Should I provide my health data for research? Citizens assessing the value of data provision

Following data and the connections they create

The promise of big (health) data is tied to the movement of data, as well as to their interoperability, i.e., their capacity to link with other data types from other sources (Leonelli, 2020). Data always need to be seen as relational in the sense that they do not “pre-exist their generation” (Kitchin, 2014, p. 19), but are connected to places, practices and material conditions of production (Gitelman and Jackson, 2013). Data never simply “represent and thus document specific aspects of the world,” but are “historical entities which (…) evolve and change as their life unfolds and merges with elements of their environment” (Leonelli, 2020, p. 6–7). In their evolving, data are what Klaus Hoeyer calls promiscuous, i.e., “relatively indiscriminate in their relation to users: what is seen as data on a thing for one purpose can always become seen as data on another aspect of that thing and be used for another purpose and by another user” (Hoeyer, 2020, p. 2). In a similar way, the value of health data is not absolute or intrinsic but is always related to the context (Fiske et al., 2023) and to the actors and purposes of use (Wilson et al., 2020).

The term data donation is often used to describe the act of making data available for research, highlighting the relational nature of this practice. It typically implies a non-commercial intent (Prainsack, 2019; see also Aitken et al., 2016) and aligns with the idea of a gift—something offered beyond immediate exchange. Yet, a gift establishes social bonds through cycles of reciprocity (Mauss, 1990), even if delayed or indirect. It thus can serve as a “systematic means for individuals and groups to articulate and reciprocate recognition, and thereby determine and shape identities” (Hummel et al., 2019, p. 29). In this sense, data donations can carry expectations, such as contributing to the welfare state with the hope of future benefits like appropriate medical care (Felt et al., 2009). While we draw on the sensitivities the donation metaphor brings, this article adopts the term data provision for research to better reflect citizens’ diverse valuations of making data mobile and reusable (Metzler et al., 2023), without assuming a fixed relational model.

With this in mind, we adapt the notion of data journeys (Bates et al., 2016; Leonelli, 2020) to trace citizens’ visions and imaginations of appropriate sites and practices for (their) health data. Data journeys designate the “movement of data from their production site to many other sites in which they are processed, mobilised and re-purposed.” (Leonelli, 2020, p. 9) The concept of data journeys is grounded in the observation that all data result from “contingent and contested social practices” (Dalton and Thatcher, 2014), which privilege specific understandings of the world. To focus on the movement of data between and through sites of data practices (Bates et al. 2016) and being attentive to the fact that “values of data are not pre-defined, but emerge and change over time and space as they flow through interconnected but different socio-cultural contexts each with their own conceptual frameworks and value systems” (Bates et al., 2015, p. 14), will sharpen our attention to how this matters to citizens. When the latter are asked to contribute to data's movement by providing data for research (or allowing the secondary use of health data), this implies that they need to make an assessment whether they accept data entering into relation with other actors, sites and practices and, possibly, obtain new meanings and realize different values. By being attentive to citizens’ anticipated data journeys, we draw on Bates and co-authors’ emphasis on the socio-material nature of data movements and Leonelli's notion of data mutability—the idea that data are continually shaped by the assemblages they become part of, which in turn influence how they are understood and valued.

Valuation constellations

When citizens consider making their health data available for secondary use via a data infrastructure, they engage in the social practice of valuation. This involves assessing the value of the data itself, the infrastructure and the relations it fosters, the actors managing it, potential future users, and the broader governance structures. These (e)valuations influence whether citizens perceive the process of data provision as reasonable and aligned with their values—such as equitable access to healthcare or minimizing power differentials. As noted earlier, the value of data is always context- and purpose-dependent. Data should be understood as continually journeying through different sites and actor constellations, rather than as a static entity. The specific value health data holds, thus, varies: it is different for tech corporations, for citizens contributing data, and for researchers using it—and even the same actor may value a dataset differently depending on the context or situation (Wilson et al., 2020, p. 2). Moreover, datasets can shift in value as they move into new relations. Accordingly, Wilson and colleagues advocate for speaking of values—in the plural—when discussing datasets, urging an investigation that is attentive to diverse, situated valuation practices. This is much in line with Tamar Sharon's (2018, p. 4) argument that a framework that simply contrasts “public benefit” with “private, corporate gain” is too narrow, overlooking other evaluative regimes at play. She reminds us of the plurality of visions of the common good that can be mobilized in the context of data provision for research.

