Abstract
This article focuses on emotions, conceptualised as emotional labour, evoked during data practices used to repurpose and enable healthcare data journeys for Finnish public healthcare. Combined approaches from critical data studies and the sociology of emotions were used to contribute to a better understanding of the mundane but often invisible work of the emotions of experts involved in data practices, such as facilitating data journeys and building data technologies. The article is based on a two-and-a-half-year ethnographic study conducted in a Finnish regional public healthcare and social service organisation. The study results were derived from the analysis of 39 interviews and fieldnotes produced by observing 170 h of various meetings, events and work activities performed by experts. The results were organised into three forms of observed experts’ emotional labour related to three phases of healthcare data journeys: (a) caring for data production and preparing data for travel, (b) managing excitement and frustration in data processing for continually building the data management system, and (c) reassuring users in making sense of obtained data analytics. The results contribute to a greater understanding of the emotions and emotional labour generated by healthcare data journeys and in relation to the volatile nature of healthcare data and the collaborative character of data practices. This work advocates for a better recognition of the emotional aspects of data practices and their implications on data-based knowledge and datafication processes in healthcare.
Introduction
In this article, I apply the concept of emotional labour to advance existing critical discussions on the repurposing of healthcare data for the management of public organisations and for building data technologies. The examination of emotions (see, e.g. D’Ignazio and Klein, 2020; Kennedy and Hill, 2018), including their management, expression and production, remains an understudied and unrecognised dimension of building data-driven healthcare. In contrast, more research attention has been paid to affective practices and care in scientific research and data science (e.g. Kerr and Garforth, 2016; Pinel et al., 2020; Puig de la Bellacasa, 2011). The increasingly salient role of managing one's own and others’ emotions has also been recognised in professional life and workplaces (e.g. Choroszewicz, 2020; Grugulis and Vincent, 2009; Hochschild, 1983).
This study was specifically inspired by D’Ignazio and Klein's (2020) call to elevate emotions as a form of knowledge to better understand how data practices are situated in a particular context, place and time. Valuing and capturing emotions as a part of data practices also challenges assumptions about the neutrality and objectivity of data and data technologies (see, e.g. Bowker and Star, 1999; Elish and Boyd, 2018; Gitelman, 2013; Pink et al., 2018).
Theoretically, this article combines approaches from critical data studies and the sociology of emotions to capture emotional aspects of data practices within a public healthcare and social service organisation and a public–private partnership. Here, emotional labour serves as a vehicle to better understand the complexity and uncertainty evoked by healthcare data as they travel from data producers and processors to IT company experts and users. Moreover, the concepts of ‘data journeys’ (Leonelli, 2020; Leonelli and Tempini, 2020) and ‘broken data’ (Pink et al., 2018) serve to capture the volatile nature of healthcare data as they move across the interconnected sites in and from which they are made. The definition of data practices operationalised here is rather broad to capture the emotional aspects of experts’ data practices across the phases of the healthcare data journey (Beaulieu and Leonelli, 2021: 36–37; Leonelli and Tempini, 2020: v–vi), including the practices of data production from patient healthcare data records for internal use within the organisation, data processing for continually building the data management system (DMS) in collaboration with a private IT company and the curation and dissemination of the obtained data analytics. The DMS serves as an example of data analytics products under development within the public–private partnership and their implementation in public healthcare systems with expected benefits for management and improvements in decision-making processes and public services.
This article is based on a two-and-a-half-year ethnographic study of a Knowledge Team (KT) and its work related to building data-driven public healthcare and social services for and within a public regional organisation. The KT was comprised of about 16 experts, including middle managers, ICT planners, data workers, subject experts and IT experts. Additionally, IT experts and managers of a private IT company were interviewed, and observations of collaborative meetings between them and the KT were carried out.
This work makes three contributions. First, theoretically, it proposes a framework to capture the emotional aspects related to the collaborative data practices undertaken at different phases of the healthcare data journey. Second, empirically, it provides an overview of the repertoire of emotions captured in the often taken-for-granted interactive and collaborative processes between data workers, IT experts, clinicians and managers. Finally, methodologically, the framework adds an important dimension to current ethnographic research on data practices and data technologies (e.g. Berg, 1997; Passi and Sengers, 2020; Pollock and Williams, 2008) by demonstrating my use of video-mediated ethnography as well as the video-mediated observation and interpretation of emotions.
I first outline the theoretical framework employed for this study. I then briefly discuss the research site, methods and materials. Subsequently, I present the results, which are organised into three forms of experts’ emotional labour related to the three observed phases of the healthcare data journey. Finally, I discuss the contributions and implications of this study for further research.
