Editorial standards

Who publishes this, and what that limits

One person writes and edits this publication: Zane Hitchcox, an independent publisher, as a sole proprietorship. There is no editorial team, no advisory board, and no clinical reviewer. Nobody here is a clinician; nobody here has worked in an operating room, used a fluid-management system, or handled regulated medical waste.

That limit sets the scope. What we do is read documents — the Code of Federal Regulations, OSHA letters of interpretation, state administrative codes, municipal sewer-use ordinances, FDA clearance records, and the peer-reviewed literature — and every clinical, regulatory or operational claim on this site is a claim about what those documents say, attributed to a specific document, with the date we retrieved it. The full statement of that scope is the first thing on About.

Not clinically reviewed is printed on every page rather than left off. If a real credentialed external reviewer is ever retained, that reviewer gets a name, a credential, an affiliation and a review date on the page they reviewed. Until then the field says what is true. There is no third option, and there will never be an unnamed “editorial team” or a “medically reviewed” badge on this site.

The seven commitments

1. Quote first. Every regulatory, clinical or safety claim appears adjacent to the quotation it rests on. We do not paraphrase a regulation in house voice without the operative text on the same screen.
2. Cite the primary document, never a summary of it. No citation to an article about a rule, a vendor’s paraphrase, or an AI overview. Ever.
3. Pull raw text. Every quotation is transcribed from the retrieved document — eCFR text, GPO XML, the agency PDF — never from a search result or a summariser. An automated summariser was observed during our own research inventing a sentence in an OSHA letter that the letter does not contain; reproducing that distortion layer is the one failure that would end this publication, because correcting it is why it exists.
4. Absence is a finding. When a document does not contain the thing everyone says it contains, we publish the keyword scan with its date and its method.
5. We report provenance, not verdicts on truth. “This figure is attributed to CDC; it does not appear in the cited CDC document, and we could not find it in any CDC publication” is our sentence. “So the real number is higher” is not. Where a claim is also contradicted by better evidence, that is a separate sentence with its own citation.
6. We name documents, not villains. The most accurate sources the ground phase found included a solidifier manufacturer and a waste hauler; among the worst were a clinician-authored trade article and a broker’s blog. The vendor/independent axis does not predict accuracy. Care does.
7. We do not name a winner, and we are not giving advice. No product recommendation, no ranking of named systems, no advertising, no affiliate link, no sponsored content, no vendor review of copy, no investor information. Nothing here is legal advice, medical advice, or a substitute for your facility’s exposure control plan, your state’s rules, or your sewer authority’s ordinance.

Sourcing rules

These are mechanical. A page that breaks one of them does not ship.

• Every number, date, dollar figure, percentage and regulatory assertion carries a citation. No exceptions. A statement that cannot be cited is rewritten as an attributed observation, marked as our judgment, or cut.
• Citations are superscript numerals resolving to a numbered Sources list at the foot of the page. Inline links alone are unauditable: they hide the publication and the year, they rot without telling anyone, and they let a writer imply a source supports more than it does.
• Every entry in Sources carries an access date, and a hyperlink wherever our retrieval log recorded the address, because regulatory pages move. Where the log kept the publisher and the date but not the address, the entry names the publisher and says so: we do not print a URL we did not ourselves follow, and we do not reconstruct a plausible one after the fact. Those entries are listed as outstanding work on the page that carries them. Format: Author(s). Title. Publication. Year;Vol(Issue):pages. URL (accessed YYYY-MM-DD). For a regulation: Issuing body. Title. Citation. Effective [date]. URL (accessed YYYY-MM-DD).
• Where we could not follow a primary URL ourselves, an archive link runs beside it and a chip says which of the two things happened, because they are not the same thing and we report them separately. A link rot — archived copy chip means the document is not at that address any more: the URL is dead, or redirects somewhere the cited content is not, or the publisher retired the system that served it. A server refused — archived copy chip means the document is almost certainly still there and the server would not give it to us — an HTTP 403, or a bot wall that answers a browser and not a client like ours. The distinction matters and we hold to it everywhere on this site: a refusal is a fact about our retrieval, not a fact about the document. Several federal agencies, including CDC, currently refuse us and serve the same page to a person with a browser. Fragile sources carry an archive link from the start.
• Sources and Further reading are two separate lists, never merged. Further reading is material we did not cite in the text.
• Manufacturer-derived and manufacturer-funded material is labelled inline, every time it appears — “according to manufacturer documentation”, “in a manufacturer-sponsored study”. It may be cited. It may not be laundered into house voice.
• A dataset’s weakness is stated on the same screen as the number, in our own words, every time the number is used — not once, at the bottom, in a footnote nobody reaches. A survey with a self-selected cohort carries its cohort caveat and its collection year wherever it appears, and no two collection years are presented as one dataset.
• Absolute numbers travel with percentages. “5 of 100 procedures”, not “5%”. A percentage with no denominator does not ship.
• “Studies show” is not a sentence we write without a number and a citation attached to that exact sentence.
• No aggregate trust counter. There is no “N of N sources verified” figure anywhere on this site, and there never will be. No trust claim we make may be one a reader can falsify in a single click.

