Baseline patient acuity variables are insufficient to explain the observed variability in the withdrawal of life-sustaining therapies in severe traumatic brain injury. Hospitaland physicianbased variability in end-of-life decision-making and prognostication are well known. Turgeon and colleagues attempt to draw similar conclusions in severe traumatic brain injury in Canadian centres. Prognostication based on the physician, rather than on patient factors, is a concern to intensive care unit (ICU) practitioners. However, a number of important methodologic concerns cloud — rather than illuminate — in Turgeon and colleagues’ retrospective study. Mortality ranged from 10.8% to 44.2% across different centres, with most deaths associated with the withdrawal of life-sustaining therapies. The variation in mortality rates, after adjustments for admission covariates (sex, age, pupillary reactivity and Glasgow Coma Scale score), was attributed to differences in withdrawal rates because of hospital and/or physician practices rather than patient acuity. Because baseline covariates did not appear to explain the differences in adjusted mortality rates, we are left to assume that no other clinical variables during ICU admission would have an impact on decision-making. These are strong conclusions in the absence of information regarding the severity and temporal evolution of the brain injury after ICU admission. Turgeon and colleagues provide no data on comparative and serial neuroimaging between patients, neurosurgical and neuroprotective interventions, such as ventricular drainage and decompressive craniectomy, and other prognostic tests (e.g., somatosensory evoked potentials). Most concerning is the lack of information about intracranial hypertension, its measurement, management and the response to medical and surgical interventions. The ability for ICU physicians to neuroprognosticate is not limited to admission variables. Without information about the evolution of brain injury, intracranial pressure and response to interventions, the data provided do not support conclusions about the observed effect of medical practices on mortality rates.
Download Full PDF Version (Non-Commercial Use)