End-of-life decision-making for patients admitted through the emergency department: hospital variability, patient demographics, and changes over time

Background: Early studies suggest that racial, economic, and hospital-based factors influence the do-not-attempt-resuscitation (DNAR) status of admitted patients, although it remains unknown how these factors apply to patients admitted through the emergency department (ED) and whether use is changing over time.

Objectives: The objective was to examine patient and hospital attributes associated with DNAR orders placed within 24 hours of admission through the ED and changes in DNAR use over time.

Methods: This was a population-based, retrospective cross-sectional study of patients 65 years and older admitted to 367 acute care hospitals in California between 2002 and 2010; the subset of patients admitted through the ED formed the primary sample. The primary outcome was placement of a DNAR order within 24 hours of admission. Associations between DNAR order placement and hospital characteristics, patient demographics, and year were tested. Descriptive statistics are reported, and multivariable logistic regression models with generalized estimating equations (GEEs) were used to account for clustering within hospitals.

Results: There were 9,507,921 patients older than 65 years admitted to 367 California hospitals over the 9-year period, of whom 1,029,335 (10.8%) had DNAR orders placed within 24 hours of admission; 83% of DNAR orders were placed for patients admitted through the ED. Among patients over 65 years admitted through the ED (n = 6,396,910), DNAR orders were used less frequently at teaching hospitals (9.5% vs. 13.7%), for-profit hospitals (8.6% vs. 14.6% nonprofit), nonrural hospitals (12.0% vs. 26.2%), and large hospitals (11.1% vs. 15.0% for hospitals in the smallest quartile for bed size; all p < 0.0001). In regression modeling adjusted for clustering and patient demographics, these trends persisted for all hospital types, except teaching hospitals. Decreased DNAR frequency was associated with race (African American odds ratio [OR] = 0.59, 95% confidence interval [CI] = 0.51 to 0.67; Asian OR = 0.70, 95% CI = 0.59 to 0.82; reference = white), ethnicity (Hispanic OR = 0.61, 95% CI = 0.55 to 0.68), sex (male OR = 0.90, 95% CI = 0.88 to 0.92), and MediCal insurance (OR = 0.70, 95% CI = 0.57 to 0.85). Statewide rates of DNAR steadily increased from 12.2% in 2002 to 14.3% in 2010, although this trend occurred primarily among white and Asian patients.

Conclusions: While statewide rates of DNAR use have increased over time among patients admitted through the ED, there is variable penetrance of this practice by hospital types, patient race, and patient ethnicity. These patterns may suggest barriers to end-of-life discussions, differences in hospital case mix, and variation in cultural or institutional beliefs and practices.