Abstract
Noninvasive advanced respiratory support (NARS) is widely used in acute respiratory failure, including at the end of life, where its use is ethically and emotionally fraught. This ethnographic study examines how decisions to initiate, sustain or withdraw NARS are negotiated within the institutional and moral complexities of a UK hospital. Drawing on fieldwork including nonparticipant observation, reflective fieldnotes and composite narratives, this paper explores how dying is not simply recognised but discursively produced through clinical interactions. Guided by Foucault's concept of power/knowledge and Orlikowski's theory of technology-in-practice, this study shows how authority circulates unevenly across professional boundaries and how machines become nonhuman actors that structure time, command clinical attention and produce epistemically privileged outputs. These outputs, especially blood gas samples, change the flow of power/knowledge and can override patient experience, shaping decisions around palliation. By tracing how clinical truth is co-produced through medical technology, discursive hierarchies and institutional logics, this paper intervenes in sociological debates on end-of-life care, technological agency and the moral labour of frontline staff. It argues that recognising dying is not merely a clinical judgement but an emergent, distributed achievement, one shaped as much by machines and metrics as by human actors.