Monday, October 27, 2008

'Stuttered' treatment withdrawal in the ICU

American Journal of Respiratory and Critical Care Medicine has a paper looking at the duration of life-support withdrawal and family satisfaction.  The data come from a multicenter trial in Washington State which looked at various aspects of end-of-life care in the ICU.  This analysis looks at ~584 family member surveys of patients who died after some withdrawal of life-sustaining therapies (WLST):  patients died in an ICU or within 24 hours of transfer out.  (There were 2000 ICU deaths in the study period, and only 820 family surveys com
pleted, and then a couple hundred exclusions based on incomplete data and deaths occuring without WLST.)  LSTs here were defined as laboratory testing (?), mechanical ventilation, renal replacement therapy, tube feeds, vasopressors, and IV hydration.  (Why lab testing and not antibiotics I'm not sure.)  

Mean patient age was 72 years, 90% white, mean hospital length of stay was 6 days, 82% were ventilated at some point, and ~50% received vasopressors.  The paper has tons of data in it, both simply descriptive about what happened to who when, as well as some analysis looking at patient characteristics which predicted all-at-once or 'stuttered' WLST and family satisfaction.

The key findings were that for about half the patients LSTs were withdrawn sequentially over time (greater than a day) - the other half had all LSTs stopped the same day.  Across the cohort, ventilation tended to be withdrawn last (a little over 1 day prior to death), with other therapies withdrawn sooner.  Renal replacement therapy was stopped more than two days, on average, before death (although only 10% of the subjects ever received RRT).  The longer a patient was in the ICU, as well as the more LSTs they were receiving, the more stuttered the WLST was (youth, non-cancer diagnoses particuarly trauma, and more family members involved in decision making also predicted a more stuttered WLST process).  

Family satisfaction was interesting.  The shorter a patient's overall ICU stay was (less then 3 days), stuttered WLST was associated with worse family satisfaction.  For longer ICU stays, stuttered WLST was associated with better family satisfaction.  Intubation in the last week of life was not associated with family satisfaction (either way), however, for intubated patients, being extubated before death was strongly associated with family satisfaction.  The use of non-invasive postitive pressure ventilation (ie 'bipap') was negatively (although not statistically significantly) associated with satisfaction.  

There are many caveats with interpreting all this, not least being selection bias given the low response rate to the family survey.  That said, these seem to be supporting data that stuttered WLST - something I imagine that those of you who see patients in the ICU see done all the time - is common and not necessarily a bad thing for many families.  The authors state their original hypothesis was that stuttered WLST would be associated with worse family satisfaction and generate some hypotheses as to why they were wrong.  The one most consistent with my own observations is that the process of limiting and withdrawing potentially LST in ICU patients expected to die is often a days' (or weeks') long process, which evolves with the patients' changing clinical course and response to therapy, and while the vast majority of times families agree with withdrawal of some/many/all LST before a patient dies, many require/benefit from time (at least days) to accomodate a patient's changing condition/terrible prognosis and to make decisions accordingly even if the physicians lead-off with straight-forward recommendations for WLST.  The observation that increasing number of family decision makers 'prolongs' things seems to support this reality....
E. Gerstel, R. A. Engelberg, T. Koepsell, J. R. Curtis (2008). Duration of Withdrawal of Life Support in the Intensive Care Unit and Association with Family Satisfaction American Journal of Respiratory and Critical Care Medicine, 178 (8), 798-804 DOI: 10.1164/rccm.200711-1617OC

Pallimed | Blogger Template adapted from Mash2 by Bloggermint