Objectives: From a cohort of patients with community-acquired pneumonia (CAP) who required admission to hospital, to describe the subset of patients having a do not resuscitate (DNR) order and to compare them with those who did not have such an order.
Design: Retrospective subset analysis of data from the pneumonia patient outcomes research team study.
Setting: Three hospitals in the United States and one in Canada.
Participants: Hospitalized patients aged 18 and older with CAP.
Measurements: Sociodemographic features, severity of illness, antibiotic therapy, length of stay, mortality, admission to special care units, and mortality attributable to pneumonia.
Results: The 199 (14.9) of 1,339 inpatients with CAP who had a DNR order written within 24 hours of admission and an additional 96 (7.2) patients who had such an order written later were compared with the 1,044 who never had a DNR order. The 199 patients with an initial DNR and 96 later DNR were older (median age 81 and 78 vs 65 years, respectively; P< .001), more likely to be white (92.5 and 90.6 vs 84.8; P = .007), and more likely to have come from a nursing home or chronic care facility (53.8 and 31.3 vs 4.5; P< .001). The two DNR groups received more antibiotics for a longer time than the never DNR patients. The DNR patients had longer lengths of stay than the never DNR patients (medians 9 and 12 vs 7 days). There were 89 in-hospital deaths among the 1,339 patients, but only 11 of these were among patients who did not have a DNR order during the first 30 days (sensitivity, specificity, and positive and negative predictive values of a DNR order for in-hospital mortality were 87.6, 82.6, 26.4, and 98.9, respectively). The 90-day mortality rates were 43.2 for the initial DNR group, 61.5 in the later DNR group, and 4.7 for the never DNR group (P< .001). Pneumonia-attributable mortality accounted for most of the in-hospital deaths but did not differ by DNR status. Only 31.7 of the initial DNR patients and 24.0 of the later DNR patients were discharged home, versus 82.6 of the other patients (P< .001). In a multivariate analysis, the following were predictive of initial DNR: age, nursing home care, active cancer, dementia, neuromuscular disorders, altered mental status, low systolic blood pressure, tachypnea, abnormal hematocrit, abnormal blood urea nitrogen, and absence of alcohol or intravenous drug abuse. In similar analyses of DNR at any time, additional predictors included aspiration, low white blood count, chronic pulmonary disease, cerebrovascular disease, and congestive heart failure.
Conclusion: Most in-hospital pneumonia deaths occur in patients who have a DNR order. DNR orders written within 24 hours of admission primarily reflect comorbid status, whereas DNR orders written later during hospitalization reflect the futility of care plus comorbidity.