The certainty ladder

Hedging is where a reference publication usually goes soft. Ours is a fixed vocabulary of six registers, adapted from the certainty-of-evidence wording used by Cochrane and the GRADE working group. A writer picks a rung; a writer does not invent a hedge. The sixth rung is the one most publications leave out, and it is mandatory here: any conclusion that is not derivable from the cited evidence is our judgment and has to say so in the sentence.

Two consequences worth stating plainly. A finding about a citation is not a finding about the world: when we report that a figure cannot be traced to the document it is attributed to, that is a statement about the record, and it is smaller than “the figure is wrong”. And where a document is genuinely ambiguous — where a provision can be read two ways and no agency or court has settled it — we publish the ambiguity as an ambiguity, label the reading as a reading rather than a holding, and name the document that would settle it.

The quotation policy

The design of this site has one job, and it is this policy made visible: to make it obvious, at a glance and without reading, which words are the government’s and which are ours. Verbatim primary text is set in monospace, on a sunk ground, behind a rule. House prose is set in serif. Nothing else on the site is set in monospace at length.

• Every quotation is transcribed from the retrieved document itself — the eCFR text, the GPO XML, the agency PDF, the state code viewer — and never from a search result, a snippet, an aggregator or a summariser.
• A document extract carries its own citation strip: the exact designation of the provision, the source it was taken from, and the date the text was current or retrieved. Beneath it run two links: the full section at the issuing body, and an archived copy.
• No quotation marks appear inside an extract. The typeface and the frame do that work, and quotation marks inside a long provision are a source of transcription error.
• Ellipses may remove words. They may never remove a qualifier, a scope limitation or a condition. Where a provision has a scoping sentence, the scoping sentence is quoted with the provision it scopes, even when that makes the extract longer than the point being made.
• Emphasis is never added silently. If we bold or italicise inside a quotation, the line beneath says so.
• Copyrighted standards are cited, not reproduced. Several documents that govern this subject — hospital design guidelines, the health-care facilities code, plumbing code sections, and the major nursing and surgical technology association guidelines — are copyrighted works sold by their publishers. We describe and cite their positions with the edition and section number. We do not reproduce their operative text at length. Where we cannot print the sentence that does the work, we say so, and in at least two cases we have deferred an entire article rather than write around a document we cannot quote.
• Provenance of the retrieval is disclosed when it is unusual. Where a live official server refused us and the text came from a web archive capture, the row or the extract says so and gives the capture date, and we check it against a second official server wherever one exists.
• Quoted United States government text is in the public domain. Third-party quotations remain their authors’. Our own prose, tables and diagrams are CC BY 4.0; the terms are at How to cite and reuse.

What the register labels mean

The three registers at Registers are maintained tables, not articles. Every row carries its source, its citation, the verbatim primary text where one exists, the date we retrieved it, and exactly one of four status labels. The vocabulary is closed. There are four labels, no others, and no free-styling.

What “the primary text” is in each register

The labels are the same everywhere. The document they are about is not, and it would be dishonest to let one word cover both cases without saying so.

