End-of-life decisions in emergency patients: prevalence, outcome and physician effect.

Background: End-of-life decisions (EOLD) represent potentially highly consequential decisions often made in acute situations, such as 'do not attempt resuscitation' (DNAR) choices at emergency presentation.

Aim: We investigated DNAR decisions in an emergency department (ED) to assess prevalence, associated patient characteristics, potential medical and economic consequences and estimate contributions of patients and physicians to DNAR decisions.

Design: Single-centre retrospective observation, including ED patients with subsequent hospitalization between 2012 and 2016. Primary outcome was a DNAR decision and associated patient characteristics. Secondary outcomes were mortality, admission to intensive care unit and use of resources.

Methods: Associations between DNAR and patient characteristics were analysed using logistic mixed effects models, results were reported as odds ratios (OR). Median odds ratios (MOR) were used to estimate patient and physician contributions to variability in DNAR.

Results: Patients of 10 458 were attended by 315 physicians. DNAR was the choice in 23.3% of patients. Patients' characteristics highly associated with DNAR were age (OR = 4.0, 95% CI = 3.6-4.3) and non-trauma presentation (OR = 2.3, 95% CI = 1.9-2.9). In-hospital mortality was significantly higher (OR = 5.4, CI = 4.0-7.3), and use of resources was significantly lower (OR = 0.7, CI = 0.6-0.8) in patients choosing DNAR. There was a significant effect on DNAR by both patient (MOR = 1.8) and physician (MOR = 2.0).

Conclusions: DNAR choices are common in emergency patients and closely associated with age and non-trauma presentation. Mortality was significantly higher, and use of resources significantly lower in DNAR patients. Evidence of a physician effect raises questions about the choice autonomy of emergency patients in the process of EOLD.