Guidelines that update annually — and the citation that quietly follows them

You cited a national guideline in 2022. The guideline was revised in 2023 and again in 2024. Your entry still says 2022 — and the URL in it now serves the 2024 text.

Guidelines, standards, statistical releases and policy frameworks behave unlike anything else in a bibliography. They are not superseded the way a book edition is, or replaced the way a preprint is. They are overwritten, frequently at the same address.

What makes grey literature different

A journal article, once published, is fixed. Its DOI resolves to that article for ever, and a later paper on the same subject is a different object with a different identifier.

An annually revised guideline is one continuous publication that changes contents. The issuing body does not usually mint a new identifier, and it very often does not preserve the old version at the old address.

  WHAT YOU CITED          2022 edition
                          the recommendation you
                          quoted, in section 4

  WHAT IS THERE NOW       2024 edition
                          section 4 renumbered,
                          the recommendation revised

  WHAT YOUR ENTRY SAYS    (Organisation, 2022)
                          + a URL that serves 2024

  no redirect notice · no version marker
  · nothing in your entry is flagged

So a reader following your citation reads the 2024 text while your sentence describes what the 2022 text said. Both of you are acting correctly and you disagree, which is the worst shape a reference can take.

A book edition announces itself. A revised guideline replaces its predecessor at the same address and says nothing.

Why the check has to be a resolution, not a lookup

There is no database record to compare against. Grey literature is largely outside the bibliographic databases: no DOI in most cases, no structured metadata, no publisher record stating that the 2024 version supersedes the 2022 one.

What there is, is a URL — and following it is the only way to learn anything. When the link is resolved, the page that comes back either names the version you cited or it does not, and a page announcing a later year than your entry is the signal.

This is the same mechanism as checking that a DOI lands on the paper it claims, applied to a document class where the link is all you have. It is also why the access date convention exists for these sources at all: with nothing else to pin the version, the date you read it is part of the citation.

The organisation-as-author complication

Guidelines are almost always authored by a body rather than a person, and that creates a second problem stacked on the first.

Your text cites (NICE, 2022); your bibliography lists National Institute for Health and Care Excellence. Those are the same author and no naive comparison of surnames will say so. It is also entirely normal for an organisation to be cited by acronym in the text and spelled out in the entry — the style guides ask for exactly that.

We handle that pairing deterministically: initials of the words, in order, plus a matching year. It is applied without asking, because there is no judgement in it — and it is disclosed on the report, because a silent merge is invisible. The wider case is in when a university becomes an author.

How to cite something that keeps changing

Four habits, and together they make the problem disappear.

1. Cite the version, not the document. Include the year, the edition or version number if the body issues one, and the section you used. "Section 4.2" ages better than a page number in a document that gets re-typeset.
2. Prefer a versioned link. Many bodies publish a permanent address for each release alongside the "current version" address. The permanent one is what belongs in a thesis; the current-version one is what belongs in a browser bookmark.
3. Record an access date. For anything without a stable identifier this is not decoration — it is the only thing in the entry that pins which text you read.
4. Archive it. A web archive snapshot of the version you used, cited alongside the live URL, survives the body reorganising its website — which they do, roughly as often as they revise the guidance.

When the update matters, and when it does not

Not every revision affects you. Guidance bodies revise for many reasons and most of them do not touch the paragraph you cited.

The question is what you used the guideline for. If you cited it for a stated figure, a recommendation, or a definition, a revision may have changed exactly that, and you need to look. If you cited it as evidence that a body publishes guidance in this area at all, the revision is irrelevant and updating the year would add nothing.

  YOU CITED IT FOR          A REVISION...

  a stated figure or a      probably changed it
  recommendation            -> read the new version

  a definition              sometimes changed it
                            -> worth two minutes

  the existence of the      did not touch it
  guidance                  -> leave the entry alone

  the 2022 position         is the whole point
  itself                    -> keep 2022, say so

Sorting your grey literature this way is quick because there is rarely much of it — a handful of documents in most theses, even in policy-facing fields. The sorting is what turns "check my guidelines" from a vague anxiety into twenty minutes of specific work.

And if your work is about the 2022 guidance — its reception, its effects, the change it introduced — then the 2022 version is your object of study and must stay, cited explicitly as that version. The same logic as reference currency by discipline: the newest is not automatically the right one.

What this does not claim

Citing a superseded guideline is not an error and the report does not present it as one, or as any suggestion of misconduct. It is a fact about a source that changed after you used it — which, for annually revised documents, is guaranteed to happen during a doctorate rather than merely likely.

Nor does the tool track guidance versions. It resolves your links and reports where they land, so that a URL now serving a different year is visible to you instead of being discovered by a reader. The judgement about whether the revision touches your argument is one only you can make, and it usually takes a minute per document.