• Registers A and B are indexes of operative clauses. The primary text is the sentence in the state rule or the sewer-use ordinance that does the work, and a row may not carry Verified or Qualified unless that sentence is quoted in the row.
• Register C is an index of enacting instruments, and its Verified label certifies something narrower. There it means that the citation, the bill number, the chapter or public-act number and the effective and compliance dates in the row were read against that legislature’s or code publisher’s own primary text on the date shown. It does not certify that we have transcribed that state’s operative command. The register carries a column showing exactly what we did transcribe, and six of its twenty-two rows say Not transcribed. Four states’ operative commands are printed there in full — Kentucky, Louisiana, Georgia and California — and transcribing the rest is that register’s standing work. We would rather publish the narrower certification and name it than let a reader assume the wider one.

Three rules ride on that table, and they are the reason it is worth publishing.

• A row may not carry Verified or Qualified without a retrieved primary quote in its quote column. Not a summary of the rule, not a one-line verdict from a secondary source, not our own recollection of what it says. Everything without a quote ships as Not checked, however confident anyone is about the answer.
• Silent is a claim about our scan, not about the law. It means the governing text we retrieved does not contain the terms we searched for. It does not mean no duty exists. The terms, the character count and the date are printed with the row so that a reader can add a term we missed and tell us.
• Rows are never deleted. A superseded row is struck through, dated, and left in place with a note saying what replaced it and when; the replacement is added beneath. Every row has a stable anchor so it can be linked into an argument and still be there later.

The labels carry meaning in text first, then in fill and border style, and only last in colour — so that a printed greyscale copy, a colour-blind reader and a screen reader all get the same information. They are deliberately not a traffic light. Green, amber and red would say “true, suspect, false”, and none of these labels says anything about whether a rule is good, bad or correctly decided.

What the claims-ledger labels mean

The claims ledger is one page of anchored rows, one row per figure or assertion in wide circulation, each followed hop by hop from where it circulates back toward the document it is attributed to. Each hop carries its document, the HTTP status the server returned, and the date we retrieved it. This vocabulary is also closed: six labels, no others.

A broken chain is a finding about the record, not about the world. None of these labels says anything about whether a claim is true. A figure we could not trace may still be correct; we are reporting that we could not find the evidence, which is a different and smaller statement.

Two fields are mandatory in every ledger row and a row without them does not publish: What the record does support, which hands the reader something they can cite instead, with its source; and This would change if, which names the specific document that would move the label, with an invitation to send it. Those two fields are the mechanism that keeps a provenance ledger from becoming a grievance register.

The review process

There is no second reader. Saying so is part of the standard, because a review process that describes people who do not exist is worse than none. What exists is a fixed set of checks run before a page ships and a fixed cadence for re-reading it afterwards.

Before a page ships

• Every quotation is compared character by character against the retrieved document, not against a note taken from it.
• Every citation is opened and confirmed to resolve to the document named, and the access date recorded is the date that check was made.
• Every number in body copy is traced to a numbered source entry, and any number that will not trace is cut.
• Every table is confirmed to have a numbered caption and a source line. A table without a source line does not ship.
• Every row in Registers A and B is confirmed to carry a quote of the operative clause if it carries Verified or Qualified. Every row in Register C is confirmed to carry its bill, chapter or public-act number read against the primary instrument, and to show in its verbatim column exactly what was transcribed — including when the answer is nothing. Every row in all three carries a retrieval date either way.
• Every hedge is confirmed to sit on one of the six rungs of the certainty ladder.

