ICMJE Authorship: Who Earns a Place and How to Resolve Disputes

A registrar spends eight months collecting data on 240 arthroplasty revisions. They rotate. Eighteen months later the paper appears with their name fourth, behind a consultant who was not involved and ahead of a statistician who is not listed at all.

Some version of this happens constantly, and it produces more lasting resentment than almost anything else in academic surgery. Authorship is the currency the specialty runs on — it determines training numbers, fellowships, appointments and promotion — and disputes about it are correspondingly bitter.

Most are entirely preventable, and the prevention takes one conversation before anyone collects anything. This post sits within our research integrity series.

The four criteria

The International Committee of Medical Journal Editors recommends that authorship be based on four criteria:

1. Substantial contributions to the conception or design of the work, or the acquisition, analysis or interpretation of data for it; AND
2. Drafting the work or revising it critically for important intellectual content; AND
3. Final approval of the version to be published; AND
4. Agreement to be accountable for all aspects of the work, ensuring that questions about the accuracy or integrity of any part are appropriately investigated and resolved.

Everyone designated as an author should meet all four, and everyone who meets all four should be listed. Those who do not meet all four should be acknowledged, not listed as authors.

The word that does the work

AND, not OR. That single conjunction is where nearly every misapplication starts.

It means data collection alone — however laborious, however many months — does not by itself confer authorship. The registrar above must also have contributed intellectually to drafting or revision, approved the final version, and accepted accountability.

That sounds harsh, and read carelessly it becomes the excuse used to demote trainees. Read properly it points the other way: if you collected the data, you are entitled to be given the opportunity to meet the other three criteria. Excluding someone from drafting and then citing their failure to draft as grounds for demotion is a misuse of the criteria, not an application of them.

Criterion 4 is also stronger than people assume. Being accountable for all aspects means you should be able to say which co-author was responsible for which part of the work. If you cannot, you are not in a position to vouch for it — and you have signed up to do exactly that.

The named malpractices

Gift authorship is so normalised in some departments that refusing it feels like the deviant act. It is worth being clear: adding a name that did not meet the criteria is a misrepresentation, and criterion 4 makes it a substantive one — that person is publicly accepting accountability for work they have not read.

Medical writers are legitimate — professionally written papers are common in industry-sponsored research and there is nothing improper about it, provided the writer's involvement and its funding are disclosed. Undisclosed, it is ghost authorship.

Where orthopaedics gets into trouble

The rotating trainee. Data collected by someone who has moved on, with the paper written a year later. They are contactable, and contacting them is not optional.

The large multi-centre series. Every site wants an author. Consider a collaborative group authorship model instead of a list of twenty names who mostly cannot describe the methods.

The industry-funded implant study. Where a sponsor's writer drafted, that must be disclosed — see Conflicts of Interest and Industry Relationships.

The statistician. Frequently ghosted. If they designed the analysis and interpreted the results, they have met criterion 1 and should be given the chance to meet the rest.

The registry study. Data custodians and registry staff often meet the criteria and are often omitted.

Author order, which ICMJE does not govern

The criteria say who qualifies, not what order they appear in. Order is convention, and conventions differ — which is precisely why it needs stating explicitly rather than assuming.

In most clinical specialties:

• First author — did the bulk of the work; usually the trainee or junior researcher
• Last / senior author — supervised and guided; usually the principal investigator
• Corresponding author — handles submission and post-publication correspondence, and takes on the accountability that goes with it
• Second and middle — descending contribution, though middle positions carry little signalling value

Joint first authorship is legitimate where two people genuinely contributed equally, and should be marked. Be aware it is discounted by some appointment panels.

Preventing the dispute

Nearly every authorship row traces back to an assumption nobody voiced. The fix is trivially cheap and almost nobody does it.

Before data collection begins, agree in writing:

1. Who is expected to be an author, on what basis
2. The anticipated order
3. What each person is responsible for delivering
4. What happens if someone does not deliver
5. What happens if someone leaves or becomes uncontactable
6. Who is the corresponding author

An email to everyone involved, sent at the start, is enough. It costs ten minutes and prevents the great majority of disputes — and if someone objects to writing it down, that itself is informative.

Revisit it when the work changes. Projects grow, people join, contributions shift. An agreement made two years ago about a different study is not binding on the current one.

Contributorship statements

Many journals now require a statement of who did what, and the CRediT taxonomy provides standard roles — conceptualisation, methodology, formal analysis, investigation, data curation, writing (original draft), writing (review and editing), supervision, funding acquisition, and others.

This is a genuine improvement on the binary of author-or-not. It makes contribution visible and specific, credits people whose input does not fit the traditional model, and makes gift authorship conspicuous — it is hard to write a contribution statement for someone who did nothing.

Use it even where the journal does not require it. Drafting one is also an efficient way to discover that your author list does not match reality.

Resolving a dispute

Escalate in order, and stop as soon as it is resolved.

1. Talk directly, early, and in person. Most disputes are misunderstandings that harden through email.
2. Go back to the criteria. Depersonalise it — the question is not who deserves it but who meets criteria 1 to 4.
3. Involve the supervisor or PI, if they are not the problem.
4. Escalate to the department or research lead.
5. Your institution's research integrity office — nearly every university and large hospital has one.
6. The journal. Editors take authorship disputes seriously and will often hold publication pending resolution. COPE publishes flowcharts editors follow for exactly this.

Do not resolve it by quietly submitting elsewhere, and do not add a name you do not believe qualifies simply to end the argument — that makes you a party to a misrepresentation.

If you are the trainee

You are the person most exposed, so protect yourself:

• Get the agreement in writing before you start. If you take one thing from this post, that is it.
• Keep your own copy of the data, the protocol and your drafts.
• Contribute intellectually, visibly — comment on drafts in writing, so your contribution to criterion 2 is documented.
• Leave contact details when you rotate, and stay reachable.
• Ask directly about the intended order rather than hoping.
• If excluded unfairly, escalate — but weigh it realistically. These disputes are draining and the power gradient is real. Take advice from someone outside the department first.

And when you become the senior author, the obligation reverses. The registrar who collected 240 revisions gets sent the draft, gets asked for comments, and gets first authorship if the work was substantially theirs. Departments where that is reliably true have no trouble finding trainees to do research.

The summary

Four criteria, joined by AND. Meet all four or be acknowledged. Collecting data does not by itself confer authorship — but it does entitle you to the chance to meet the rest, and denying that chance is a misuse of the criteria rather than an application of them.

Agree it in writing before anyone collects anything, write a contributorship statement even when not required, and remember that criterion 4 means every name on the paper is publicly vouching for work they should be able to account for.