Datafication of public healthcare, emotions and emotional labour
Public healthcare has been undergoing numerous changes to comply with the expectations and opportunities provided by datafication, both in Finland and internationally. The datafication of healthcare (Hoeyer, 2019; Hogle, 2016; Ruckenstein and Schüll, 2017) has rested upon intensive data work to make healthcare data available for multiple uses and new data technologies (see, e.g. Grön, 2021; Halford et al., 2009). In Finland, like in other Nordic countries, the personal identity number – assigned to all citizens at birth or at immigration – has become a key technology that has enabled data interoperability among public authorities, private businesses and their data repositories (Alastalo and Helén, 2022). Its widespread use has also stoked equally widespread enthusiasm over public databases in Finland as a national resource for economic growth (Tarkkala et al., 2019; Tupasela et al., 2020). In Finland, the building of data-intensive healthcare has been fuelled by regional and governmental authorities and organisations, think tank organisations and private companies through numerous strategy papers and roadmap proposals (Helén, 2019). One prominent pilot study involved the organisation of healthcare and social data retrieved from public service providers’ databases and national registers into so-called Social and Healthcare Information Packets (SHIPs) that are comparable (see, e.g. Helén, 2019). These SHIPs provided a new rationale for building data analytics-based managerial tools for management and decision-making in healthcare and social service organisations in Finland.
As a result of this widespread enthusiasm over repurposing and reusing healthcare data in Finland, it is particularly relevant to examine labour-intensive data practices to understand the social construction and interrelations between classifications, databases, people and technologies. Some key issues here relate to the importance of reflexive practices in unmasking the assumptions and motivations inherent in data practices within specific local contexts (Bowker and Star, 1999; D’Ignazio and Klein, 2020; Gitelman, 2013). Both data infrastructures and data themselves are contingent and relational; they are interrelated through the continual practices of their enactment, standardisation, transformation and expansion within particular contexts (Bowker and Star, 1999).
The framework of data journeys allows for capturing the continuous process of the social construction of data across phases and sites, from data gathering and recording to processing and cleaning to sense-making and visualisation (Beaulieu and Leonelli, 2021; Leonelli, 2014). Data and their prospective usability are also impacted by different types of encounters and practices throughout their journeys (Beaulieu and Leonelli, 2021: 36–37; Leonelli and Tempini, 2020: vi). As data move, they are subject to new requirements and transformations to facilitate the production of information that will satisfy epistemological needs different from those for which the data were originally produced. It is also during these journeys that data are decontextualised, and their limitations and fragile nature become visible (Pink et al., 2018). The concept of broken data facilitates capturing potential difficulties – related to, for example, original data attachments and belongings to patients and clinical work (see, e.g. Berg, 1997; Pinel and Svendsen, 2021) – in transforming and enabling healthcare data journeys through local data infrastructures into data analytics products, such as the DMS, for the production of data analytics-derived information for knowledge-based management and decision-making in public organisations.
Some scholars (Bonde et al., 2019; Neff et al., 2017) have highlighted often hidden and marginalised aspects of data practices that are based on communication and collaboration within diverse teams of experts to produce, repurpose and make sense of data. This can be especially relevant for healthcare data due to their volatile and varied nature. The repurposing of healthcare data relies on dynamic relations between technical and clinical competencies as well as flexible organisational structures that facilitate interdisciplinary collaboration in data practices (Bonde et al., 2019). Thus, collaborative work with healthcare data performed in each phase of the journey can inspire as well as require different forms of emotional labour to allow communication and interaction between experts.
Via these theoretical insights, this research departs from the assumption that the collaborative endeavour of repurposing healthcare data is enmeshed in the emotions of different experts who permit the movement of healthcare data and the usability of data analytics-derived information. To date, a few scholars have addressed the issue of affects and care in data science and research (see, e.g. D’Ignazio and Klein, 2020; Kerr and Garforth, 2016; Puig de la Bellacasa, 2011; Taylor, 2020). Furthermore, in their research on non-experts’ emotional engagements with data and data visualisations, Kennedy and Hill (2018) demonstrated that emotions constitute a vital component of relating to and making sense of data. Their study significantly broadened current understandings of data and datafication processes by accounting for – what they call – ‘the feeling of numbers’ as an integral component of building trust in numbers in the contemporary era of big data and quantification. This study built on and extended this body of research to capture emotions in data practices in public healthcare.
In sociological theory, because of their social character, emotions have been theorised as emotional labour to highlight the amount of work involved in managing one's feelings, or those of someone else, in response to societal demands or job-specific demands (Hochschild, 1983, 2002). Hochschild's (1983, 2002) approach emphasised that emotions are embedded in social settings to satisfy the need to invoke or suppress inner emotions in relation to particular situations and to produce desirable emotional responses from others. Emotions are relational; that is, emotions are shaped by relations and interactions with others, with objects and with the environment (Ahmed, 2004: 4). They are understood as social and cultural expressions of feelings (Koivuniemi, 2010: 9; Probyn, 2005: 11), such as empathy, shame, sorrow, frustration, anger, (dis)satisfaction and excitement. Emotions comprise both verbal and non-verbal expressions and are often performed regardless of one's own particular feelings.
This sociological perspective focuses on emotional labour as a job feature in the service economy – one that is based on interaction and communication. These communicative acts are rarely free and spontaneous but are instead guarded by social norms that people seek to accommodate in their behaviour through emotion management. Emotional labour thus covers both conscious and unconscious self-control practices aimed at presenting one's emotions in ways that are socially expected and situationally appropriate (Hochschild, 2002: 9). Emotions are managed and produced for a wide range of reasons, such as professional or organisational norms and regulations, social rules related to specific relationships and situations and personal history.
