Fictional Evidence, Real Patients: The New Phase of Market-Based Medicine

In an earlier article, I described how evidence-based medicine (EBM) can slide into market economy-based medicine when diagnostic thresholds, trial designs, and treatment strategies are tuned more to commercial interests than to patient benefit. That analysis still assumed one fundamental safeguard: that the studies being cited actually exist. Today, even this assumption is under threat. A new and more dangerous phase has begun, in which fabricated references—citations to studies that were never conducted—enter the biomedical literature and quietly accelerate the shift toward market-driven medicine.

For years, market forces have influenced how real data are produced and interpreted. Diagnostic criteria are lowered to expand treatment-eligible populations. Trials are designed as add-on studies that almost cannot fail. New, expensive drugs are tested in ways that favor them over older, cheaper ones. All of this distorts medicine, but at least it operates on a substrate of genuine studies. Fabricated references change the game. They allow the appearance of evidence without the burden of actually generating it. A trial that was never run can still be cited as if it showed that a particular stent, anticoagulant, or biologic is effective.

This mechanism is especially dangerous because it works in the same direction as existing market incentives. Pharmaceutical and device companies benefit from the construction of dense, impressive-looking reference lists that suggest a robust evidence base for their products. Paper mills and careless use of AI writing tools now provide a cheap way to generate such lists, complete with plausible-sounding titles, real author names, and credible publication years. Once these citations are embedded in review articles or guidelines, they are rarely questioned. Clinicians see a forest of references and understandably infer solidity where there may be none.

The result is a shift in where distortion occurs. Previously, the main problem was bias in trial design, selective reporting, and aggressive interpretation of marginal findings. Now the problem reaches deeper, into the very existence of the purported data. Evidence-based medicine becomes vulnerable at its foundation. If guidelines recommending broader use of new antihypertensives or anticoagulants are partly supported by studies that do not exist, the line between science and marketing is no longer blurred—it is crossed.

What can be done to prevent this new phase of market-based medicine from taking hold? Moral appeals to researchers are not enough. The system itself must change.

First, reference verification must become a standard part of editorial practice. Journals should automatically check every citation against major databases before a paper enters peer review. References that cannot be found should be flagged, corrected, or removed. This will not stop sophisticated fraud, but it will greatly reduce the casual introduction of invented references via AI tools or sloppy drafting.

Second, transparency around reference integrity should be built into indexing and evidence synthesis. Article records could carry a simple indicator of whether their references have passed automated verification. Systematic reviewers and guideline panels should not only evaluate study design and risk of bias, but also the reliability of the citation network underneath their conclusions. Where key references cannot be verified, this uncertainty should be explicitly acknowledged rather than hidden beneath dense bibliographies.

Third, existing literature must be audited. High-impact areas—cardiology, oncology, infectious disease—should be systematically screened for clusters of fabricated references, particularly in review articles that directly influence guidelines. Where fabricated citations materially change a paper’s apparent evidentiary support, corrections or retractions are necessary. This is laborious work, but it is essential if we are serious about repairing the evidence base instead of merely admiring it.

Fourth, professional norms and training should evolve. Use of AI writing tools should be openly declared, and authors must be held fully responsible for the existence and accuracy of every reference in their work. Education in research methods should now include “evidence verification”: not only how to search for literature, but how to confirm that cited studies exist and actually say what is claimed.

Finally, clinicians need to adopt a more systemic skepticism. The label “evidence-based” should no longer be accepted at face value. The deeper question is: what functional system produced this recommendation? What incentives shaped the trials, the reviews, and now even the references themselves? If the useful result of the publishing system has drifted from truthful representation of reality to throughput and profit, its outputs must be treated with corresponding caution.

Fictional evidence does not harm fictional patients. It harms real ones. Preventing the next phase of market-based medicine requires realigning the system’s criterion of success—from generating publishable claims that favor products, back to generating verifiable knowledge that serves the living organism whose fate is at stake.

This article was developed in response to the research letter “Fabricated citations: an audit across 2·5 million biomedical papers” published in The Lancet on May 7, 2026 (DOI: 10.1016/S0140-6736(26)00603-3).

You can learn more by reading our e-book or listening to our audiobook

Mykola Iabluchanskyi Yabluchansky