The first paper from a trial does not exhaust every question its data could answer. Reanalysis may test a different idea, combine datasets or check an earlier result. To understand what that adds, readers need to follow the participants and datasets as carefully as they follow the publications.
- Publication
- The BMJ · 2 September 2026
- Object of study
- Research publications, not a new treatment comparison
Original publication: 2 Sep 2026 · The date above refers to this brief.
Identified secondary publications
Number of publications
What the investigators counted
Taherifard and colleagues examined 336 Johnson & Johnson trials available for data sharing through YODA by the end of 2021. They identified 1,167 secondary publications through June 2025: 958 internal and 209 external. Of the trials, 265 had some secondary publication; 152 had an external one. “Internal” included an original investigator or a sponsor-affiliated author. The work received Johnson & Johnson support through Yale; several authors had YODA or company roles.
Source 1 ↗Source 2 ↗More publications do not mean more independent participants
Imagine a hypothetical trial with 500 participants. One paper reports the main endpoint, another explores age groups and a third builds a prediction model. There are now three papers, but there are still 500 original participants. Adding the paper-level sample sizes would count the same people more than once.
A new analysis can be scientifically useful without constituting an independent replication. In your study notebook, record the trial identifier, dataset and recruitment period. Those fields help reveal overlap that different titles and author lists can hide. This is especially useful when reading a review that draws on several reports from the same trial.
External research and new data answer different questions
A team outside the original investigation may bring a new question or analysis. That is different from collecting a new sample in another setting. You can have independent investigators working with old data, or the original investigators recruiting a genuinely new cohort.
Our comparison table separates these possibilities. Ask both “Who did the work?” and “Where did the observations come from?” Neither question should disappear into the broad label “independent research.” For a particular claim, the most informative next study depends on whether the uncertainty concerns the analysis, the sample, or both.
Why a publication count cannot prove the effect of sharing
This study describes a shared-data portfolio; it does not randomly compare sharing with withholding data. Its total also includes research by original investigators, rather than measuring only reuse enabled by external access. A paper count therefore cannot be read as a causal estimate of what the platform created.
Source 1 ↗What useful reuse should make easier to check
Our suggested reporting standard starts with a short chain: identify the source datasets, state the new question, explain how the analysis differs from the original, and link the analysis plan and reproducible code where access arrangements allow. A reader should be able to tell whether a result is confirmation, extension or exploration.
For future evaluations of sharing, we would also want failed requests, incomplete projects and analyses that did not become papers to be visible. Publication alone misses that work. The longer-term hope is that a participant’s contribution can support more careful questions without blurring which findings have truly been tested in new people. That is a research aim, not a patient benefit demonstrated here.
Three ways a study can be revisited
| Approach | What changes | What it can help check |
|---|---|---|
| Reanalysis | Analysis of an existing dataset | Sensitivity to analytical choices |
| Secondary question | Question asked of existing observations | An additional hypothesis within available measurements |
| Replication in new participants | New observations and possibly a new team | Whether a finding recurs beyond the original sample |
Our conceptual comparison. These are not mutually exclusive publication categories assigned by the BMJ investigators.
Your questions, answered
Is a secondary analysis less valuable than the first paper?
Not necessarily. Value depends on the question, the suitability of the data and the analysis. The label tells you when or how the data are being reused, not whether the conclusion is sound.
Can several papers from one trial go into a meta-analysis?
They can inform extraction from that trial, but repeated reports should not silently become independent participant groups. Check how the review identifies overlapping studies and selects the outcome data.
Does an outside research team guarantee a replication?
No. Team independence and a new participant sample are separate properties. Record both before deciding what kind of confirmation a paper supplies.
What should I save when I read a reused-data study?
Save the original trial identifier, the dataset source, the new research question, the prespecified plan if available, and a note describing how the analysis differs from earlier reports.
Limits of this interpretation
- One sponsor’s portfolio and potentially missed publications limit the scope of the findings.
- Publication counts do not directly measure methodological quality or benefit to patients.
- Complete supplementary and trial-level records were not independently assessed in this source check.
Sources & transparency
- Taherifard, Mooghali, Hakimian et al. (2026): Secondary use of shared clinical trial data
PubMed abstract and disclosures, publisher-indexed PDF results and the publisher’s research summary checked. Direct retrieval of the complete article/PDF failed; supplements and trial-level data were not independently assessed. A short factual account and original numerical chart are used. · Accessed 29 Sep 2026
DOI: 10.1136/bmj-2026-100460 - PubMed record for the Taherifard et al. study
Bibliographic record, abstract and competing-interest disclosures checked. This is another access route to the same study, not independent corroborating research. · Accessed 29 Sep 2026
Prepared and source-checked with AI on 29 September 2026. Press-news Team is our collective publication byline, not a claim of medical credentials or human review. No human editorial or clinical review has taken place. This is educational reporting on research methods, not medical advice. Access limitations are listed with each source. Reading frameworks and proposed follow-up tests are our commentary, not additional study findings. The photograph is illustrative.
Source check: AI source check — selected primary research and reported numbers
Clinical review: Not applicable to this educational guide
Suggest a correction