CONSORT 2025's New Statistical Items Land in Statisticians' Laps

On April 28–29, 2025, CONSORT 2025 was published simultaneously in JAMA, The Lancet, The BMJ, Nature Medicine, and PLOS Medicine — the first revision since 2010. SPIRIT 2025 dropped at the same time, updating the protocol-reporting standard for the first time since 2013. ICMJE revised its foundational manuscript recommendations in January 2026. And in May 2026, FDA finalized ICH M11 CeSHarP — the machine-readable structured protocol standard — completing the most compressed simultaneous reset of trial-reporting infrastructure in recent memory.

For biometrics teams, the CONSORT changes are not someone else’s editing problem. Three of the seven new or revised checklist items sit squarely in the statistician’s lane: a revised item 3 requiring disclosure of where the SAP can be accessed; a new item 21b on analysis population definition and group assignment; and a new item 21c on how missing data were handled. That last one matters beyond checkbox compliance — it asks teams to actually document the method, not just assert that missing data were “handled appropriately,” a phrase that has historically done a lot of heavy lifting in CSR statistical sections.

The compliance gap these items are designed to close is real. A large-scale analysis of 21,041 RCT articles found that only 2.22% provided protocol access information and just 9.7% reported randomisation sequence generation — both areas now explicitly targeted by CONSORT 2025. SAP templates, CSR boilerplate, and TLF shells built against CONSORT 2010 will need auditing.

The regulatory convergence layer is CeSHarP: where SPIRIT 2025 sets the human-readable standard for what belongs in a protocol, CeSHarP operationalises it in machine-readable form for regulatory exchange. Teams updating protocol templates for one should be updating for both simultaneously — they govern the same content in parallel registers.

Two open questions deserve active tracking. First, ICMJE updated its recommendations in January 2026, but the specific changes are not surfaced on the landing page; the annotated changelog sits in the News & Editorials section and requires a PDF download to verify scope. Whether the revision touches AI-assisted writing disclosure or data-sharing mandates — both live issues for manuscript teams — is currently unconfirmed. Second, ISMPP is running a member survey on AI governance in publication workflows, signalling that biopharma organizations are actively formalizing policies in this space. The unresolved question is whether biometrics-generated statistical outputs — outputs that AI tools increasingly mediate in synthesis and drafting — will be treated as “AI use” requiring disclosure. Most organizations have not yet decided.

Both CONSORT 2025 and SPIRIT 2025 are designated living guidelines, hosted with companion tools SEPTRE and COBWEB at consort-spirit.org. Living status means the checklist can update without waiting another fifteen years — which is either reassuring or a reason to build more flexible SOP revision processes, depending on your organization’s current appetite for both.

Protocol read: CONSORT 2025 has finally moved the missing-data-and-SAP-access burden onto checklist items journals will enforce — biometrics teams who waited to see whether the standard would actually bite need to stop waiting.

What to do now:

• Audit SAP templates and CSR boilerplate against CONSORT 2025 items 3, 21b, and 21c — the new missing-data documentation requirement is operational, not aspirational.
• Update protocol templates against SPIRIT 2025 and ICH M11 simultaneously; they govern the same content in parallel registers and will be enforced on the same submissions.
• Read the ICMJE January 2026 PDF directly — the landing page won’t tell you what changed.