Afterwards

• Every article, every register and every maintained index carries Published, Last reviewed and Next review dates in a metadata block above the text. Next review is twelve months after publication, except where a register states a faster cadence, and it is a promise rather than decoration: it is either kept or removed. The short standing pages that describe the publication rather than a body of evidence do not carry the block — the homepage, About, Contact, Corrections, Change log, How to cite and Privacy. Every substantive change to any of those is dated at Change log instead.
• Last reviewed moves only when a human has actually re-read the page against its sources. A review that changes nothing still moves the date and is still logged, because reviewed-and-unchanged is information.
• Registers A and B are reviewed annually, on 1 March. Register C is reviewed twice a year, on 1 March and 1 September, because state legislation moves faster than administrative code. This cadence is deliberately slower than the quarterly and thirty-day cadences these pages could have promised. A published review date that is missed is a worse signal than no date at all, and one person tracking fifty legislatures on a thirty-day trigger will miss it.
• Version numbers. 1.0 is first publication. A minor bump (1.x) is new material, updated figures, added or re-verified rows or new sources, with the conclusions unchanged. A major bump (2.0) means a conclusion changed, and the change-log entry has to say which conclusion and why.
• Supersession keeps the URL. A replaced page stays where it is and gains a banner linking forward. Nothing is silently deleted. That failure mode is what created this subject in the first place: the numbers everyone still quotes come from a trade article that no longer resolves, so the figures circulate and the document cannot be checked.
• Every substantive revision is logged at Change log with its date, the page, what changed and why. Typo fixes are not logged; nothing else is exempt.

The correction policy

If we get something wrong, tell us and we will fix it in public. Corrections are dated, permanent, and describe the error rather than hiding it. We do not delete a wrong sentence and move on; the notice says what the page originally said and what it says now. Send corrections to corrections@biodrainmedical.com with the page title and the date on it. We aim to answer within five working days and to publish a documented correction within seven, whether or not the person who found it wants credit.

The mechanics are fixed. A correction produces a dated notice at the top of the corrected page, permanently; an entry at Corrections, newest first, permanently; a version bump; and an entry at Change log. Never a silent edit. The notice has one form — the date, then what the page originally said, then what is right — with no apology theatre and no vagueness about which sentence was wrong.

Corrections launches empty and dated, and it says so. An empty dated index is credible. A fabricated one is fraud, and no correction will ever be invented here to make the page look worked.

What we will not publish

This list is not aspirational. Each line is a build rule, and several of them cost us articles we would otherwise have written.

• No product recommendation, ranking or named-brand comparison of fluid-management systems, canisters, solidifiers or liners — not in text, tables, captions, alt text or source lines. A manufacturer is named only where the identity of the party is itself the fact being reported, and where that happens, the agency’s own standing position that it does not approve or endorse products is quoted on the same screen.
• No advertising, sponsorship, affiliate link, paid placement, guest post, sponsored content or link sale, and no call to action inside editorial copy. The full position is at Independence and funding.
• No medical advice and no legal advice. Nothing here is a substitute for your facility’s exposure control plan, your state’s rules or your sewer authority’s ordinance.
• No procedural how-to and no troubleshooting instruction. We do not tell anyone how to perform a task in a clinical setting.
• No first-person experience claim of any kind. No “we tried it”, no clinical anecdote, no assertion about what facilities generally do. The first person is reserved for our own procedure and our own judgment, and it is labelled when it carries judgment.
• No invented reviewer, credential, affiliation or advisory board, and no unnamed “editorial team”.
• No claim we could not trace — not as a finding, not as an aside, and not as a foil. Several widely repeated figures in this field trace to nothing we could obtain. They are absent from this site entirely rather than reprinted with a rebuttal, because reprinting a claim to knock it down still puts it into circulation with our name on it.
• No verdict on a jurisdiction we have not read. A state we could not verify appears as a row that says we have not checked it, with no citation and no bill number, rather than as a confident sentence sourced to a tracker.
• No bill number, chapter number or date that has not been read against that legislature’s own primary source.
• No photograph, no stock image and no AI-generated image, ever. The only images on this site are diagrams we drew, each numbered, captioned, and carrying a source line naming the documents it is derived from.
• No investor information, no statement about any company’s securities, and no reference entry about any manufacturer’s regulatory clearance record.
• Nothing copied or paraphrased from any previous owner of this domain, from any era, including as a foil. The history of the domain is set out with its sources in The Domain Record; none of that material is republished here.
• No trend where there is no trend. A surveillance dataset with a convenience sample and a shifting denominator is not presented as a year-over-year comparison, and a figure is not separated from the population it was measured in.