npj Series Raises the Bar for Clinical Trial Submission Standards
The npj series journals require trial registration before the first participant is enrolled, and they make the trial number a mandatory element of both the abstract and methods section. Manuscripts describing primary results from unregistered trials will not be considered. That policy places registration at the center of scientific credibility, not as a formality to be patched in after a study is finished.
The definition used here is broad and clinically practical. Any research project that prospectively assigns people, or groups of people, to an intervention with or without a concurrent control group qualifies as a clinical trial when it is designed to examine the relationship between a health-related intervention and a health outcome. Drugs, surgical procedures, devices, behavioral treatments, educational programs, dietary interventions, quality-improvement initiatives, and process-of-care changes all fall within that scope. In other words, the journal treats the infrastructure of clinical care itself as a potential intervention when it is tested prospectively.
Trial registration is not just about transparency. It is a safeguard against outcome switching, selective reporting, and retrospective reshaping of hypotheses after data are known. When a protocol is publicly registered before enrollment, readers can judge whether the published results followed the study as planned or drifted toward a more favorable narrative. In a field where small analytic changes can alter how a treatment appears to perform, that distinction has direct implications for patient care and research integrity.
Protocols, Statistical Plans, and the Paper Trail Behind Primary Outcomes
Submissions must include the most recent clinical trial protocol in English, signed and dated. If the protocol was amended, the changes must be visible, and the rationale for those changes must be provided. A separate statistical analysis plan is also required when it exists, and it must be detailed enough to allow another analyst to reproduce the work in a similar dataset. For editors and referees, this is not administrative clutter; it is the document trail that shows whether the primary endpoint analysis remained anchored to the original design.
The journal also requires machine-readable versions of original documents when the source material is not in English. Sensitive content may be redacted when necessary, but only if the redaction does not obscure the design, analysis, or outcome reporting. Editors reserve the right to request unredacted documents when needed for review. That policy protects both confidentiality and scientific auditability, two principles that often collide in multicenter or commercially sponsored research.
Primary outcome analyses must match the prespecified protocol and statistical analysis plan unless an unusual circumstance justifies a deviation. Any deviation from the registry entry or protocol must be explained in the Methods section. Published trial reports must also include the protocol alongside the final paper after acceptance. For clinicians reading the final article, the protocol is therefore not an accessory; it is part of the evidentiary record.
Interim Analyses and the Limits of Midstream Interpretation
Interim analyses are acceptable only when they were prespecified in the protocol. Preliminary reads from ongoing trials that were not planned in advance may be considered case by case, but unplanned post hoc or exploratory analyses must be labeled clearly as such in both the abstract and the manuscript text. This matters because interim looks can be biologically persuasive while still being statistically unstable. A therapy that appears to work early may later fail once the full event curve is observed, especially when the outcome depends on time-to-event patterns or delayed toxicities.
Clinical readers understand the temptation here. Early signals can be compelling, especially when a treatment seems to alter a serious outcome or when safety concerns emerge abruptly. Yet unplanned analyses can also exaggerate benefit, inflate false-positive findings, or distract from the actual prespecified endpoint. The journal’s rules force investigators to separate what was planned from what merely looked interesting after the database began to speak.
Reporting Frameworks Span Randomized, Observational, and Biomarker Studies
The editorial policy extends well beyond randomized trials. Randomized studies are expected to follow CONSORT 2025, and cluster randomized trials must comply with the appropriate extension. Non-randomized trials are encouraged to use CONSORT principles as well. Case reports and case series used within a trial context must conform to the relevant N-of-1 extension, while treatment delivered outside a formal trial, such as compassionate use or expanded access, must meet ethical approval and consent expectations aligned with CARE and the Declaration of Helsinki.
Observational studies must follow STROBE, systematic reviews and meta-analyses must follow PRISMA, biomarker studies linked to clinical outcomes must follow STARD or an applicable extension, and oncology biomarker work should adhere to REMARK. Prediction models for diagnosis or prognosis must follow TRIPOD or TRIPOD-AI when machine learning is involved. Studies that emulate a target trial to estimate causal effects must follow TARGET. The policy reads like a methodologic map because each design carries its own failure modes, and each demands a different reporting discipline.
Data Sharing, Trial Registration, and What Reviewers Expect to See
Data availability statements are required for all original research manuscripts. When custom code is used, a separate code availability statement must also be included. Editors and peer reviewers may request source data and code during submission, and failure to provide them can slow or derail review. For trials that began enrolling participants on or after 1 January 2019, the registry record must include a data sharing plan, and any change to that plan after registration must be reflected both in the manuscript and in the registry itself.
A practical trial report, under these rules, needs more than a results paragraph and a p-value. It should include the trial registration number, the type of trial, the prespecified primary endpoints, a summary of safety data, and a clear statement about whether the trial met its prespecified endpoints. The main text should also contain a CONSORT-style flow diagram, baseline participant characteristics, adverse event reporting, and exact dates for first enrollment, last enrollment, and data lock.
The policy’s direction is unmistakable: clinical research is expected to be reproducible before it is persuasive. The primary endpoint does not stand alone; it is tied to registry records, protocol amendments, prespecified analyses, and data-sharing commitments. For investigators, that means the manuscript must reflect the trial as it was truly designed and conducted. For clinicians, it makes the published report more usable, because the reader can see exactly what was planned, what changed, and what the data support without having to guess at the space between those points.