Association of early withdrawal of life-sustaining therapy for perceived neurological prognosis with mortality after cardiac arrest

Abstract

Background: Withdrawing life-sustaining therapy because of perceived poor neurological prognosis (WLST-N) is a common cause of hospital death after out-of-hospital cardiac arrest (OHCA). Although current guidelines recommend against WLST-N before 72h (WLST-N<72), this practice is common and may increase mortality. We sought to quantify these effects.

Methods: In a secondary analysis of a multicenter OHCA trial, we evaluated survival to hospital discharge and survival with favorable functional status (modified Rankin Score ≤3) in adults alive >1h after hospital admission. Propensity score modeling the probability of exposure to WLST-N<72 based on pre-exposure covariates was used to match unexposed subjects with those exposed to WLST-N<72. We determined the probability of survival and functionally favorable survival in the unexposed matched cohort, fit adjusted logistic regression models to predict outcomes in this group, and then used these models to predict outcomes in the exposed cohort. Combining these findings with current epidemiologic statistics we estimated mortality nationally that is associated with WLST-N<72.

Results: Of 16,875 OHCA subjects, 4265 (25%) met inclusion criteria. WLST-N<72 occurred in one-third of subjects who died in-hospital. Adjusted analyses predicted that exposed subjects would have 26% survival and 16% functionally favorable survival if WLST-N<72 did not occur. Extrapolated nationally, WLST-N<72 may be associated with mortality in approximately 2300 Americans each year of whom nearly 1500 (64%) might have had functional recovery.

Conclusions: After OHCA, death following WLST-N<72 may be common and is potentially avoidable. Reducing WLST-N<72 has national public health implications and may afford an opportunity to decrease mortality after OHCA.

Keywords: Cardiac arrest; Neurological prognosis; Prognostication; Resuscitation; Withdrawal of care.

Figure 1

Daily totals of subjects who died after out-of-hospital cardiac arrest, stratified by cause of death. Dashed line indicates the 72-hour threshold before which withdrawal of life-sustaining therapy for anticipated neurological prognosis was considered to be “early”.

Figure 2

A) The proportion of cases exposed to WLST-N versus rates of functionally favorable survival within each hospital. Exposure to WLST-N was negatively associated with functionally favorable survival (coefficient −0.23; P = 0.01) B) The proportion of cases exposed to WLST-N<72 versus rates of functionally favorable survival within each hospital. Exposure to WLST-N<72 was negatively associated with survival (coefficient −0.32; P < 0.01)

Figure 3

The association of between each unexposed subject’s propensity for exposure to WLST-N<72 based on baseline clinical characteristics and in-hospital mortality. The 95% confidence band of the spline curve at the highest propensity scores does not include a 1.0 probability of in-hospital mortality, indicating excess mortality associated with exposure after adjustment for baseline clinical characteristics.