End-of-life care in the ICU: Treatments provided when life support was or was not withdrawn

Study objective: To compare and contrast use of technology, pharmacology, and physician variability in end-of-life care of ICU patients dying with or without active life support. Design: Retrospective cohort study. Setting: Two medical-surgical tertiary-care ICUs in a Canadian regional referral teaching hospital. Participants: One hundred seventy-four patients who died between July 1, 1996, and June 30, 1997. Intervention: Data abstraction from medical records. Results: Patients in whom life support was withheld or withdrawn (138 of 174, 79%) were older (65 ± 16 years vs 55 ± 18 years; p < 0.05 [mean ± SD]). Once the decision to withdraw life support was made, death occurred in 4.3 h (2.1 to 6.5 h; mean [95% confidence interval]). Patients who had active life support treatment until death received more support measures including inotropic agents (36 of 36 vs 21 of 138; p < 0.05), dialysis (4 of 36 vs 2 of 138; p < 0.05), and mechanical ventilation at the time of death (36 of 36 vs 81 of 138; p < 0.05). Physician differences (> 10-fold) were detected for prescribed doses of morphine and sedative agents whether or not life support was withheld or withdrawn. The median cumulative dose of morphine prescribed during the final 12 h was larger (fivefold) in patients undergoing withdrawal of life support. No documented discussion of life support withdrawal was noted in one case. In the remaining patients, the 10 staff physicians were documented to be involved in 77% (range, 54 to 94%) of the end-of-life discussions. Conclusions: Differences were evident in technologic and pharmacologic support and in physician prescribing habits in patients for whom life support was or was not withheld or withdrawn. Substantial variability was noted in physician documentation of physician-family interactions surrounding the withdrawal of life support.