In the service economy, the majority of jobs rely on communication and interaction processes, which require and include an extensive amount of emotional expression, management and production (Hochschild, 1983). This emotional labour, often invisible to others, facilitates communication and therefore contributes to the production of material goods (Hardt, 1999: 94), such as classifications, datasets, data visualisations or technologies.
In this study, I addressed the research question of what emotions are evoked as healthcare data move from the KT members regarded as data producers and processors through the IT company experts and the DMS to different groups of users by examining the often taken-for-granted socio-cultural expression and production of emotions elicited during experts’ collaborative practice with healthcare data. In doing so, I approached the data practices and building of data technologies as a situated practice (Haraway, 1988) – one that is also collaborative and originates at a particular place and time in discrete social contexts, including in a public–private partnership, one affected by the service economy, and in a Finnish regional healthcare and social service organisation. This situated practice emerges as an interactive process permeated by numerous promises, expectations and goals that might be difficult or impossible to meet (Grön, 2021; Tupasela et al., 2020), therefore evoking emotions.
Ethnographic study: Research site, methods and materials
Fieldwork was conducted during 2019–2021 to examine the data-driven practices and technologies of a regional public healthcare and social service organisation. 1 Research access to the organisation was provided by the KT, which played a central role in assisting the organisation in work related to data production and data processing for information production and in developing and adopting new technologies for knowledge-based management and decision-making in this public organisation. During the two-and-a-half-year-long fieldwork, the KT was divided into two teams 2 that were situated at different physical locations and under different middle managers. The KT comprised about 16 experts, 3 including middle managers, ICT planners, data workers (also subject experts in, e.g. SHIPs) and IT experts. The IT experts’ responsibilities were associated with technical expertise in digital infrastructure. The ICT planners’ and data workers’ responsibilities were particularly broad and ranged from monitoring and training in data-recording practices across the organisation, assisting practitioners in technology adoption and appropriation, organising the training of practitioners, reporting to national standardised institutions and data work. Some KT members also represented the organisation in the communication and co-development of technologies with private and in-house IT companies.
Building the DMS was one of the KT's prominent tasks, which commenced a few years prior to the beginning of the fieldwork of this study. The DMS utilises healthcare and social data retrieved from patients’ records, the public organisation's administrative data and national registers, which are organised into SHIPs. In practice, the building of the DMS was organised into two types of collaborative meetings held online: meetings focused on discussing planned developments to data analytics and maintenance meetings focused on resolving service tickets related to data outputs. These meetings were organised and chaired by the expert team at the private IT company, which included five to seven experts who worked in R&D, project management, IT and sales direction. The DMS serves here as an example of data technologies that are particularly relevant to building data-driven healthcare and social systems in Finland and beyond, for which technologies provide the promise of better-quality and cost-efficient public services. The studied DMS was collaboratively produced via a public–private partnership to serve the public organisation's managerial needs for data analytics-derived information for knowledge-based management and decision-making. The public organisation invested its experts’ time and expertise, including laborious data work, in the co-development of the DMS, which was a relatively innovative and expensive commercial data analytics product. 4
The study began by observing events involving the KT, such as team meetings and the abovementioned collaborative meetings to build the DMS, as well as interviewing KT members, former and current organisational leaders and managers, academic actors, a few technology users at the public organisation and experts and leaders of private IT companies that were collaborating with the public organisation. In total, 39 interviews 5 with 36 interviewees were conducted. The observations 6 lasted until June 2021 and covered 170 h of observations of KT members and IT company experts at a range of events and meetings. The observed events provided an exceptionally wide and rich perspective on the data work and digital infrastructure of this public organisation. This extensive fieldwork, including both interviews and insightful observations, also allowed for recognising and interpreting the emotional aspects of experts’ speech without facial expressions. Before mid-March 2020, when the COVID-19 pandemic prompted the first lockdown in Finland, I met, observed and interviewed the KT members personally at numerous face-to-face events that they had either organised or attended. During the pandemic, all the observed events were held via videoconferencing without cameras. The online character of the meetings and events provided me with easier access as an invisible outsider, but it also hindered the observation of bodily experienced affects (Martin, 2013). Thus, my extensive video-mediated observations were analysed in the context of research material collected through face-to-face interviews and observations prior to the pandemic.
The collected research material provided a broad overview of the healthcare data practices and journeys within both the public organisation and the public–private partnership; however, my research material and ethnographic practice differed from those employed in previous research on data practices in laboratories (i.e. Kerr and Garforth, 2016; Pinel et al., 2020; Puig de la Bellacasa, 2011). As the data production by the KT members often included the processing of somewhat personal information, I had no access to these first-hand practices. Therefore, my research material covered the verbal accounts of data practices expressed during the interviews with KT data workers and other KT members as well as IT company experts in addition to observations of meetings and events and the activities that occurred during these events.
Consequently, in my analysis, I focused on emotions related to data practices that enabled the repurposing of healthcare data and the data journeys from data production and processing by the KT data workers for the organisational management and healthcare staff as well as for the collaborative building of the DMS. My interest in studying the emotional aspects of data work emerged during fieldwork due to my previous research and was thus not a view from nowhere (Haraway, 1988). I focused specifically on emotions as socially manifested in speech acts and interactions with others, in objects and within the broader organisational and public–private partnership (Ahmed, 2004, 2010; Koivuniemi, 2010). I consider speech acts – and their absence when they are expected – to be carriers of emotions. The captured emotions were present in experts’ speech in the interviews and in the observed events during which they described their work tasks, including the amount of time and attention devoted to and the necessary skills for working with data and building the DMS. During the interviews, some interviewees expressed and commented on their emotions, particularly those related to their work tasks.
