van Veen E, van der Jagt M, Citerio G, et al. Occurrence and timing of withdrawal of life-sustining measures in traumatic brain injury patients: a CENTER-TBI study. Intensive Care Med. 2021;47(10):1115-1129. doi:10.1007/s00134-021-06484-1
Read the Full Article
Background: In patients with devastating brain injury, the Neurocritical Care Society recommends 72 hours of observation prior to formal decisions on the withdrawal of life sustaining measures (WSLM). Prognostication in patients with severe traumatic brain injury is challenging, and early decisions regarding WLSM can be a self-fulfilling prophecy for poor outcome. The purpose of this study is to explore WLSM patterns in TBI and identify variables that are associated with early WLSM.
Methods: This prospective cohort design involved multiple centers over Europe and Israel from 2014 to 2017. Seven geographic regions were defined. Patients studied were 16 years or older, presented to the hospital within 24 hours after TBI and had an indication for CT brain. All patients included were ICU admissions with the primary end point of mortality. The data was compared between ICU survivors and those who died, and deaths related to WLSM vs. death from other causes. Key characteristics between patients receiving early WLSM vs. later WLSM were analyzed, with early defined as less than 72 hours from injury. Imputed time was used if key times and dates for WLSM were not recorded, and sensitivity analyses were performed with and without imputed data. Using the IMPACT model (International Mission for Prognosis and Analysis of Clinical Trials in TBI), the likelihood of mortality as well as poor outcome were calculated.
Information on 2022 patients was obtained, and 1998 patients with known ICU discharge status were included for analysis. ICU mortality occurred in 267 patients (13%), and 229 deaths occurred following WLSM (86%). Half of these patients (51%) experienced early WLSM. 44 patients had imputed data for time and date of WLSM, and 10 patients had imputed data for time of WLSM. The median age was similar between groups (61 years in the early WLSM group, and 60 years in the later group). The median time to WLSM was 24 hours (IQR 12-37) in the early group and 214 hours in the later WLSM group.
Lack of pupillary reactivity significantly predicted early WLSM. Unilateral lack of pupillary reactivity was associated with an OR 4.60 (CI 1.74-12.17) for early WLSM and bilateral lack of pupillary reactivity with an OR 6.61 (CI 3.56-12.27). An analysis of patients without missing data on timing of WLSM (i.e., without imputed data) showed an increased OR of early WLSM with one unreactive pupil, however this was not statistically significant. Patients in the early WLSM group had a higher rate of predicted mortality and poor outcome based on IMPACT scoring (60% vs. 29%). With regards to geographical region, there was not an association between region and timing of WLSM.
This study addresses the important issue of the timing of WLSM in the setting of TBI and highlights potential factors that may influence early vs. late WLSM. Given that guidelines for devastating brain injury propose a 72-hour observation period prior to WLSM, it is reasonable to define less than 72 hours as early. This study found that 51% of patients (a total of 117) who received WLSM had early WLSM. In 44 patients, the time and date data were imputed, but even with consideration of this limitation, the rate of early WLSM remains striking. Furthermore, when data were missing regarding timing of WLSM, ICU discharge date and time were used instead, which are reasonable surrogate variables.
The median time for early WLSM was 24 hours (IQR 12-37), far shorter than the 72-hour observation period that is recommended. Correlations were made between lack of pupillary reactivity and early WLSM. There is a strong implication here that pupillary reactivity influenced discussions regarding prognosis. Unfortunately, this trial does not specifically address details regarding documentation and content of goals of care discussions from providers. There is no information regarding role of specialists such as neurologists and neurointensivists in these patient cohorts. Expert opinion in discussions of neuroprognostication would likely influence timing of WLSM, and further investigation is warranted.
Shannon Hextrum, MD
Assistant Professor of Neurosurgery
Tulane University, School of Medicine