Hospital Culture and Intensity of End-of-Life Care at 3 Academic Medical Centers.

ImportanceThere is substantial institutional variability in the intensity of end-of-life care that is not explained by patient preferences. Hospital culture and institutional structures (eg, policies, practices, protocols, resources) might contribute to potentially nonbeneficial high-intensity life-sustaining treatments near the end of life.ObjectiveTo understand the role of hospital culture in the everyday dynamics of high-intensity end-of-life care.Design, setting, and participantsThis comparative ethnographic study was conducted at 3 academic hospitals in California and Washington that differed in end-of-life care intensity based on measures in the Dartmouth Atlas and included hospital-based clinicians, administrators, and leaders. Data were deductively and inductively analyzed using thematic analysis through an iterative coding process.Main outcome and measureInstitution-specific policies, practices, protocols, and resources and their role in the everyday dynamics of potentially nonbeneficial, high-intensity life-sustaining treatments.ResultsA total of 113 semistructured, in-depth interviews (66 women [58.4%]; 23 [20.4%] Asian, 1 [0.9%] Black, 5 [4.4%] Hispanic, 7 [6.2%] multiracial, and 70 [61.9%] White individuals) were conducted with inpatient-based clinicians and administrators between December 2018 and June 2022. Respondents at all hospitals described default tendencies to provide high-intensity treatments that they believed were universal in US hospitals. They also reported that proactive, concerted efforts among multiple care teams were required to deescalate high-intensity treatments. Efforts to deescalate were vulnerable to being undermined at multiple points during a patient's care trajectory by any individual or entity. Respondents described institution-specific policies, practices, protocols, and resources that engendered broadly held understandings of the importance of deescalating nonbeneficial life-sustaining treatments. Respondents at different h