CONSORT 2025 Replaced CONSORT 2010, and Seven Items Are New

CONSORT is a reporting guideline, not a design standard. It does not tell you how to run a trial; it tells you what a reader needs on the page to judge one. The 2025 revision is the first since 2010, and it is a replacement rather than an amendment.

The headline number is 30 items. Seven are new: data sharing (item 4), authors’ conflicts of interest (5b), patient and public involvement (8), eligibility of sites and of the people delivering the intervention (12b), how harms were assessed (15), who was analysed and how missing data were handled (21b and 21c), and intervention delivery and concomitant care (24a and 24b).

Three were revised: item 3 now asks where the statistical analysis plan can be accessed; item 10 now covers important changes after the trial started, including outcomes or analyses that were not prespecified; item 26 now asks for the number analysed and the number with available data at each time point. One was deleted: generalisability, folded into the limitations item.

The open science material — registration, protocol and analysis plan access, data sharing, funding and conflicts — is now grouped together rather than scattered, which is the structural change most authors will notice first.

The new items are, almost without exception, the things that let a reader tell a trial’s result from a trial’s presentation.

Prespecification is the clearest case. Item 10 now asks explicitly about outcomes or analyses that were not prespecified. A trial that reports its secondary endpoint as though it had always been the primary one is not detectable from the paper unless the paper is made to say so, and that is what the revised item does.

Harms are the second. Item 15 asks how harms were assessed, not merely what was found — because a trial that looked for adverse events weekly and one that recorded whatever participants volunteered produce numbers that cannot be compared, and until now the paper did not have to say which it was.

Patient and public involvement and data sharing are the third kind: things that funders and journals had begun to require separately, now written into the reporting standard so they are reported consistently rather than wherever each journal happened to ask.

How a journal applies it. The five journals that published CONSORT 2025 simultaneously — The BMJ, JAMA, The Lancet, Nature Medicine and PLOS Medicine — are also among the journals most likely to hold a submission to it. Many journals ask for the completed checklist as a submission file and will return a trial report without one before it reaches review.

The practical risk is not refusal; it is delay. A trial submitted without the new harms-ascertainment detail or without the analysis plan location comes back as a request, and the request arrives after the manuscript has been sitting in a queue. Building the seven new items in at writing is cheaper than retrofitting them at revision.