Withdrawal of Life-Sustaining Therapies in Children With Severe Traumatic Brain Injury

Neuroprognostication in severe traumatic brain injury (sTBI) is challenging and occurs in critical care settings to determine withdrawal of life-sustaining therapies (WLST). However, formal pediatric sTBI neuroprognostication guidelines are lacking, brain death criteria vary, and dilemmas regarding WLST persist, which lead to institutional differences. We studied WLST practice and outcome in pediatric sTBI to provide insight into WLST-associated factors and survivor recovery trajectory ≥1 year post-sTBI. This retrospective, single center observational study included patients <18 years admitted to the pediatric intensive care unit (PICU) of Erasmus MC-Sophia (a tertiary university hospital) between 2012 and 2020 with sTBI defined as a Glasgow Coma Scale (GCS) ≤8 and requiring intracranial pressure (ICP) monitoring. Clinical, neuroimaging, and electroencephalogram data were reviewed. Multi-disciplinary follow-up included the Pediatric Cerebral Performance Category (PCPC) score, educational level, and commonly cited complaints. Seventy-eight children with sTBI were included (median age 10.5 years; interquartile range [IQR] 5.0-14.1; 56% male; 67% traffic-related accidents). Median ICP monitoring was 5 days (IQR 3-8), 19 (24%) underwent decompressive craniectomy. PICU mortality was 21% (16/78): clinical brain death (5/16), WLST due to poor neurological prognosis (WLST_neuro, 11/16). Significant differences (p < 0.001) between survivors and non-survivors: first GCS score, first pupillary reaction and first lactate, Injury Severity Score, pre-hospital cardiopulmonary resuscitation, and Rotterdam CT (computed tomography) score. WLST_neuro decision timing ranged from 0 to 31 days (median 2 days, IQR 0-5). WLST_neuro decision (n = 11) was based on neurologic examination (100%), brain imaging (100%) and refractory intracranial hypertension (5/11; 45%). WLST discussions were multi-disciplinary with 100% agreement. Immediate agreement between medical team and caregivers was 81%. The majority (42/62, 68%) of survivors were poor outcome (PCPC score 3 to 5) at PICU discharge, of which 12 (19%) in a vegetative state. One year post-injury, no patients were in a vegetative state and the median PCPC score had improved to 2 (IQR 2-3). No patients died after PICU discharge. Twenty percent of survivors could not attend school 2 years post-injury. Survivors requiring an adjusted educational level increased to 45% within this timeframe. Chronic complaints were headache, behavioral problems, and sleeping problems. In conclusion, two-thirds of sTBI PICU mortality was secondary to WLST_neuro and occurred early post-injury. Median survivor PCPC score improved from 4 to 2 with no vegetative patients 1 year post-sTBI. Our findings show the WLST decision process was multi-disciplinary and guided by specific clinical features at presentation, clinical course, and (serial) neurological diagnostic modalities, of which the testing combination was determined by case-to-case variation. This stresses the need for international guidelines to provide accurate neuroprognostication within an appropriate timeframe whereby overall survivor outcome data provides valuable context and guidance in the acute phase decision process.