Does Access to Science Matter for Patenting?

A scientific article can be legally available to the public yet difficult for an industrial researcher to obtain. For patent practitioners, that gap matters twice: when identifying the knowledge relevant to patentability, and when understanding the research on which an invention builds.

Our recent study in Scientometrics (co-authored with Abdelghani Maddi and Ahmad Yaman Abdin) examines the second question through 497,098 USPTO patents and 3.2 million inventor-cited scientific references, corresponding to nearly one million distinct publications.¹ Open access research is 38% overrepresented among these references after adjustment for disciplinary publishing practices. The connection to published science is particularly strong in biotechnology and pharmaceuticals. Together, these findings make literature access a practical consideration for research-intensive patent portfolios.

Following the Science into the Patent

Not all patent citations tell the same story. Front-page references may reflect disclosure obligations and examination. References within the specification can provide a closer indication of the scientific knowledge associated with the invention.

We therefore examined inventor-supplied references appearing in patent text, excluding examiner-only citations. Using the Marx–Fuegi dataset, we linked these references to OpenAlex records to identify the cited publications' disciplines and open access status. Research on in-text citations supports this approach, while recognising that citations remain an incomplete record of knowledge flows.²

A citation cannot establish that its authors' findings caused an invention. Drafting choices also shape the record. Nevertheless, a specification provides evidence of the research brought into the account of an invention, across patents associated with companies, universities and public research organisations.

The Open Access Advantage

Only 29% of the scientific references studied were classified as open access. The majority were therefore outside that category. The significant result emerges when this share is compared with the prevalence of open access in the relevant scientific disciplines.

Our Normalized Open Access Index makes that comparison. Its overall value of 1.38 means that open access publications are 38% more strongly represented in patent-linked literature than their disciplinary prevalence would predict.³ This is a measure of relative representation, rather than a causal estimate of what making an individual article open would achieve.

Separate studies help assess the causal question. Bryan and Ozcan found that the NIH public-access mandate increased patent citations to affected research by 12–27%. Probst and colleagues, studying the US Department of Energy's mandate, reported an average increase of 42% in patent use of affected publications, with small firms the main beneficiaries.⁴ These estimates concern particular policies and populations; they support the proposition that access barriers can affect industrial use of research, without turning our own 38% figure into a policy effect.

Why Biopharma Stands Out

Biology, chemistry and medicine account for approximately two thirds of the references in our study. Our Reliance on Science Index shows how unevenly that connection is distributed across technologies.

The index compares a category's share of patents citing scientific literature with its share of USPTO patents overall. Peptides have a value of 12.34; sugars and nucleic acids, 11.08; and microorganisms and genetic engineering, 9.95. Biopharmaceutical organisations, including Ionis Pharmaceuticals, Incyte Pharmaceuticals and Biogen Idec, also appear among those with the highest indices.⁵

These figures measure relative representation among science-linked patents, not the number of references per patent. They identify areas where published science is especially closely connected to patenting. For a biotechnology company choosing research partners or developing a patent portfolio, the ability to consult that science merits attention alongside the expertise needed to interpret it.

What the Sci-Hub Comparison Reveals

Adding publications covered by Sci-Hub increases the proportion of references accessible without payment from 29% to 85%.⁶ That striking difference requires careful interpretation.

The analysis matches publication identifiers against a February 2022 coverage snapshot. It does not observe downloads, subscriptions or reading behaviour, and cannot establish whether an inventor used Sci-Hub or whether a paper was available there at the relevant time. Shadow-library coverage must also be distinguished from authorised open access.

The finding therefore concerns the availability of the cited literature. It does not demonstrate dependence on shadow libraries by particular companies. Its practical relevance is the breadth of the access gap it exposes: organisations should examine whether their authorised access arrangements support the range of literature their research may require.

Practical Access and Public Availability

Although our empirical analysis concerns US patents, European patent practice offers a useful distinction. Under Article 54(2) EPC, the state of the art includes what was made available to the public before the filing date. The EPO Guidelines expressly explain that requiring payment for access does not prevent a web page from forming part of the state of the art, provided it is in principle available without a confidentiality obligation.⁷

An access barrier can therefore matter operationally even when the publication is legally public. Open access can make research easier to discover and examine; subscription status alone does not determine its relevance as prior art. Our citation results provide no basis for restricting searches to freely accessible literature.

The distinction also helps keep claims about patent disclosure precise. Article 83 EPC requires disclosure sufficiently clear and complete for the skilled person to carry out the invention.⁸ Easier access to supporting science may assist a team's examination of the technical background, but an article's open access status cannot itself establish compliance with that requirement. Our study did not test sufficiency, validity or enforceability.

Implications for Patent Teams

The practical response starts with literature discovery. Research and patent teams should combine authorised repositories with publisher platforms and subscription resources, so that access arrangements do not unnecessarily narrow the scientific material they examine.

In collaborations, access deserves explicit attention. A university laboratory, a biotechnology start-up and an external patent team may have different subscriptions and permissions. Establishing whether participants can consult the relevant publications, and arranging appropriate access where needed, can support a shared examination of the research background. These are proposed implications of the evidence, rather than interventions evaluated in our study.

For policymakers, the findings connect open access to industrial research capability. The mandate studies suggest particular benefits for firms with fewer resources, while our analysis identifies the science-intensive fields in which access may be especially consequential. Sustainable repositories and authorised publication routes consequently belong in discussions of innovation infrastructure.

Patent strategy depends partly on understanding the science surrounding an invention. Our results give practitioners a reason to ask how effectively that science reaches the people developing and assessing it. In biopharma especially, dependable access is a research capability worth examining before a missing paper becomes a missing part of the technical picture.

References

1. A. Maddi, A. Y. Abdin and F. De Pretis, “Open science, open innovation? The role of open access in patenting activity”, Scientometrics 131 (2026), 5949–5964, pp. 5949 and 5951–5952. For the authors' earlier general-audience discussion, see “Comment le libre accès aux publications scientifiques nourrit l'innovation industrielle”, The Conversation, 28 October 2026, republished by Sorbonne Université.

2. M. Marx and A. Fuegi, “Reliance on science by inventors: Hybrid extraction of in-text patent-to-article citations”, Journal of Economics & Management Strategy 31 (2022), 369–392; K. A. Bryan, Y. Ozcan and B. Sampat, “In-text patent citations: A user's guide”, Research Policy 49 (2020), 103946.

3. Maddi, Abdin and De Pretis, above, pp. 5952 and 5956–5957, Figure 2.

4. K. A. Bryan and Y. Ozcan, “The impact of open access mandates on invention”, The Review of Economics and Statistics 103 (2021), 954–967 (author manuscript); B. Probst, P. M. Lohmann, A. Kontoleon and L. D. Anadón, “The impact of open access mandates on scientific research and technological development in the U.S.”, iScience 26 (2023), 107740, Summary and Results.

5. Maddi, Abdin and De Pretis, above, pp. 5953–5956, Tables 1 and 3; p. 5958, Figure 3(A).

6. Ibid., p. 5951; pp. 5956–5957, Figure 2(A).

7. Article 54(2) EPC; EPO Guidelines for Examination (2026), G-IV, 7.5.1, on payment, confidentiality and public availability.

8. Article 83 EPC.