Publish or Perish in Biotech: Why 2025 Became the Year the Scientific Literature Started Eating Itself

There is a number worth sitting with for a moment: 13,000. That is how many scientific papers were retracted in 2023 alone, according to Ivan Oransky, co-founder of Retraction Watch, the organisation that has spent fifteen years documenting the slow haemorrhage of integrity from the published scientific record. In 2025, Nature retracted six papers in a single year — marking a significant escalation even for the world's most scrutinised journal. And those are the retractions we know about. The ones that were caught, reported, and processed through the machinery of journal correction — a machinery that, by most accounts, is running years behind the problem it is trying to address.

In biotech and medtech specifically, the stakes of this crisis are not merely reputational. They are financial, clinical, and in some cases lethal.

How We Got Here

The retraction rate for European biomedical science papers increased fourfold between 2000 and 2021, with two-thirds of those papers withdrawn for reasons relating to research misconduct — data and image manipulation, authorship fraud, and fabricated experimental results. The trend did not slow after 2021. As of August 2025, a corpus of 55,000 unique retracted papers had been catalogued from official journal sources — and researchers studying the phenomenon consistently note that the growth rate of retracted papers is accelerating faster than the growth rate of legitimate publications. The literature is, by measurable proportion, becoming less reliable over time.

The mechanism driving this is not difficult to understand, even if it is uncomfortable to say plainly. Paper mills are commercial operations that mass-produce research articles for paying authors — scientists under career pressure to publish or perish. These companies have industrialised the research process: harvesting public databases, applying standardised analytical pipelines, generating AI-written introductions and discussions, and selling complete manuscripts with guaranteed authorship slots.

These services use AI-supported production techniques at scale and sell fake publications to students, scientists, and physicians under pressure to advance their careers. A study published in 2025 analysing 17,120 publications in PubMed developed a red-flagging method with 94% sensitivity for identifying fake papers — and applied it to random samples from 2020 and 2023 to estimate the scale of the problem. The results were not reassuring.

The Medtech Problem Specifically

For investors and developers working in medical technology, the retraction crisis is not an abstract concern about academic integrity. It is a direct operational risk.

Consider the pipeline of a typical medtech venture. Clinical claims are built on published evidence. Regulatory submissions reference peer-reviewed literature. Investor pitch decks cite foundational studies. Due diligence processes lean heavily on the published record as a proxy for scientific validity. If that record is corrupted — if a significant and growing proportion of the studies underpinning commercial claims were produced by paper mills, contain manipulated data, or will eventually be retracted — then the entire downstream edifice of investment, development, and regulatory approval is being built on ground that may not hold.

The consequences extend beyond the retracted articles themselves: retracted findings continue to circulate in citations, policy documents, and downstream research. A paper retracted in 2024 may have been cited dozens of times between its publication and its retraction, each citation propagating the false finding further into the literature and into the commercial decisions built upon it. Analysis of retraction patterns has found that approximately one quarter of papers citing retracted work are themselves retracted papers — suggesting that fraud does not merely contaminate individual studies but propagates through the citation network in ways that are difficult to trace and harder to reverse.

For a medtech founder building a company on a clinical claim, for an investor conducting diligence on a biotech Series A, and for a regulatory affairs team preparing an EMA submission, this is not a background concern. It is a foreground one.

The Publish or Perish Engine

This crisis did not emerge randomly. It is a direct consequence of how we evaluate and reward scientific work. When hiring committees prioritise publication counts over research quality, when funders use impact factors as proxies for merit, and when institutions measure prestige by citation metrics, powerful incentives for gaming the system are created. Paper mills exist because academia incentivises publication in ways that make fraud economically rational.

The incentive structure is particularly acute in fields where the distance between academic output and commercial application is short — and where the stakes of both career advancement and investment returns are high. Biotech and medtech sit precisely at this intersection. The researcher who needs publications to secure tenure, the startup that needs publications to secure investment, and the paper mill that sells publications to both are participating in the same broken economy, even if only one of them is acting dishonestly.

AI is powering an arms race in the world of research misconduct, making it easier for scientific fraud to occur — and, simultaneously, for editors to identify and root out problematic papers. The detection tools are improving. But a recent landscape study found more than 32,700 suspected fake papers linked to organised paper mills, and concluded that fraudulent outputs are growing faster than corrective measures can keep up. The race is not yet won by the honest side.

What This Means for Investors and Founders

The practical implication for anyone making decisions based on published biotech or medtech research is straightforward, if uncomfortable: the published record can no longer be taken at face value without independent verification. This does not mean the science is wrong. Most of it is not. It means that the presence of a peer-reviewed publication, even in a respected journal, is no longer sufficient evidence of the claim it contains.

Due diligence in this environment requires something that neither a citation count nor an impact factor can provide: an independent, honest assessment of whether the commercial claims being made about a technology are actually supported by the evidence base beneath them. Not what the pitch deck says the evidence shows. What the evidence actually shows — traced to its sources, assessed for integrity, and evaluated in the context of what else the field knows.

This is where honest market research, conducted by people who understand both the science and the commercial context around it, becomes not a nice-to-have but a risk management tool. The investor who funds a medtech venture built on a paper-mill citation is not merely making a bad investment. They are potentially funding a regulatory process, a clinical programme, or a commercial launch that will eventually encounter the retraction of its foundational evidence — at which point the cost is not just financial. It is reputational, legal, and in clinical contexts, potentially patient-facing.

From Paper Claims to Real Evidence

The solution to the retraction crisis at the systemic level is a reform of the incentive structures that created it — a reform that is underway, slowly, through changes to hiring criteria, funding requirements, and publishing norms. That reform will take years.

The solution at the level of the individual investor, founder, or research institution operating right now, in a market where the literature cannot be trusted uncritically, is more immediate: build the independent verification into the process before the decision is made, not after the retraction notice arrives.

For a biotech startup, this means ensuring that the evidence base for your core clinical claims has been independently assessed for integrity and currency — not just cited. For an investor, it means incorporating literature verification into due diligence as a standard step alongside financial and legal review. For a university commercialisation team, it means being able to distinguish, clearly and defensibly, between the published claims your spinout is built on and the evidence that would survive serious scrutiny from a regulatory body or a sceptical investor.

The gap between what the literature says and what the evidence actually supports is, in the current environment, wider than it has ever been. Closing that gap — for your specific technology, your specific claims, your specific market — is not a matter of academic integrity. It is a matter of building something that will still be standing when the dust from the retraction wave settles. And in 2025, that wave is very far from done.