Objective: To assess the predictive value of respiratory and haemodynamic variables and opinion of the intensivist for determining how soon death occurs after withdrawal of life-sustaining treatments (WLST).
Design: Multicentre prospective observational study.
Participants and setting: 83 consecutive adult intensive care patients at John Hunter and Calvary Mater Hospitals, Newcastle, New South Wales, for whom a decision was made to withdraw life-sustaining treatment between March 2007 and March 2008.
Main outcome measures: Data were collected before initiation of palliation. Primary outcome was to recognise in a multivariate analysis the parameters associated with a time to death < or = 60 minutes after WLST.
Results: 81 patients underwent WLST: 79 died, and two survived to be discharged from hospital. Thirty-six patients (45%) died within 60 minutes of WLST, and 45 (55%) survived 60 minutes or longer. Mean ICU stay before WLST was 4.8 days (range, 1-85 days). Mean time from WLST to death was 6:31 h (range, 1 minute to 31 days). A modified University of Wisconsin assessment tool showed no statistical association with the time from WLST to death (P = 0.09). The adapted United Network for Organ Sharing tool, systolic blood pressure, APACHE II score, ventilatory dependence, oxygen disruption, Glasgow Coma Scale (GCS) score and staff specialist opinion all showed a statistically significant association with time from WLST to death (P < 0.05).
Conclusions: It is possible to predict the time from WLST to death accurately using a tool that combines GCS, respiratory and haemodynamic parameters and intensivist opinion. These results require validation in a large multicentre study.