The research data were analysed thematically. I began the analysis by reading through all the interviews and fieldnote observations and identifying instances in which experts’ emotions specifically related to healthcare data were evoked. I then reread these research data and grouped the emotions expressed by different experts, specifically organisational management members, including middle managers, other KT members (i.e. data workers, ICT planners and IT experts) and IT company experts. I then examined the empirical material in light of theoretical approaches from critical data studies concerning the concepts of broken data and data journeys. This stage of my analysis resulted in the identification of three forms of emotional labour related to the data practices in each of the three phases of the healthcare data journey, which are discussed in detail in the next section.
Results: Forms of emotional labour
The results are structured based on three phases of the healthcare data journey. The first focused on data production from patient healthcare data records, the organisational data repository and other pre-existing data for internal use within the public organisation. The second phase focused specifically on data processing for continually building the DMS in collaboration with the private IT company. The third phase focused on curation and making sense of obtained data analytics across the public organisation.
Caring for data production and preparing data for travel
My fieldwork experience with the public organisation was permeated by my observations of the KT's extensive manual work with data and local expertise in data-recording practices, which were central to the first phase of repurposing healthcare data. While the data-recording practices of the healthcare staff and the KT's data work were invisible to me, I witnessed numerous discussions about the tremendous effort, time, attention and skills the KT had put into data production so that the data could be used for purposes other than those for which they were originally collected (Isin and Ruppert, 2019).
Specifically, the KT data workers expressed a strong sense of attachment and commitment to their everyday work with the data as well as their responsibility for the produced data, which echoed Carol Gilligan's (1982) feminist framework of the ethics of care (see also Taylor, 2020). This was evident in their verbal interactions, statements and activities through which they demonstrated their care for data as an everyday labour and their feeling of obligation to assist the organisational management and the healthcare staff in their decision-making processes through the production and delivery of the requested data. This corresponds with the results of a study by Pinel et al. (2020) on a research laboratory whose staff engaged in caring practices aimed at enhancing the value and flourishing of data. During the interviews and observed events, some of the KT data workers voiced their frustration and feeling of helplessness when some information requests from healthcare staff could not be completed or when they were urged by the managers to prioritise tasks other than practical work with data. I also observed that the data workers were left alone to make burdensome trade-offs between how much work to devote to particular data work tasks, how many resources to invest in these tasks and which tasks to either drop or complete faster and perhaps less comprehensively than they would have preferred.
The data workers’ care for data was also expressed in their relational orientation towards data production as a collective endeavour rather than an individualised work task. This was evident in the KT data workers’ volunteering for additional data work despite having an already busy work schedule or offering help each other in completing extremely intensive and tight-schedule requests for data production. They also credited each other's work and welcomed assistance when needed. Additionally, they emphasised the importance of the healthcare staff's mundane work put into data-recording practices.
At times, the data workers appeared to be frustrated with the lack of understanding within the public organisation regarding their mundane work with the data. For them, a seemingly simple request for particular information from healthcare staff or the organisational management could take anywhere from a couple of hours to a day or more to be fulfilled. This data production process also involved uncertainty with regard to identifying relevant pieces of data that were scattered across different databases, reports and information systems as well as their combination and their interpretation. Some requests were short notice and were accompanied by impatient follow-up requests, which only added to feelings of frustration – and sometimes also incompetency – among the data workers. One data worker expressed frustration with being criticised for the pace of their work or produced data as follows: The main job is that when we are asked for information that is not available in any of the ready-made reports, we need to investigate where this piece of data was recorded, how it was recorded and from where it can be retrieved. And then we need to go through with an IT expert to plan a new report in terms of how it will be done, stored and retrieved by users. (…) These requests for information are usually not easy to fulfil, as they require the aggregation of a great many different things that are not easily combined, and because of this, we are criticised for taking so long [to deliver requested information]. [Interview with KT data worker]
The data workers found it difficult to explain to organisational management or other healthcare staff that some of their information requests could not be immediately and directly addressed as they required creative data work (Pink et al., 2018: 3), which information systems were not able to perform on their own. In this case, they appeared to draw on what Kennedy and Hill (2018) called ‘the feeling of numbers’ to track, identify and assemble data from various databases within continuously changing contexts and data-recording practices. Thus, it can be argued that these data were continually incomplete, breakable and repairable (Pink et al., 2018), with some missing data or incorrectly recorded data, some recorded or corrected later or some never recorded at all. This contingency of healthcare data was related to the changing configurations of people, organisations, national legislation, processes and technologies that made data production for re-use and movements within the data infrastructures of the organisation challenging (see, e.g. Leonelli, 2020); however, this messy everyday reality of data practices (see, e.g. Elish and Boyd, 2018; Leonelli, 2014) was rather invisible to organisational management and healthcare staff.