In order to better understand how citizens develop a position on the question whether they should provide their digital health data for research and to account for the multiplicity and context-dependence of values, we decided to follow the suggestion of Waibel et al. (2021) to think in valuation constellations when studying citizens’ assessments. This means that we do not focus on singular moments of valuation, treating them as isolated events as this would “neglect the interconnectedness of moments across situations and social fields” (Waibel et al., 2021, p. 34). Instead, we see citizens’ valuations of the health data platform and the data provision process as “temporally and spatially situated” and pay close attention to how “valuation situations are linked in various ways” (p. 35) and what role interactions between different participants in our research play in valuation practices. Discussion groups—particularly due to their duration (Felt, 2016)—offer ample space for the development of complex interactions, allowing us to observe multiple valuation constellations and the shifts between them.

Valuation constellations have several components: the valuee (what exactly gets valued when thinking about a health data platform and providing data for secondary use), the valuator (in their diverse potential roles as citizens), and the audience toward whom the expression of value is (symbolically) directed (Waibel et al., 2021, p. 38f.). Furthermore, rules of valuation (p. 41f.), often defined by the setting in which a valuation takes place, and infrastructures of valuation (“the material contexts that facilitate or restrict valuations, p. 48), matter. Citizens’ assessment of the provision of health data for research is thus assumed as neither clear-cut nor stable across situations during a discussion. In taking the stance of the valuator and speaking to various projected audiences and about specific of parts of the platform infrastructure and its functionalities, citizens can assume diverse roles in making situated assessments. They might also change roles within one conversation, e.g., at one time speaking as a citizen concerned about data protection, in another moment drawing on previous experiences as patients or speaking in the name of others who are not at the table. Furthermore, actors might value different parts of the data infrastructure differently, embracing some elements while being concerned about others or even rejecting them. An analytic angle that is attentive to the “triad of valuator, valuee and audience” (p. 35) helps us grasp the complexities and multilayeredness of assessments and to observe how valuations are interconnected. To understand such multi-situational valuation constellations will then be essential for building a robust health data infrastructure as imagined in the framework of the EHDS.

Multiplicity of data values

The statement that health data holds intrinsic value has become a prevailing trope in recent years (Fiske et al., 2023), a sentiment that was also well reflected in our empirical material. Among our participants, there was a broad and quasi-immediate consensus that data is, by nature, a valuable asset. This assessment echoed the market repertoire of justification that often frames data as key driver of wealth creation (Sharon, 2018). One such instance from our discussions illustrates this perspective, with three participants swiftly agreeing that, in contemporary society, data can be equated to money—highlighting its undeniable economic significance.

DW1_IP2: I have to say, data are the currency at the moment. There is nothing more valuable

The combination of assumed high monetary value of data and concerns about its potential exploitation prompted citizens to expect stringent protective measures. When asked where they saw the appropriate, i.e., sufficiently protected, place for health data, one participant used the analogy of “the national bank, because these data are worth pure gold, and the national bank protects the gold reserve and money” (DW4_IP1). Citizens frequently used analogies to banking to emphasize the need for a high-security environment to protect their data. This comparison also emerged in their visions of data protection and control processes, pointing to banks as institutions that have already developed trusted and secure systems. By positioning themselves as clients who entrust their savings to a bank, participants were able to express both their expectations and concerns about the handling, security, and protection of their data.

The high general value of health data was rarely questioned by citizens—as long as their thinking remained within the general market logic. In this frame, data were seen as intrinsically valuable, and citizens adopted roles such as the informed participant or the client who trusts institutional safeguards, much like with other high-value assets. However, when asked to assess the specific value emerging in concrete, situated practices, citizens became more cautious—both in their perceived ability to assess such scenarios and in their willingness to take on an active role in them. The following participant opens such a reflection:

DW3_IP3: What would I donate? I cannot judge that, if I am not a physician, yeah? Then maybe it would also not make sense, no?

Many participants shared the view that they would not be able to select specific datasets for secondary use in a meaningful way. Still, they wanted to be able to decide on the general path the data could take, and with this the relations the data could enter into. We can see this particularly well in the following exchange.

DW3_IP3: The selection is quite the challenge (…) for the patient: what do I share and what do I not share, no?

