Abstract 103: Hospital-Level Differences in Use of Do Not Resuscitate Orders in PaWith Pneumonia, Acute Myocardial Infarction, and Congestive Heart Failure in California

    loading  Checking for direct PDF access through Ovid


Background: Hospitals vary in their use of Do Not Resuscitate (DNR) orders. Whether this variability in DNR use is consistent across multiple diagnoses at a hospital or is unique to certain diagnoses is unknown.

Methods: Using the California Office of Statewide Health Planning and Development (OSHPD) Database for the years of 2010 and 2011, we identified all patients 18 years or older who had a DNR order placed within 24 hours of admission. We selected patients with one of three primary diagnoses: pneumonia (PNA), acute myocardial infarction (AMI), and congestive heart failure (CHF). Hospitals with less than 10 patients with these diagnoses were excluded. A Generalized Linear Mixed Model (GLIMMIX) with hospital random effect was used to determine the unexplained variation in DNR across hospitals, after adjusting for patient age, gender, race, and primary diagnosis. Analyses were stratified by primary diagnosis. Based on the GLIMMIX model, we calculated intra-class correlation coefficients (ICCs) in each cohort of primary diagnoses.

Results: 265 hospitals had over 10 patients with one of the primary diagnoses. Risk and reliability-adjusted DNR rates ranged from a little over 0.8% to 54.5% with a mean DNR rate of 10.1% (Figure). The ICC of the full cohort of patients was 0.22; ICCs for PNA AMI, and CHF were 0.19, 0.18, and 0.25, respectively. All ICCs were ~0.20, suggesting that 20% of the unexplained variability in DNR use was attributable to hospital differences after accounting for patient-level factors. For each condition, there was a strong positive pairwise correlation between risk and reliability-adjusted DNR rates for the other 2 diagnoses, also supporting a hospital-level effect.

Conclusion: Approximately 20% of the unexplained variability in risk and reliability-adjusted DNR rates is attributed to differences across treating hospitals. This may reflect variability in hospital culture factors, such as aggressiveness of care, early care limitation, and patient-doctor relations. Failure to account for these differences may unduly impact hospital ranking measures.

Related Topics

    loading  Loading Related Articles