The lack of recognition of data workers’ mundane work with data seemed to be, at least partly, linked to enthusiasm on the part of organisational management regarding the potential of their healthcare data. At times, their way of talking or inquiring about the data seemed to evoke the assumption that the healthcare data were sitting in a repository, fully formed and ready for harvesting; however, in practice, the repurposing of the healthcare data was far from a clear-cut technical task, despite what some members of organisational management might have assumed. The excerpt below (from my fieldnotes) illustrates a typical situation from the observed meetings during which the data workers attempted to make their data practices understandable for organisational management: During the team meeting, the data worker was explaining at length, with a fatigued voice, that it took three weeks to work with data, but there is still lots to do before the data can be loaded into the DMS. They hoped for the data to be ready before midsummer [about one month from that point]. The data worker continued expounding the reasons for this additional work related to preparing data to be fed into the DMS.
When the data worker finished, the manager asked with surprise and curiosity, ‘So, what are those challenges? Isn’t it so that the information system can account for these changes on its own?’
The data worker heaved a light sigh of frustration but maybe also slight satisfaction that the manager showed interest in their data work. She repeated in other words what she had said before and concluded by emphasising the assertion that ‘information systems are not able to do those changes on their own’. [Fieldnotes, spring 2020]
The process of data production also required collaboration between the data workers and the IT experts of the KT to channel the flow of data both within and beyond the organisational infrastructure. The data workers highlighted that data work is a highly collaborative endeavour done together with the KT's IT experts to find, for example, the origins of problems with the data, to fulfil information requests or to prepare new data reports. While the IT experts facilitated the technical and logistical movement of data across sites as well as data visualisations, the data workers were needed to define, track and assess the necessary pieces of data. During the interview, the data worker explained the importance of their requisite local expertise of databases and data-recording practices to the data production as follows: IT experts do not have knowledge of what and where data are recorded. There might be some missing data, for example. IT experts need to be told to ‘also take this piece of data, make a circle there, and put this here’. Then, we might wonder together how to illustrate some pieces of information. Then, I might offer advice about which database some particular pieces of data can be found (…) I know very well into which table what types of data are recorded. (…) We [data workers] have such profound knowledge of users’ practices in terms of data recording. [Interview with KT data worker]
Managing excitement and frustration in data processing for continually building the DMS
The emotional labour of managing excitement and frustration related to data processing was also evident during the online meetings between the KT members and the IT company experts, which focused on healthcare data processing for continually building the DMS to produce data analytics-derived information for the knowledge-based management of the public organisation. The meetings were highly interpretative in nature (see also Neff et al., 2017), whereby the experts discussed and agreed on the possibilities and limitations of the healthcare data and the DMS. While the IT company experts provided expertise in the technical aspects related to healthcare data processing in the DMS, the KT members provided their expertise and experience in working with volatile and complex healthcare data in the context of clinical and data-recording practices. The KT also provided consultations to the IT company experts during or outside of the online meetings regarding healthcare terminology so that new data analytics could be properly developed.
In the interviews, some KT members expressed frustration about the misleading use of healthcare terminology, such as ‘treatment path’ or ‘treatment period’, in the DMS in contrast to the original meanings of these terms in their own databases, which were used by healthcare staff. During the interviews, they voiced concern that such terminological misuse was confusing not just to practitioners but also to them and that it also created more manual data work for them, as they had to adjust their data to the changed meanings of these terms as applied in the DMS. Ultimately, the KT members, including the medical expert, carried the responsibility of adequately using healthcare data when they were recontextualised (Pinel and Svendsen, 2021) but nonetheless sufficiently reflected the practices linked to their origins.
During the online meetings, the repertoire of emotions expressed by the KT members, including the medical expert, 7 in their speech ranged from excitement, satisfaction and gratitude to frustration, irritation and disappointment. The KT members occasionally provided general gratitude for and appraisals of the progress of developments, such as ‘good job’ and ‘developments are going in a better direction’; however, these comments were rarely followed by more specific feedback on the developments. In the course of time spent at the organisation, I learned that this initial satisfaction was, amongst other reasons, the result of the lack of time taken by the KT members to test new features more thoroughly. Instead, they would sometimes politely inquire about other issues of greater importance to them in their everyday data work, such as about a schedule of data loadings or a release of particular reports.
Unless there were enormous problems with the obtained data analytics in the DMS, the KT members, besides the medical expert and key experts, hardly ever made any comments even though they were asked for feedback by the IT company experts. During the observed events, I noticed numerous instances in which the IT company experts presented a new feature of the DMS and attempted to initiate a discussion about it by asking for feedback and questions. As the majority of developments were originally requested by the medical expert, he usually spoke first. His feedback was predominantly general and encouraging and only critical when developments did not work or progress within the expected timeframe. The criticisms or feedback of other KT members, if given at all, were voiced cautiously, e.g. ‘We ask you not to blame you but to learn about [the feature under discussion]’. It seemed as though the KT members did not want to express criticism about the developments or felt uncomfortable doing so.
