Open Science and Data Sharing: Where Biomedical Research Is Heading
The NIH now requires data sharing plans for most funded research. The "Nelson memo" changed federal open access policy overnight. The direction of travel is clear - but the tension with commercial interests in pharma research is real and unresolved.
In August 2022, the White House Office of Science and Technology Policy issued a memorandum directing all federal agencies that fund research to make the results of that research freely available to the public immediately upon publication - no embargo period. The document, known as the "Nelson memo" after OSTP director Alondra Nelson, did not create a new law. It updated existing open access policy. But its effect was immediate: every major federal research funder in the United States began developing implementation plans, and the 12-month embargo that had been standard practice for years was eliminated for federally funded work. This was the most significant shift in scientific publishing policy in decades, and its full effects are still unfolding.
The NIH Data Sharing Mandate
The NIH's updated data management and sharing policy, which took effect in January 2023, requires that all NIH-funded research include a data management and sharing plan - describing what data will be generated, how it will be documented, and where and when it will be shared. For most research involving human participants, this means depositing data in an NIH-designated repository within a specified timeframe after publication.
The mandate is significant not just because it requires data sharing, but because it requires planning for data sharing from the beginning of the research - which changes how studies are designed, how consent is obtained, and how data management is resourced. Investigators who treated data management as an administrative afterthought are now required to integrate it into the scientific plan from grant application stage.
FAIR Data Principles
The FAIR principles - Findable, Accessible, Interoperable, and Reusable - were published in Scientific Data in 2016 and have become the standard framework for evaluating data sharing quality. Findability requires that data be assigned persistent identifiers and indexed in searchable resources. Accessibility requires that the data or its metadata be retrievable through standard protocols. Interoperability requires the use of shared vocabularies and formats. Reusability requires sufficient provenance documentation and licensing information for others to use the data legitimately.
In practice, much "shared" data falls short of the FAIR standard. Datasets deposited in general-purpose repositories without data dictionaries, without code books, and without documentation of preprocessing steps are technically shared but practically unusable. FAIR compliance requires effort and resources that not all research groups allocate adequately.
Open Access Publishing
The shift to open access publishing - making journal articles freely available without subscription - has been accelerating for a decade, driven by funder mandates, institutional policies, and the Nelson memo. The two dominant models are "gold" open access (the article is free to all readers immediately upon publication, funded by an article processing charge paid by the author or their institution) and "green" open access (the author posts a version of the accepted manuscript to a repository, with or without an embargo). A third model - diamond open access - involves journals that charge neither authors nor readers, sustained by institutional or funder support.
The article processing charge model has created its own distortions. For-profit publishers have responded to mandates by launching high-volume open access journals where the business model is acceptance, not selectivity. Predatory journals - exploiting the APC model to publish without genuine peer review - have proliferated. The Directory of Open Access Journals (DOAJ) maintains a quality-filtered list, but researchers and evidence synthesisers need to be aware that "open access" and "quality-reviewed" are not synonyms.
The Tension With Commercial Interests
The open science agenda sits in direct tension with commercial interests in pharmaceutical and biotech research. Clinical trial data is a commercial asset. Detailed patient-level data from drug development programmes, if made publicly available, enables competitors to perform secondary analyses that could be used in regulatory submissions or market access applications for competing products. The legal and strategic frameworks that govern data exclusivity - market exclusivity periods, trade secret protections - were designed precisely to protect this asset.
The result is a split landscape: academic and publicly funded research is moving rapidly toward open science norms, while commercially funded research - which accounts for the majority of late-phase clinical trial activity - remains largely closed, subject to selective disclosure through regulatory submissions and voluntary sharing platforms like ClinicalStudyDataRequest.com.
What This Means for Evidence Synthesis
For systematic reviewers, the open science shift is directionally positive. Pre-registration is more common, reducing outcome reporting bias. Open access reduces the cost of obtaining full-text articles. Data sharing, where it happens, enables more granular meta-analyses using individual patient data. But the landscape is uneven, and reviews that draw on a mix of fully open academic studies and commercially funded trials with selective reporting still face the same grey literature and publication bias problems that have always existed.
At NousLab, we are building evidence synthesis infrastructure that is designed to incorporate open science outputs - pre-registered protocols, open data, preprints - alongside conventional published literature, treating each source with appropriate scrutiny. See how NousLab handles diverse evidence sources.
