Intensive Care (ICU) survivors discharged from ICU to the general ward at night have a higher mortality 1,2. We sought to clarify which factors, including night-shift discharge, influence outcome following ICU discharge in a metropolitan hospital, using a cohort study of critically-ill patients between 1/1/1999-30/4/2003. Patients were excluded from analysis if they (a) died in ICU, (b) were transferred to another hospital, (c) had an ICU length of stay <8 hours, or (d) age <16 years. Logistic regression was used to derive a predictive model based on the following variables: patient demographics, severity of illness following ICU admission (APACHE II mortality-risk, p m), final diagnosis, discharge timing including premature or delayed (>4 hours) ICU discharge, and "limitation of medical treatment" orders. The outcome measures were patient status at hospital discharge and ICU readmission rate. Of the 1870 ICU survivors, 92 (4.9%) died after discharge from ICU. Patients discharged to the ward during the night-shift (2200-0730 hours) had a higher APACHE II score and crude mortality. The difference in APACHE II p m did not reach statistical significance. No significant calendar or seasonal pattern was identified. Logistic regression identified night-shift discharge (RR=1.7; 95% CI 1.03-2.9; P=0.03), limited medical treatment order (RR=5.1; 95% CI 2.2-12) and admission APACHE II p m (RR=3.3; 95% CI 1.3-7.6) as independent predictors of patient outcome following ICU transfer to the ward. Conclusion: At the time of ICU discharge to the ward three factors are predictive of hospital outcome: timing of ICU discharge, limited medical treatment orders and initial illness severity.

There is growing concern that blood transfusion may be associated with adverse outcomes in critically ill patients. Timing of transfusion in relation to intensive care unit (ICU) stay may be important in designing and understanding transfusion studies. The objective of this study was to determine the timing of red blood cell transfusion in relation to admission to an Australian ICU and to describe associations with transfusion requirements. We undertook a retrospective, observational, single-centre cohort study of all patients admitted to the ICU at The Northern Hospital, Melbourne, Australia, between 1 January and 31 December 2008 in order to measure the timing of transfusion in relation to ICU admission and the demographic and outcome data of the cohort. 674 individual hospital admissions were analysed. Overall, 28% (188/674) of patients admitted to ICU received a red cell transfusion during their hospital stay. A total of 55 (28.5%) patients were transfused either before and/or after ICU discharge but never in the ICU. Thirty-five percent (258/741) of red cell units were transfused outside the ICU. The median number of red cell units transfused was three units per patient (interquartile range 1 to 5). There was no difference between transfused and non-transfused groups in either crude mortality or severity-adjusted mortality. In approximately one-third of ICU patients in our study transfusions occurred before admission to, and/or after discharge from, the ICU. This has implications for designing and interpreting transfusion studies in the ICU and requires confirmation in a multi-centre study.