Purpose: To identify and describe provider and organizational norms contributing to hospital-level variation in end-of-life treatment intensity.
Methods: We conducted a mixed-methods case study of two U.S. academic medical centers (AMCs) in the same state and health care system to hold regulatory/financial factors constant. At each AMC, we conducted 4 weeks of shadow observation of rounds, family meetings, and clinical care in the medical ICU and completed semi-structured interviews with patient, relative, provider, and administrator informants. We qualitatively analyzed field-notes and interview transcripts using constant comparison.
Results: At the low-intensity AMC we observed treatment of 80 patients in a 16-bed mixed medical-surgical ICU and interviewed 24 informants. At the high-intensity AMC we observed treatment of 81 patients in a 24-bed medical ICU and interviewed 30 informants. The case mix at the high-intensity AMC included a greater number of chronically ill elders. We observed relative parsimony of diagnostic testing, explicitly tied to goals of treatment and structured by teaching rounds at the low-intensity hospital, which contrasted with profligate diagnostic testing without consideration for how the data might be used to inform decision making, implemented during work rounds at the high-intensity hospital. Providers at the low-intensity AMC focused on their decision-making role and the usefulness of institutionalized transfer policies. They evaluated life-sustaining treatments (LST) based on their effectiveness in achieving outcomes early in the ICU course. In contrast, providers at the high-intensity AMC externalized the locus of control to patients, relatives, referring providers, and specialists who they believed expected “aggressive” care. There was more open-ended LST use at the high-intensity AMC, rationalization of continued treatment on the basis of sunk costs, and discussion of care goals when clinicians felt there were no additional LSTs to offer. Palliative care, a mature service at the low-intensity AMC, was rarely consulted; instead providers perceived palliative care as core intensivist competency and followed nurse-initiated policies and protocols for LST withdrawal. Palliative care was a rarely-consulted new service at the high-intensity AMC; instead, providers consulted ethics about concerns regarding futility.
Conclusion: Variations in the use of explicit treatment goals, clinician self-efficacy for making LST decisions, and the maturity of hospital-based resources and policies may contribute to the observed variation in end-of-life treatment intensity observed between these two AMCs.
Candidate for the Lee B. Lusted Student Prize Competition