• John D Santamaria, Graeme J Duke, David V Pilcher, D James Cooper, John Moran, Rinaldo Bellomo, and Discharge and Readmission Evaluation (DARE) Study.
• 1 Intensive Care Unit, St. Vincent's Hospital (Melbourne), Fitzroy, Australia.
• Am. J. Respir. Crit. Care Med. 2015 May 1; 191 (9): 1033-9.

RationalePrevious studies suggested an association between after-hours intensive care unit (ICU) discharge and increased hospital mortality. Their retrospective design and lack of correction for patient factors present at the time of discharge make this association problematic.ObjectivesTo determine factors independently associated with mortality after ICU discharge.MethodsThis was a prospective, multicenter, binational observational study involving 40 ICUs in Australia and New Zealand. Participants were consecutive adult patients discharged alive from the ICU between September 2009 and February 2010.Measurements And Main ResultsWe studied 10,211 patients discharged alive from the ICU. Median age was 63 years (interquartile range, 49-74), 6,224 (61%) were male, 5,707 (56%) required mechanical ventilation, and their median Acute Physiology and Chronic Health Evaluation III risk of death was 9% (interquartile range, 3-25%). A total of 8,539 (83.6%) patients were discharged in-hours (06:00-18:00) and 1,672 (16.4%) after-hours (18:00-06:00). Of these, 408 (4.8%) and 124 (7.4%), respectively, subsequently died in hospital (P < 0.001). After risk adjustment for markers of illness severity at time of ICU discharge including limitations of medical therapy (LOMT) orders, the time of discharge was no longer a significant predictor of mortality. The presence of a LOMT order was the strongest predictor of death (odds ratio, 35.4; 95% confidence interval, 27.5-45.6).ConclusionsIn this large, prospective, multicenter, binational observational study, we found that patient status at ICU discharge, particularly the presence of LOMT orders, was the chief predictor of hospital survival. In contrast to previous studies, the timing of discharge did not have an independent association with mortality.

24 keywords...

hide…