Objectives: To evaluate the outcomes of hospitalized patients with do-not-resuscitate (DNR) orders and to identify variables that may elucidate the high mortality of patients with DNR orders.
Methods: Among a nationally representative sample of Medicare patients hospitalized with congestive heart failure, acute myocardial infarction, pneumonia, cerebrovascular accident, or hip fracture, we retrospectively studied in-hospital and 180-day mortality and hospital lengths of stay for patients without DNR orders, with early (day 1 or 2) DNR orders, and with late (day 3 or later) DNR orders, before and after adjustment for sickness at hospital admission and patient and hospital characteristics.
Results: In-hospital mortality for patients with DNR orders exceeded that for patients without DNR orders before adjustment (59% vs 8%, P < .001), and after accounting for differences in sickness at admission and patient and hospital characteristics (40% vs 9%, P < .001). Sicker patients were assigned earlier DNR orders. Yet, patients with early DNR orders had a lower adjusted in-hospital mortality (31% vs 49%, P < .001) and shorter hospital stay (10 vs 18 days, P < .001) than did patients with late DNR orders.
Conclusions: Hospitalized older patients with DNR orders have a much higher mortality than predicted by admission demographic and clinical characteristics. The differential association of early and late DNR orders with mortality indicates that DNR orders represent a heterogeneous group of interventions that may be a marker of unmeasured sickness and a determinant of quality of care. A better understanding of what the DNR order represents and its effect on patient care is needed to ensure optimal use.