Infrastructure Revisited: An Ethnographic Case Study of how Health Information Infrastructure Shapes and Constrains Technological Innovation.

Background: Star defined infrastructure as something other things "run on"; it consists mainly of "boring things." Building on her classic 1999 paper, and acknowledging contemporary developments in technologies, services, and systems, we developed a new theorization of health information infrastructure with five defining characteristics: (1) a material scaffolding, backgrounded when working and foregrounded upon breakdown; (2) embedded, relational, and emergent; (3) collectively learned, known, and practiced (through technologically-supported cooperative work and organizational routines); (4) patchworked (incrementally built and fixed) and path-dependent (influenced by technical and socio-cultural legacies); and (5) institutionally supported and sustained (eg, embodying standards negotiated and overseen by regulatory and professional bodies).

Objective: Our theoretical objective was, in a health care context, to explore what information infrastructure is and how it shapes, supports, and constrains technological innovation. Our empirical objective was to examine the challenges of implementing and scaling up video consultation services.

Methods: In this naturalistic case study, we collected a total of 450 hours of ethnographic observations, over 100 interviews, and about 100 local and national documents over 54 months. Sensitized by the characteristics of infrastructure, we sought examples of infrastructural challenges that had slowed implementation and scale-up. We arranged data thematically to gain familiarity before undertaking an analysis informed by strong structuration, neo-institutional, and social practice theories, together with elements taken from the actor-network theory.

Results: We documented scale-up challenges at three different sites in our original case study, all of which relate to "boring things": the selection of a platform to support video-mediated consultations, the replacement of desktop computers with virtual desktop infrastructure profiles, and problems with call quality. In a fourth subcase, configuration issues with licensed video-conferencing software limited the spread of the innovation to another UK site. In all four subcases, several features of infrastructure were evident, including: (1) intricacy and lack of dependability of the installed base; (2) interdependencies of technologies, processes, and routines, such that a fix for one problem generated problems elsewhere in the system; (3) the inertia of established routines; (4) the constraining (and, occasionally, enabling) effect of legacy systems; and (5) delays and conflicts relating to clinical quality and safety standards.

Conclusions: Innovators and change agents who wish to introduce new technologies in health services and systems should: (1) attend to materiality (eg, expect bugs and breakdowns, and prioritize basic dependability over advanced functionality); (2) take a systemic and relational view of technologies (versus as an isolated tool or function); (3) remember that technology-supported work is cooperative and embedded in organizational routines, which are further embedded in other routines; (4) innovate incrementally, taking account of technological and socio-cultural legacies; (5) consider standards but also where these standards come from and what priorities and interests they represent; and (6) seek to create leeway for these standards to be adapted to different local conditions.