The practical ethics of repurposing health data: how to acknowledge invisible data work and the need for prioritization

Abstract

Throughout the Global North, policymakers invest in large-scale integration of health-data infrastructures to facilitate the reuse of clinical data for administration, research, and innovation. Debates about the ethical implications of data repurposing have focused extensively on issues of patient autonomy and privacy. We suggest that it is time to scrutinize also how the everyday work of healthcare staff is affected by political ambitions of data reuse for an increasing number of purposes, and how different purposes are prioritized. Our analysis builds on ethnographic studies within the Danish healthcare system, which is internationally known for its high degree of digitalization and well-connected data infrastructures. Although data repurposing ought to be relatively seamless in this context, we demonstrate how it involves costs and trade-offs for those who produce and use health data. Even when IT systems and automation strategies are introduced to enhance efficiency and reduce data work, they can end up generating new forms of data work and fragmentation of clinically relevant information. We identify five types of data work related to the production, completion, validation, sorting, and recontextualization of health data. Each of these requires medical expertise and clinical resources. We propose that the implications for these forms of data work should be considered early in the planning stages of initiatives for large-scale data sharing and reuse, such as the European Health Data Space. We believe that political awareness of clinical costs and trade-offs related to such data work can provide better and more informed decisions about data repurposing.

Medical Secretaries’ Registration Work in the Data-Driven Healthcare Era

“…Organization of the work presented in this paper can be seen as a consequence of the blurring line between medical and nonmedical data. While clinical documentation is clearly the physician's responsibility, the administrative requirements of clinical documentation go beyond the clinical domain to the extent that some researchers question whether requirements for secondary data purposes take precedence over the primary clinical purposes in the design of EHRs [2]. In the example of our case, a trajectory-oriented registration regime in which all activities are registered according to their relation to a specific illness trajectory increases the socio-technical complexity in the documentation workflow, as this logic often does not match the clinical intervention as perceived by the clinician.…”

“…As the ambitions and prospects for data use increase, so does the demand for more and increasingly accurate and complete clinical-administrative datasets. This development persistently puts healthcare professionals under pressure to deliver quality documentation [2,3] and challenges the organization of those tasks. This is manifestly emphasized by the issue of clinician burnout, which has been extensively reported for the past decade.…”

Reconciling healthism and techno‐solutionism: An observational study of a digital mental health trial

“…In Nelly Oudshoorn’s study of patient self‐diagnosis through telemonitoring, for example, individual motivation and responsibility is emphasised in formal accounts, following the logic of healthism, and the care work of home nurses is thus made invisible (2008). The concept of invisible labour in digital health has also been applied to the production, completion, validation and sorting of data sets in data‐intensive clinical settings (Bossen et al., 2019; Green et al., 2022).…”

Putting data to work for precision medicine

“… 10 Further, attempts to repurpose clinical data for administrative or research purposes may drain resources from clinical care, where additional data needs consume clinical resources. 11 The C-MOT relies on EHR data that is routinely entered and does not require any additional data entry. However, analyses of preliminary EHR data in C-MOT have exposed data weaknesses and deficiencies in such areas as the systematic documentation of treatment toxicity and ECOG performance status, as well as systematic radiological documentation of the Response Evaluation Criteria In Solid Tumors.…”