In this interaction, the participants shifted the focus from the concrete data to be selected towards deciding on its anticipated pathway, thereby reinstating their active role. Several potential dimensions can be identified here: the research purpose, which would allow them to assess whether the potential new contexts and relations of health data are appropriate or not; the concrete research project as information from which they could infer the appropriateness; and the actor groups who may use the data, thereby touching on the problematized configuration of data use by commercial actors, which has been well described in the literature (Aitken et al., 2016; Cascini et al., 2024).

Having the right and possibility to define the general pathway would prevent researchers from doing “whatever [they] want” with the health data provided for research, thus sustaining the data value as much as their active position. This formulation was also used in another group discussion. In both cases, the participants clearly did not want to provide data without knowing where it might end up and without being able to assess the purpose or context of use. They did not want to delegate the decision of what may be done with ‘their data’ to someone else and risk a value shift, given that the value of health data is always related to its context and purpose (Fiske et al., 2023). This was further elaborated by the following participant:

DW4_IP3: Not just generally, yes, here you have my data, do whatever you want with it. I would also like to know for what. (…) What came out, whereby have I maybe helped with my data, yes?

Assessing data relations: Transparency, symmetry, disentanglement

As indicated in the previous section, when citizens anticipated data pathways, they assessed the relations they wanted to articulate and strengthen through their data provision (Prainsack, 2019), and whether their concept of data value mapped with the value systems of the contexts they flow through (Bates et al., 2015). To be able to make such an assessment, citizens demanded transparency on the situation and modes of value generation, given that data value is related to context, purpose of use, and actors involved (Fiske et al., 2023; Wilson et al., 2020). In their explorations, our participants largely embraced the role of generous research supporters, seeing themselves as actively helping and contributing to research. For many, it was without question that health data collection is connected to value prospects and the expectation of future benefits. And in principle, they saw nothing wrong with this, as efforts and resources would need to be invested by others in collecting health data and making it available through developing an enabling infrastructure. However, in exchange for providing what they perceived to be their health data, they wanted to be taken seriously as equal partners. This included being addressed with unambiguous statements—“talk straight!” (INT_IP8)—and have transparency on data circulation, the aims and logics of value generation and the actors involved in it. All this would allow them to assess whether activities are in line with their own understanding of data value. Conversely, a disconnected sense of “shooting data up into the cloud” (APP_test_IP4) without understanding where it has gone, who is harnessing it and for what reason, can lead to feelings of anxiety and would not necessarily invite providing data for research.

A second crucial dimension in citizens’ assessments of the relations that the data provision forms or strengthens was whether or not the relations are symmetrical, which allows the integration of value systems that sometimes seemed incommensurable. Across the material, participants drew on the notion of the donation to make sense of data provision, which means to give it away for free with the expectation that it is used for the collective good (Skovgaard et al., 2019). However, many participants were wary about making health data available for free when they could theoretically be transformed into a commodity and taken out of circulation for the collective. The pharma industry, in particular, was perceived as engaging in such problematic transformations. In one discussion, participants were rather harsh in their assessment of the pharma industry:

DW2_IP1: Well, they are crooks, yes.

Value generation and value distribution, thus, must be symmetrical for citizens to consider providing their data in the first place. Symmetry in this context means data provision and data use do not change the tacitly assumed framework. If citizens provide data freely, it must be ensured that benefits can also be freely returned. Symmetry can be framed collectively, in the sense of a “social benefit [that] is to be achieved” (DW4_IP1). For some participants, it was important to emphasize that benefits must be public and fair and not just materialize for those who can afford expensive medication, reproducing preexisting inequities in healthcare. The outcomes of research based on data donation thus should “not just benefit the rich people, but really [be available to] everyone who needs it” (DW3_IP3). Expectations of symmetry can also relate to more direct beneficial effects, such as more personalized healthcare, receiving information on research that was conducted on the basis of one's data, or, as in one case, indirect benefits of preferential treatment, based on seeing themselves as part of the chain of value generation: “If I am taken to the hospital (…) then there should be a benefit for me” (INT_IP5). In other cases, participants identified the potential for individual benefits when discussing rare or life-threatening diseases or chronic conditions, through the faster circulation of data and, subsequently, research insights into the clinical/practitioners’ realm.

