of a documented fact, a qualified reviewer or the convergence of several credible technical sources may support an interpreted answer, provided the interpretation is shown as an interpretation and its limitations are visible. Most of what is genuinely useful about dental decision-making lives here, which is why the presentation rules are strictest at this grade.
Grade C: research lead only
A single commercial source, a manufacturer's marketing material, a forum account or an individual observation can identify something worth investigating. Material at that grade cannot support a published claim. Material at this grade is used to decide what to research and is then either corroborated at Grade A or B, or dropped.
Grade D: never published as fact
Grade D covers unverified assertions, scraped summaries of other sites and text produced by a model without a traceable source. Figures with no methodology are not evidence at any strength. They do not appear on a page with a hedge attached; they do not appear at all. Automation may assist with organization, structure and drafting, but it is never the source of a factual claim and never the final reviewer of a clinical one.
The conflict rule
When credible sources genuinely disagree, there are exactly three permitted responses. Picking whichever source is most convenient is not among them. Explain the disagreement and what turns on it; narrow the claim until it is something all credible sources support; or leave the record unpublished until the question resolves. Silently choosing a side is the failure this rule exists to prevent.
- Explain the disagreement, and what would settle it for an individual reader
- Narrow the claim to the part that is not in dispute
- Withhold publication and record why
Numbers get a stricter rule than words
A number carries more apparent authority than the sentence around it, and it is the thing a reader remembers, quotes and plans around. So the Guide publishes no national average price and no estimate of what your dentist will charge. A dollar figure appears only when it is a fee copied from a named public fee schedule, with the schedule and its year shown beside it. No clinical success, survival or failure rate appears at all. The one place a percentage appears is where a plan term is itself a percentage, because coinsurance cannot be explained without one.
Where a number would otherwise help, the page gives the structure instead: the components of a cost, the factors that move it and the questions that make a quote comparable. An automatic check on every page blocks publication of one that breaks this rule.
Staleness is treated as an accuracy problem
An answer that was correct when written and has not been checked since is not a correct answer, it is an unverified one. Records carry review intervals set by how quickly the underlying evidence moves, and a record that has passed its interval without review is flagged for recheck, not quietly continuing to assert itself. Any figure or fact not verified in the current review cycle is treated as a hypothesis until it is re-fetched.