I became particularly intrigued by the performative character of these numerous moments of silence (for more, see Dupret, 2019), which appeared to be related to the power relations within the public organisation and the public–private partnership (for more, see Choroszewicz and Alastalo, 2021). The interviews with the KT members and the KT's own team meetings provided me with insights into the instances of silence in the collaborative meetings, which appeared to be linked to the following situations: (a) the KT members had no time to acquaint themselves with the issue and thus had nothing to say about it; (b) they were mostly present as listeners and were thus engaged in other work tasks during the meetings; (c) they wished to express critical comments or questions but felt pressured not to do so; (d) they were impeded by the hectic pace and technical language of technical demonstrations by the IT company experts; (e) they had no access to DMS testing; or (f) they were overwhelmed by and frustrated with the laborious collaboration with few benefits from the DMS for their everyday work.
These instances of silence appeared to allow a smooth transition through the agenda meetings, but they also often left the IT company experts with unclear and rather limited and homogeneous feedback. Furthermore, they prevented a multiplicity of views, voices and expertise from being recognised in the development of data analytics, and thus they facilitated the achievement of consensus about data processing in the DMS as only the goals and interests of organisational management, and the expertise of a narrow group of key experts were included. Consequently, the medical expert in particular became almost solely responsible for the provision of ideas and feedback for the IT company experts.
Occasionally, the IT company experts had to dispel the anger of the medical expert, who sometimes fairly directly expressed his frustration and disappointment with developments of data analytics or development delays, which appeared to result from a disconnect between what he expected to feel, that is, satisfaction with well-functioning technical features, and the feeling of disappointment with the actual state of developments. On other occasions, the medical expert politely noted that some developments were based on faulty assumptions or were underdeveloped. After all, he had shared his visions of the necessary data analytics with the IT company experts on many occasions, and in exchange, he had expected them to, as he put it, ‘find ways to implement them technically’. Still, as his visions were complex and ambitious, they often led to disappointment with unfulfilled aspirations.
In these situations, like in the example of call centre workers (D’Ignazio and Klein, 2020: 193), the IT company experts stayed calm and did not escalate the emotional reactions of the medical expert. Instead, they waited until his anger had subsided to suggest certain actions to move forward and to resolve the issue under debate. Still, the IT company experts refrained from making any immediate promises on developments or schedules during the online meetings. Instead, they remained calm and replied vaguely: ‘I need to think about how this can be implemented’ or ‘I can’t say anything for sure before I check it’.
The IT company experts also occasionally advertised new technological solutions for the KT that they did not request (assumedly) free of charge, which could be regarded as a way to stimulate customers’ excitement and gratitude. Additionally, as the DMS was under continuous development, the IT company experts were asked and were expected to propose new solutions to the public organisation's informational needs. The IT experts were developing new data analytics on a limited pool of test data for demonstration to the KT. If they were approved by the KT, they were developed further and possibly integrated into the DMS. There was always uncertainty in terms of whether the proposed data analytics would make sense from the perspectives of the KT and the medical expert and whether they would work on the proper datasets. When the IT company experts were demonstrating their ideas, they asked the KT ‘not to mind the faulty numerical results but rather focus on the functioning of the technological solution’. While the KT remained predominantly silent, the medical expert often expressed excitement about the proposed developments: The IT company expert went through the research and proposed a new solution to the issue discussed at previous meetings. He demonstrated the solution during a collaborative meeting and asked for feedback and questions. The medical expert spoke first and praised the solutions. He stated, ‘Great job. I believe this is not only of use for us but at the national level, and this issue has been sought after so long and no one has yet proposed a solution to it. So, you are the pioneers’.
The IT company expert asked other KT members present at the meeting for feedback, but they remained silent. When no additional feedback was offered, he thanked everyone for what feedback had been received and promised further work on the implementation of the proposed solution. [Fieldnotes, autumn 2020]
While the IT company experts never directly voiced any signs of frustration concerning, for example, the medical expert's numerous grand, futuristic visions for the necessary data analytics in the collaborative meetings, they alluded to them during the interviews. They talked about many ‘vague’ and ‘fluctuating’ developments, perspectives and agreements on development priorities. Still, the IT company experts, as service providers, tried to demonstrate their understanding of the lack of clarity concerning the prioritisation and formulation of development needs by the KT as representatives of the customer organisation in the collaboration. By doing so, they appealed to aspects of emotional labour (Hochschild, 1983) in their work that were associated with active listening and attending to the customers’ needs without challenging them to strive to sustain the ongoing public–private partnership. The IT company expert emphasised his own and his company's efforts – as the service provider – put into provision for the best possible service to this public organisation as follows: It can be a bit confusing when one development is a clear priority for a specific person, and we think about it a lot and work on it, but then this person is absent, and the others have different opinions about it. (…) But, of course, it is understandable that a customer's job is to demand a lot and expect their demands to be met right away. And so our task is to manage these sometimes contradictory needs. (…) I try to think all the time about how we can assist them [in this particular organisation] better (…) We value their knowledge and expertise so much (…) We collaborate much more closely with them, and we listen much more sensitively to all the needs that arise. [Interview with IT company expert]
Reassuring users in making sense of data analytics within the public organisation
As the DMS relied on structured and carefully defined data, the data loaded into the DMS required continuous work both immediately and long after the monthly data loadings, as even assumedly well-defined data appeared to behave unexpectedly after their journey into the DMS. While much work on automating the data flow from the local infrastructure of the public organisation to the DMS was accomplished throughout my two-and-a-half years of fieldwork at the organisation, the KT data workers’ labour-intensive activities related to data curation and validation after the monthly data loadings remained manual. Over time, data validation requests became scheduled in advance, which provided the data workers with more stability. Still, three to five data workers were tasked for a couple of days almost every month with data validation.
