Abstract TMP83: Propensity Score Matching to Estimate Supported Outcomes in Intracerebral Hemorrhage Patients with Withdrawal of Life Support
Background: Observational reports suggest that a self-fulfilling prognostic pessimism may lead to withdrawal of life support (WOLS) in patients with more severe intracerebral hemorrhages (ICHs) who might otherwise have acceptable clinical outcomes. Our objective was to estimate what outcomes in a cohort of ICH patients might have been if they did not have WOLS.
Methods: Multivariate modeling was used to create a propensity score (PS) for WOLS in a Seattle single center cohort of ICH patients with hospital discharge modified Rankin Scale (mRS) as the primary outcome. Using nearest neighbor matching, individual ICH patients with WOLS were matched to individual ICH patients without WOLS and baseline variables and outcome were compared.
Results: The cohort comprised 590 ICH patients with mean age of 67 years, 42% women, 76% white and 18% WOLS. Factors used to create the PS for WOLS included age, pre-ICH mRS, GCS, ICH volume, intraventricular hemorrhage, pre-existing hypertension, diabetes and atrial fibrillation, first temperature and intubation. Matches were possible for 78 WOLS/non-WOLS pairs. Groups were well matched on all PS factors, mean age (67 years), GCS (6.4), ICH volume (59cc) and % intubation (59). Discharge mRS in the both groups varied from 3-6 and was for the WOLS group 1.3%, 2.6%, 6.4% and 90% respectively; in the matched non-WOLS group discharge mRS was 6.4%, 32%, 24% and 37% respectively (difference p<0.0005). Using mRS <= 4 at discharge as an acceptable outcome, % in WOLS/non-WOLS were 4%/38% (difference p<0.0005).
Conclusions: The proportion of ICH patients with WOLS that might have had an acceptable outcome without WOLS was ~38%. This discharge “acceptable outcome” of mRS <= 4 assumes eventual recovery to better function in many. The 34% absolute difference in WOLS/non-WOLS outcomes suggests that for every 3 fewer ICH patients with WOLS, one more patient might have an acceptable clinical outcome. This proportion would vary depending on an individual patient’s interpretation of what constitutes an acceptable outcome.
- © 2012 by American Heart Association, Inc.