Asbjørn Malte Pedersen is the leading author behind the article entitled “Cultivating Data Practices Across Boundaries: How Organizations Become Data-driven”. Here is an interview with him conducted on this occasion.
EUSSET: Asbjørn Malte, your paper “Cultivating Data Practices Across Boundaries: How Organizations Become Data-driven” was awarded the 2024 David B. Martin Best Paper Award during ECSCW 2024 – congratulations! Your paper is based on an empirical, ethnographic study in the field of healthcare business intelligence. Could you provide some additional background information about this paper? What led you to focus on data work and boundary objects in your research?
Asbjørn Malte: The paper is one strand of a larger research project called Making Data Work Visible, whose overarching aim is to reveal and investigate the largely invisible human labor that turns data into practical value in healthcare. Within that program, I conducted ethnographic fieldwork inside a Danish regional Business Intelligence unit. The unit’s mission is to repurpose healthcare data from electronic health records, finance, and logistics into ‘Data Reports’ (self-service analytic tools) and a regional data warehouse that clinicians, medical secretaries, and managers can use.
Initially, I set out to understand what working with data involved in this particular unit. Throughout the process, I realized two things: their data-driven technologies could be understood as boundary objects enabling cooperation across different boundaries between healthcare professionals, data professionals, administration, and managers. This concept became a natural analytical lens for me to understand the work revolving around the different data objects. And 2) beyond data extraction, analytics, and data warehousing, a lot of the unit’s work focused on disseminating these products. I observed BI developers teach nurses how to interrogate a dashboard and promote their BI setup through workshops, annual events, and merchandise, which was all very surprising to me. But this also made me realize that the Data Reports were not only doing a specific kind of socio-technical work as Boundary Objects but also needed work to become useful in practice. In short, I followed the practice, and the practice led me to Boundary Objects and to the complementary notion of collaborative boundary work that the BIU engages in to make those objects actually travel in everyday healthcare.
EUSSET: How would you summarize the main points of the paper to someone new to your work?
Asbjørn Malte: One of the central points is that we cannot assume that strategies, high-quality data, or having user-friendly data products are enough to disseminate data throughout the organization. It requires work beyond the technical, as well as intra-organizational dynamics, where multiple actors must collaborate across boundaries and take on new roles. We label these efforts ‘cultivating data practices across boundaries’ and point to three specific categories of (data) work: 1) ‘Mobilizing interest’ concerns the BI unit’s activities, events, and artifacts to make others interested in using data and demonstrate the potential to cultivate a cultural shift; 2) ‘building local capabilities’ designates their work of providing end-users (healthcare professionals and administrative workers) with capabilities necessary to work with data and apply different data-driven technologies in practice; and 3) ‘propagating data locally’ concerns the consolidation and extension of new collaborative data practices where healthcare professionals take on new responsibilities to use, implement, and disseminate data locally in their departments.
Regarding self-service analytic tools, the intention is often to shift the work from the IT department to the end user. In the healthcare sector, this raises questions about professional boundaries and the competencies required to succeed in practice, as well as questions about resources. Conceptually, the paper contributes by (a) showing how Boundary Objects become objects-in-use, (b) extending the literature on healthcare data work to include data professionals, and (c) providing an empirical counterpoint to the otherwise normative Business Intelligence and Self-Service BI literature.
EUSSET: What would you say were the main challenges to conducting this kind of empirical study? Any lessons learned that you might want to share?
Asbjørn Malte: First and foremost, to gain confidentiality and familiarity with my field and the people I met along the way. Building rapport was, of course, essential, but COVID-19 added another wrinkle: for months, fieldwork moved to Teams-based staff meetings and remote screen-sharing sessions. However, my experience was that both BI staff and healthcare professionals were curious and open to the study. Another challenge was disciplinary translation. To understand a data warehouse conversation, a medical secretary’s DRG vocabulary, and an ICU nurse’s workflow, you must become bilingual in clinical shorthand and SQL logic. My takeaway is that naivety can be constructive in the beginning, but long-term ethnography gives you enough time to acquire the dual literacy you need in the long run.
EUSSET: …and what are the positive experiences you have had in the field?
Asbjørn Malte: Without a doubt, all the passionate people I have met along the way: BI developers celebrating that they identified and corrected a small error in a table; a medical secretary proudly demonstrating her customized reporting tools used for quality assurance in her department; and clinicians critically assessing data visualizations to determine a patient’s options and treatment. Witnessing the impact firsthand is the upside of deep immersion, and those moments where data moves from an abstract data warehouse to a concrete purpose of use are professionally and personally rewarding. They remind you of all the hard labor and passion that goes into healthcare and data every day.
EUSSET: The paper presents an intriguing contribution regarding the role of data in healthcare systems and the work practices surrounding it. This topic is particularly relevant given the recent surge in machine learning, natural language processing, and data science applications. What are your thoughts on the potential opportunities and risks associated with the automatic processing of data in this context?
Asbjørn Malte: There is enormous promise: automated pattern-detection can uncover adverse-event hotspots faster than monthly audits; predictive models can help allocate staff before a surge hits; and natural-language processing can surface clinically relevant details hidden in free-text notes. But automation amplifies whatever data goes in. If registration practices are uneven, algorithms may reinforce local biases or silently exclude clinically meaningful nuance. Equally, the seductive clarity of a probability score can obscure the assumptions and labor that went into its creation. That is why it is so important to keep the hermeneutic, interpretive side of data work visible; teaching end-users to stay ‘critically attentive’ to data provenance, model limits, and ethical implications. Automatic processing should augment, not replace, situated clinical judgement.
EUSSET: Is this a line of research you plan to continue pursuing, or do you intend to shift your focus to something entirely different?
Asbjørn Malte: Although the Making Data Work Visible project has formally ended, I still think data work is an important part of the discussion about the digitalization of the healthcare sector, so for now, I still have a bit of focus on data work in healthcare BI. Moreover, I remain fascinated by how health-sector digitalization keeps redrawing professional boundaries. My next step is to follow this thread into clinical AI-enabled prediction and decision support: How does work change when the “object” is an algorithm rather than a dashboard? What new competencies will front-line clinicians need to make sense of probabilistic outputs? What does it take to successfully implement clinical AI? In other words, I will stay with datafication in healthcare but follow it into its emerging ML and AI articulations.
EUSSET: Is there anything you’d like to share with the community members?
Asbjørn Malte: First, a heartfelt thank you to the ECSCW award committee. Receiving the David B. Martin Best Paper Award as an early-career researcher is humbling and motivating. I also owe much to the anonymous CSCW reviewers whose constructive comments undeniably made the paper – and me – better. Finally, I would like to extend my gratitude to the community for always being so engaging and accommodating.
EUSSET: Thank you so much for sharing your insights with us. Your research and insights are of great value to healthcare practitioners and the EUSSET community. We express our congratulations and wish all the best to you for the future.
The interview was conducted by Mateusz Dolata.