Reading the EHDS regulation carefully, and in particular Section 60, leaves a sense of discomfort that is not really about legal drafting, but about how knowledge is produced and legitimized. The text is cautious, layered with safeguards, and repeatedly reassures the reader that intellectual property, trade secrets, and commercial confidentiality remain protected. And yet, between the lines, there is a clear normative direction: data sharing for secondary use is not just permitted, it is encouraged, normalized, and framed as a collective good.
Section 60 is emblematic of this tension. It does not mandate disclosure of proprietary datasets, including clinical trial data, but it establishes an expectation that such data may, under certain conditions, be made available through regulated access mechanisms. The burden subtly shifts. Refusal is no longer a default right exercised silently; it becomes a decision that must be justified, documented, and potentially scrutinised. In regulatory terms, this is a soft but consequential shift in power.
From the perspective of a data holder operating in a competitive environment, this raises an obvious question: why would any rational organisation voluntarily make high-value proprietary data accessible, even under controlled conditions, when those data are the result of substantial financial investment and strategic risk-taking? Clinical trial data, in particular, are not merely scientific artefacts; they are assets that underpin regulatory exclusivity, lifecycle management, and future development strategy. The EHDS text recognises this in principle, but it simultaneously introduces a framework in which the legitimacy of withholding such data is no longer taken for granted.
What makes Section 60 particularly interesting is not what it explicitly requires, but what it seems to anticipate. It anticipates a future in which health data are increasingly treated as infrastructure rather than property, and where the moral and political weight of public interest grows relative to private control. This is not accidental. EHDS is part of a broader European project that seeks to rebalance innovation, public health, and sovereignty over data. In that sense, the regulation is less about immediate data flows and more about setting norms that will harden over time.
However, this normative shift comes with consequences that are barely discussed in the regulation itself. One of the most important is selection bias — not the familiar kind arising from study design or patient inclusion criteria, but a systemic bias rooted in incentives, governance, and commercial reality. Under EHDS, the datasets that become accessible for secondary use will not be a random or representative sample of all health data generated in Europe. They will disproportionately reflect data that are easier to share, less commercially sensitive, publicly funded, or politically defensible to expose.
In contrast, data that are tightly coupled to competitive advantage — early development trials, innovative care pathways, proprietary platforms, and privately governed real-world datasets — will be selectively absent, restricted, or heavily transformed. This absence will not be accidental. It will be the rational outcome of actors responding to the incentive structure created by EHDS. The result is a non-ignorable selection mechanism, where data availability is correlated with dimensions that matter for interpretation: maturity of therapies, standardisation of care, and alignment with public-sector delivery models.
Section 60, together with other parts of the regulation that encourage data sharing even when data are privately held, implicitly assumes that safeguards and access controls are sufficient to neutralise competitive risk. But this assumption underestimates the strategic nature of data withholding. Organisations do not only protect data because of re-identification risk or trade secrets in the narrow sense; they protect data because availability itself can shape narratives, benchmarks, and expectations. Making some data visible while other data remain invisible can influence regulatory science, HTA methodologies, and ultimately perceptions of value.
A particularly plausible scenario is one in which EHDS-enabled evidence increasingly reflects healthcare systems and therapeutic areas where public provision dominates and innovation is incremental rather than disruptive. Over time, this could lead to a form of epistemic lock-in, where what is measurable becomes what is considered normal, and what is not shared becomes marginal in evidence-based decision-making. The danger here is not malicious misuse, but quiet drift: large-scale analyses with impressive sample sizes that subtly privilege certain models of care and penalize others.
There is also a temporal dimension to consider. In the short term, EHDS will likely amplify existing disparities between countries with mature data infrastructures and those with fragmented or hybrid systems. In the longer term, as EHDS norms stabilise, access to funding, regulatory goodwill, or collaborative opportunities may become implicitly linked to a willingness to share data. At that point, the line between encouragement and obligation may blur, even if the legal text remains unchanged.
None of this implies that EHDS is misguided or bound to fail. On the contrary, the regulation addresses real problems: duplication of effort, underuse of existing data, and fragmentation of evidence generation across Europe. But, its success will depend less on the volume of data made accessible and more on the intellectual honesty with which limitations are acknowledged. If EHDS-derived evidence is treated as a curated, politically shaped layer of reality rather than a neutral mirror of European healthcare, it can be enormously valuable. If not, it risks creating a false sense of completeness.
Section 60 therefore deserves attention not because it forces disclosure, but because it signals a shift in how legitimacy around data ownership and access is constructed. The most important question is not whether companies will share proprietary data, many will not, and rationally so. The real question is whether regulators, HTA bodies, and researchers will remain alert to the biases introduced by who shares, who does not, and why. EHDS will shape not only what data are available, but what kinds of questions can be asked with confidence.
In that sense, the regulation is less a technical instrument than a long-term experiment in evidence governance. Its outcome will depend on whether Europe is willing to confront the uncomfortable reality that more data does not automatically mean better knowledge, especially when the act of sharing itself becomes a selective filter.