Furthermore, the DMS provided data analytics, but it did not provide a guide about how to make sense of and use them within the public organisation. My fieldwork illustrated that making sense of data analytics is a collective process (see also Neff et al., 2017) and was thus achieved via collaboration between the KT data workers, the IT company experts and the healthcare staff. First, the KT members actively requested first-hand training from the IT company experts when new data analytics were introduced to the DMS. This training was given to the KT and some central users from organisational management. During the training sessions, the KT members and other users were rather quiet, even when the IT company experts attempted to initiate a discussion.
Second, some KT data workers were given the responsibility of training healthcare staff in the use of the DMS, which appeared to be emotionally draining for them. As a part of this training, the data workers needed to also perform the invisible work of encouraging, comforting and reassuring practitioners of their capacity to master the DMS even though they themselves had an insufficient understanding of it. They emphasised their proficiency in data-recording practices, the local data infrastructure and some knowledge of the data repository created to support data journeys to the DMS, but they had no knowledge about what happened to their data thereafter. Thus, some of them expressed anxiety about being placed in situations in which they were unable to answer practitioners’ questions regarding the exact origins and processing of data or the potential applications of the resulting data analytics. Specifically, they experienced stress and embarrassment about being unable to address the practitioners’ doubts concerning their ‘feeling of numbers’ (Kennedy and Hill, 2018) produced by the DMS during one-on-one or small-group training in DMS use as requested by practitioners. While the data workers needed to encourage practitioners to use the DMS and offer future assistance in its use, they confessed their limited knowledge of the DMS, which was embarrassing for them. The KT data worker disclosed her feelings of embarrassment and frustration associated with the training of practitioners in the use of the DMS to obtain and make sense of data analytics as follows: Practitioners need someone to come and explain to them where these data outputs come from and how they should be used. They need to be trained about it. I told them that I do not know that; I do not have this type of expertise. (…) At the moment, training is completely missing here in terms of how they should use different types of information. These data analytics are nice to watch, but there is no information about where data come from. Practitioners immediately ask for it (…) For example, a chief nurse asked me this question, and I did not know what to say because I do not know where it comes from. It is embarrassing. It is embarrassing that we do not have a guidebook for how it could be checked. [Interview with KT data worker]
Furthermore, making sense of the data analytics obtained from the DMS also rested on the collaborative efforts between the KT data workers and the healthcare staff. The KT data workers demonstrated some particular data analytics that would be of special interest to specific practitioners who requested a training session. In exchange, the practitioners provided feedback on the data analytics and were expected to propose further data analytics that would be of particular benefit to their work in the units; however, there was always uncertainty in terms of whether the proposed data analytics would ever be developed. During the interview, the practitioner was aware of limited resources invested in tailoring the DMS to the informational needs of the healthcare staff, and she also expressed her enthusiasm about making sense of data analytics and their further development as follows: [The DMS] provides possibilities to monitor patient-oriented activity metrics, so, well… it provides this positive surprise that, ‘Wow, great! We can get this and this from [the DMS]’. But my opinion is that these [data analytics] reports were made for specific users. (…) I developed a better understanding of data analytics after I got an opportunity to finger and twiddle with [the DMS] and its reports on my own and together with the KT members. (…) Our unit’s goal is that we would have our own reports, including our own metrics, but I have no idea how long it would take to get them. (…) It feels like the resources put into the implementation of [the DMS] are always underestimated. [Interview with a head nurse of the public organisation]
Discussion and conclusion
I have examined emotions as emotional labour related to healthcare data journeys to advance our comprehension of understudied emotional aspects of data practices in different phases of these journeys. Each phase involved collaborative and interdisciplinary work as well as decision-making by different experts within not only the public organisation but also the public–private partnership in which many data technologies are contemporaneously built. The collected empirical material enabled me to situate data practices and data journeys within broader organisational and public–private partnership contexts, which are rarely accessible to the public or the research community. This issue is central due to the increasing importance of public data and new data technologies for managing public organisations and for the delivery of public services (Grön, 2021; Halford et al., 2009; Helén, 2019; Hogle, 2016).
The combination of the sociology of emotions and critical data studies provided a theoretical framework for naming and recognising the emotional aspects of data practices. Doing so allowed for elaborating on the concepts of data journeys (Beaulieu and Leonelli, 2021; Leonelli, 2014; Leonelli and Tempini, 2020) and broken data (Pink et al., 2018) in relation to the healthcare data as well as demonstrating how data practices are embedded in the social world in which they exist and that include experts’ emotional labour at different phases of data journeys. This labour is also related to the collaborative and interdisciplinary character of data practices in the healthcare sector. Furthermore, the contributions made by Bonde et al. (2019) and Aula (2019) have been extended by recognising that data practices are situated within not only the same organisation but also more broadly within the power dynamics of the service economy and are thus reflected in public–private partnerships.
