npj series sets a high bar for clinical research submission and trial transparency

Clinical research submitted to the npj series now faces a clearly defined editorial threshold: prospective trial registration, protocol alignment, and reporting discipline are not optional extras but core requirements for consideration. The journal’s policy makes plain that it will consider all types of clinical research, yet it also narrows the path to publication by demanding that investigators show their work from the first participant enrolled through final data lock. The emphasis is not cosmetic. It reflects a publishing environment in which incomplete trial registration, missing protocols, or loose outcome reporting can erode confidence in even well-designed studies.

Prospective registration sits at the center of that framework. Investigators must register clinical trials in an acceptable registry before enrollment of the first participant, and the trial number must appear in both the abstract and methods section. Primary results from nonregistered trials will not be considered. In practical terms, this prevents post hoc reshaping of endpoints after data are known, a problem that can distort efficacy estimates and safety interpretation. It also places the registry record alongside the manuscript as an accountability tool, allowing editors and reviewers to compare what was planned with what was ultimately analyzed.

The policy adopts the ICMJE definition of a clinical trial, capturing any prospective assignment of people or groups to an intervention with or without concurrent control groups in order to study a relationship between a health-related intervention and a health outcome. That scope is deliberately broad. Drugs, surgical procedures, devices, behavioral treatments, educational programs, dietary interventions, quality improvement interventions, and process-of-care changes all fall under the same umbrella when they are tested prospectively against a health outcome. Safety events and pharmacokinetic measures are included too, which means the evidentiary burden extends beyond symptom scores or survival curves alone.

A trial without a protocol is not enough.

The manuscript package must include the most recent English-language protocol, signed and dated, plus any amendments and a summary of why changes were made. If a statistical analysis plan is separate from the protocol, it must also be submitted in English and detailed enough for another analyst to reproduce the work in a similar dataset. Where the original documents are in another language, machine-readable versions are required. Editors reserve the right to request unredacted versions when needed. This is a direct response to a familiar problem in clinical publication: the temptation to simplify or obscure analytic decisions after results become available. By demanding the protocol and SAP up front, the journal ties the published paper back to a pre-specified analytic route.

The policy also states that the analysis of the primary outcome should match the prespecified plan in the final protocol and SAP, except in unusual circumstances, and any deviation from the trial registry or protocol must be justified in the Methods section. Interim analyses must be prespecified as well. Unplanned post hoc or exploratory analyses are not banned, but they must be labeled clearly in the abstract and manuscript text. For readers, that distinction matters because the credibility of an endpoint depends not only on its statistical significance, but on whether it was chosen before the data were seen. Without that guardrail, even a statistically elegant result can be clinically fragile.

Reporting standards are treated with the same seriousness. Randomized trials must conform to CONSORT 2025, cluster randomized trials to the relevant extension, and nonrandomized trials are encouraged to follow CONSORT principles as well. Observational studies must follow STROBE. Systematic reviews and meta-analyses must follow PRISMA. Biomarker studies tied to clinical outcomes are directed to STARD or the appropriate extension, and oncology biomarker studies are expected to adhere to REMARK. Prediction-model papers must follow TRIPOD, or TRIPOD-AI when regression or machine learning methods are used. Observational studies that explicitly emulate a target trial must follow TARGET. The cumulative effect is obvious: different study designs require different methods of accountability, and the journal is mapping each design to a reporting framework that fits its own risks of bias.

AI-driven clinical research receives a separate layer of scrutiny. Studies using artificial intelligence in clinical medicine are encouraged to follow the MI-CLAIM-GEN guideline and to submit the accompanying checklist. That expectation is sensible in a field where model development, validation, deployment, and clinical interpretation can easily blur together. Transparent reporting becomes even more important when the intervention is not a drug or device with a fixed molecular identity, but a model whose behavior may shift with the training dataset, threshold settings, or implementation environment.

Data availability and code availability are now part of the publication contract. Every original research manuscript must include a Data Availability Statement explaining what data exist, who can access them, how access is obtained, and whether any restrictions apply. If custom code is used, a separate Code Availability Statement is required. Supporting source data and code must be available to editors and peer reviewers when requested. For clinical trials that began enrolling on or after 1 January 2019, a data sharing plan must also be included in the registration record, and any later change must be reflected both in the manuscript and in the registry. This is where publication policy begins to influence trial culture itself: researchers must think about sharing before the first patient is enrolled, not after acceptance.

Competing interests and sponsor involvement are also handled explicitly. Authors must declare any financial or non-financial competing interests in a dedicated section, and the corresponding author is responsible for submitting the statement on behalf of all authors. Funding must be disclosed separately, along with the role of the sponsor in study design, manuscript preparation, and any other stage of the research. If the sponsor had no role, that must be stated outright. These disclosures do not eliminate bias, but they make it visible.

For a journal that accepts a wide range of clinical research, the message is unmistakable: publication will depend not only on scientific novelty, but on a study’s traceability from registry entry to protocol to analysis to final manuscript. The policy leaves little room for ambiguity on trial registration, reporting standards, and data access, and that rigidity is the point. Any manuscript that cannot show its pre-specified design, its analytic decisions, and its data governance in full will face delay or rejection before review can even begin.


Reporting by Dr. A. Nikhil Krishnan, MBBS for Medicine World Foundation, with additional data provided by British Journal of Cancer.