In contrast to this, asymmetrical value generation and distribution means that corporate interests stop the return of benefits to the collective realm or that corporations yield economic gains based on something that was given in a gift logic. There were some instances of citizens developing strategies for dealing with perceived asymmetries, for instance, by jokingly turning the provision relationship around—“pharmaceutical companies, they should donate something to us, no?” (DW3_IP3)—or demanding concrete savings—“I would also like to see a discount on my medicine” (DW2_IP1). In this way, the potential asymmetry in the generation and distribution of value was remedied by them changing the logic and demanding economic benefits: if the pharma industry makes economic benefits and the returns are also based in an economic logic, the exchange's character is again symmetrical and, thus, appropriate.

Finally, having the possibility of initiating the end of potential data journeys and disentangling established relations was another key element of citizens considering providing health data for research.

APP_test_IP4: If my data will be anonymized, if they could trace it back to me (…) but also if you wish to withdraw your consent - if it's possible (…) - what traces will be left of you?

A key notion here is “the trace,” i.e., data traces, that can come to haunt us (Thylstrup, 2019). Both instances of traces/tracing point to the assumption that even if data has been anonymized or deleted, there may still be an imprint that does not disappear. This is related to data mutability and data having been integrated into broader datasets, a concern that the following citizen expressed:

INT_IP1: What does withdraw actually mean when the data is already shared, when it's already part of (…) a larger bucket of other data. Is it really withdrawn from there or does it just mean, data generated anew will not be added anymore?

Choice by proxy: Governance and responsibility

In this section, we shift the attention to the relations citizens envisaged with those in control of their data. By looking at how they assessed the governance of data provision, we engage with another layer of valuation that mattered for citizens finding a position towards secondary use of health data. As Waibel and colleagues have highlighted, moments of valuation are interconnected across situations (Waibel et al., 2021). The user role foreseen for citizens in the health data platform had primed them as empowered: they were supposed to collect and manage their health data in their platform account and could in detail decide with whom to share what (or not). Many initially approached the data provision process from the active and empowered user position inscribed in the narrative of the health data platform and projected it also onto data provision, expecting fine-grained decision-making possibilities.

DW4_IP3: I would not simply generally provide my data so that any scientific research can be done with it, but I want to be informed: what kind of enterprise is this; is it (…) for the pharmaceutical industry or is it a syndrome that is being investigated (…). Every time, my data are being used, I want to know, for what.

In contrast to the above-mentioned ideal vision of a more dynamic consent, our analysis in the project is based on the data provision process built on a broad consent approach (Steinsbekk et al., 2013), which pre-framed citizens’ choices. Citizens could only select which of their collected health data types they wanted to make available for research. However, they could neither select nor exclude those accessing the data (e.g., a specific research project or a group of actors, such as commercial companies) or define a specific research purpose. Instead, the choice on context and purpose was outsourced to a Data Access Committee (DAC). From the project side, the DAC was envisaged to determine the scope of data access, by whom, and for what purpose, according to specific and publicly available data access criteria. By citizens, the DAC was perceived as a powerful agent in the data provision process, as “they make the ultimate decisions. It just seems very authoritative” (INT_IP1). Such a configuration means that citizens need to trust an entity to act on their behalf, deciding who can access the data they collected and provided.

Understanding the DAC as proxy for citizens’ choice and decision-making, citizens thus began to assess the DAC in lieu of those actors, places, and purposes the data would ultimately be entangled with. To come to such an assessment, they requested detailed information about the DAC's composition, the criteria for accessing data, and how rigorously these would be applied. The trustworthiness of the DAC and the reliability with which it would apply the specified data access criteria were seen as depending on its composition, as “the promised procedures are only as good and reliable as the people responsible for them” (INT_IP5). Moreover, for citizens, positioning themselves as active and critical, it was not only about the composition of the DAC as an institution but also about the affiliations and background of its members, how the committee came into being in the first place and who was involved in this decision.

If citizens accepted that the DAC as proxy makes decisions about who may use data for research and for what purposes, they also expected a robust distribution of responsibility regarding the risks inherent in data provision. In a dedicated section of the informed consent for data provision, citizens can read about (1) the central risk of disclosure of private information and the potential but concrete consequences, e.g., feeling stress, anxiety, or embarrassment due to disclosed information, and (2) about being informed of a data breach within the timeframe given by the GDPR. In principle—and in line with the direct form of communication mentioned in the previous section—such transparency was appreciated.

INT_IP1: I like that it's just very clearly stated, these are the risks (…); but also, we’ve taken these measures (…), we’re following the law and obviously there's always remaining risks. But I just felt (.) I just liked reading it; it felt very honest.