The experts’ emotional labour functioned as a vital element of human interaction in these practices. The study results indicated that emotional labour was elicited by unrealistic expectations and simplistic notions of healthcare data as being readily available for harvesting and reuse. The emotional labour perspective facilitated the capture of the work involved in seeking to cope with uncertainty and the messiness inherent in everyday tasks concerning data work and the construction of data technologies, specifically inducing anxiety, stress and frustration. Kennedy and Hill (2018) noted that emotions play a vital role in seeing and understanding data, an observation for which this study provided some evidence, specifically for experts who work with data and data technologies. Emotional labour appeared to be linked to experts’ time- and resource-consuming work with data, which is often accompanied by numerous limitations, failures, expectations and frustration. As the contingency of healthcare data is perpetuated by the variability of medical records due to typically multiple interpretations of the same health conditions (Cabitza et al., 2019), the mundane everyday work of experts who must contend with volatile healthcare data and the emotional labour these data entail will not necessarily be improved by optimising the recording of data. It is thus important to recognise the emotional work of these experts as an integral – and ongoing – dimension of data practices throughout data journeys.
Each phase of the data journey evoked a different form of the emotional labour of the experts. In the first phase, collaboration within the KT in the public organisation was pronounced and emphasised the experts’ care for data in the processes of data production and preparation for travel. This form of emotional labour highlighted the data workers’ strong feeling of responsibility for data production in line with the informational needs of the organisational management and the healthcare staff. It also stressed data workers’ emotional investments in caring relations for healthcare staff's data-recording practices and data (see, e.g. Kerr and Garforth, 2016; Pinel et al., 2020; Puig de la Bellacasa, 2011), which need to be recognised and valorised within public organisations as it enhances the value of the produced data.
In the second phase, the management of excitement and frustration emerged as a vital element of collaborative efforts between the KT and the IT company experts in data processing for continually building the DMS. This form of the experts’ emotional labour facilitated the creation of a ‘correct’ climate that enhanced the collaborative data practices based on communication and the sharing of knowledge among experts from the public organisation and the IT company. The management of excitement and frustration was also enmeshed in the power dynamics of the service economy (Hardt, 1999; Hochschild, 1983) in which the customer organisation experts, especially those in higher organisational positions, were provided greater leeway to draw from diverse emotional expressions. In contrast, the IT company experts, as service providers, appeared to engage in forms of emotional labour that sustained long-lasting relationships with a customer organisation. Instances of silence among the KT members contributed to the reproduction and mapping of current power relations within the public organisation in the development of data analytics that aligned with the needs and goals of organisational management. Instances of silence also prevented the views of organisational management from being challenged or complemented with a stronger view of the healthcare staff's needs concerning data analytics.
In the third phase, reassuring users in making sense of data analytics obtained from the DMS was collaboratively achieved by the KT, the IT company experts and the healthcare staff. The KT data workers needed to perform acts of reassuring practitioners of their capacity to use the data analytics produced by the DMS while managing their own embarrassment related to having an insufficient understanding of the system. This emotional labour emphasised the interdependence between data work competences, clinical practice and technical competence in data-based knowledge production and decision-making.
Finally, the study provided methodological clues regarding the ways in which emotional aspects can be captured by video-mediated ethnography. When access to facial and bodily expressions was unavailable, I shifted my focus towards listening and detailed notetaking on utterances, including their context, tone of voice, words used and instances of silence. For example, I paid special attention to the verbal statements and interactions, order of utterances, tone of voice and ways in which the experts were prompted to speak up. Second, I examined the instances of silence when speech was expected. The emotions behind both the utterances and the instances of silence were interpreted in the context of their affective backgrounds from multi-sited ethnography, including face-to-face observations and interviews. Furthermore, the methodological approach based on relationally engaging with collected material became pronounced in this study not only in understanding the experts’ emotional reactions in video-mediated observations but also in critically examining my own emotional reactions concerning the emotional atmosphere of the setting and surroundings of the observed events and interactions. Video-mediated ethnography served as a multi-sensory act, including my own emotional responses and experiences, that rested heavily on my prior acquaintances with the research subjects, which enabled me to interpret emotions without access to facial and bodily expressions.
Via these insights, this study aligns with current scholarship (D’Ignazio and Klein, 2020; Kennedy and Hill, 2018; Pinel et al., 2020; Puig de la Bellacasa, 2011; Taylor, 2020) that has advocated for the recognition of emotions and feelings-based approaches in datafication processes in healthcare for the purpose of challenging rational and neutral perspectives on data practices and data technologies. The recognition of emotions and emotional labour expands the existing scholarship on the tensions and ambiguities inherent in the secondary use of Finnish healthcare data (Aula, 2019; Grön, 2021) and the development of the health data economy (Tupasela et al., 2020). It also facilitates the recognition of the continuous work and meaning making inherent in datafication and shifts attention to its processes rather than its end-products, such as data analytics or data technologies.
Footnotes
Acknowledgements
I would like to thank the public organisation and its experts and organisational leaders as well as the IT company experts for providing access to the observed meetings and events as well as for their agreement to be interviewed. I thank Dr Marja Alastalo for her collaboration in the collection of research material during the fieldwork. The article has been greatly improved by the suggestions of the anonymous reviewers and the editorial hand of Matthew Zook and Sachil Singh. I appreciate the time and attention they devoted to engaging with this article.
Declaration of conflicting interests
The author declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The author disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This work was supported by the Academy of Finland (Grant No. 317303, 336074).
