Hospital to Hospital Transfers of Cerebral Hemorrhage: Characteristics of Early Withdrawal of Life-Sustaining Treatment

Background Large intracerebral hemorrhages (ICHs) are associated with significant morbidity and mortality. Patient transfer to higher level centers is common, but care in these centers rarely demonstrably improves morbidity or reduces mortality. Patients may rapidly progress to brain death, but a large number die shortly after transferring because of withdrawal of life-sustaining treatment (WOLST). This outcome may result in poor resource use and unnecessary cost to patients, families, and institutions. We sought to determine clinical and radiographic predictors of early death or WOLST that may alter potential transfer. Methods We performed a retrospective review of patients admitted from outside medical centers to the neurosciences intensive care unit at Saint Marys Mayo Clinic Hospital in Rochester, MN, from January 2014 to December 2019. Patients ≥ 18 years old with a spontaneous ICH were included. Exclusion criteria included trauma, subarachnoid hemorrhage, and subdural hematoma. We identified patients who died or underwent WOLST within 24 h of transfer. Descriptive characteristics of patients and ICH were collected. Data were analyzed with univariable, multivariable, and logistic regression. Predictive modeling was performed. An additional case-matched study was completed to evaluate for characteristics further. Results A total of 317 consecutive patients were identified. Forty-two patients were found with early death or WOLST within 24 h of transfer. Do not resuscitate/do not intubate (DNR/DNI) code status (odds ratio [OR] 5.23, confidence interval [CI] 3.31–8.28), anticoagulation use (OR 2.11, CI 1.09–4.09), and lower level of consciousness at presentation based on Glasgow Coma Score (OR 1.41, CI 1.29–1.54) and Full Outline of Unresponsiveness (FOUR) score (OR 1.34, CI 1.26–1.46) were associated with WOLST. Associated characteristics on the computed tomography scan included midline shift (OR 4.64, CI 2.32–9.29), hydrocephalus (OR 9.30, CI 4.56–18.96), and intraventricular extension (OR 5.27, CI 2.60–10.68). Case matching restricted to midline shift demonstrated similarity between patients with aggressive care and WOLST. DNR/DNI code status, warfarin use, ICH score, and composite FOUR score were the best predictive characteristics (area under the curve 0.942). Conclusions Early death or WOLST after ICH within 24 h of presentation was most associated with DNR/DNI code status, warfarin use, ICH score, and lower level of consciousness at presentation. These characteristics may be used by clinicians to guide conversations prior to transfer to tertiary care centers.