6 

How Can I Decide That a Head-Injured Patient Can’t Be Salvaged?

Donald W. Marion

BRIEF ANSWER

Background

The ability to predict within hours after injury that a patient with severe TBI will die or remain in a persistent vegetative state despite our best treatment efforts is desirable for several reasons. Modern intensive care can sustain vital functions in most patients indefinitely, and persistent vegetative survival may last for months or years, causing prolonged grieving of family members. Medical care of TBI patients has been estimated to cost as much as $84 million per year, with the majority of the money spent on those with the most severe injuries, who are unlikely to attain a meaningful recovery. Early and accurate prediction of outcomes could allow for redistribution of health care dollars to those who are most likely to benefit.

During the last 20 to 30 years, more than 60 published studies have examined the sensitivity and specificity of various early prognostic indicators. Specific components of the physical and radiologic examination, demographic factors, electrophysiologic findings, and chemical and metabolic profiles of the brain as inferred from jugular venous blood or cerebrospinal fluid (CSF) analysis have been evaluated as potential prognostic indicators. Most studies conclude that a combination of these measures provides the most accurate prediction of outcome and that reevaluation 24 hours after injury significantly improves predictive sensitivity.

Clinical Determination of Brain Death

The diagnosis of brain death implies that no brain function is detectable and, most importantly, that this condition is irreversible. The most widely accepted criteria for this diagnosis include (1) establishing the cause; (2) ruling out or correcting medical conditions that can depress the neurologic examination (e.g., severe electrolyte, acid-base, or endocrine abnormalities; hypothermia; hypotension; drug intoxication; poisoning; ingestion of systemic neuromuscular paralytic agents); (3) computed tomography (CT) findings that confirm severe brain injury or swelling, and (4) a careful neurologic examination that confirms the absence of cortical and brainstem function.1 The Glasgow Coma Scale (GCS) score should be 3, and the pupils nonreactive to light. No oculocephalic, corneal, or cough reflex is present, and cold caloric stimulation does not cause tonic deviation of the eyes. An apnea test may be done to confirm absence of respiratory effort even when the PaCO₂ reaches 60 mmHg. Confirmatory tests advocated by some, and recommended for children, include the atropine test (no change in heart rate after intravenous infusion of 1 mg of atropine), electroencephalography (EEG) (no activity at levels higher than 2 μV for 30 minutes of recording, with sensitivity set at 2 μV/mm), cerebral angiography (no intracranial blood flow), transcranial Doppler (absence of diastolic flow or reverberating flow), and nuclear imaging (absence of intracerebral uptake of tracer). Some also advocate repeating a test after 6 to 24 hours.

Literature Review: Early Prognostic Indicators

Physical Examination and Age

Patient age and two elements of the physical examination (the motor component of the GCS score and pupillary size and reactivity) are consistently found to independently correlate with the severity of brain injury and therefore with outcome. Choi et al² reviewed the 12-month outcomes of 555 patients with severe TBI who were prospectively included in their TBI database (class II data). Among 23 prognostic indicators, they found that age, GCS motor score, and pupillary reactivity were the most important prognostic indicators and could be used in combination to predict death or vegetative survival with an accuracy rate of 90%. However, 12 of 121 patients predicted to have a poor outcome were neurologically normal or had only moderate disabilities at 12 months. Mamelak et al³ developed a predictive model by using stepwise logistic regression analysis to define the clinical variables most important for predicting 6-month outcome in 672 TBI patients entered prospectively into a database. The authors then applied the model to 108 patients randomly selected from the database to determine its predictive accuracy (class II data). They, too, found that age, best motor score, and pupillary reactivity were the most significant predictors of outcome. These elements were included in their model, which had a sensitivity of 100%; in the 108 patients in whom it was tested, the model correctly identified all survivors and made no false-negative predictions of nonsurvival in patients who actually survived. However, when applied during the first 6 hours after injury, its specificity was only 43%; 57% of patients who were predicted to have a good outcome actually died or were vegetative by 6 months after injury. When applied at 24 hours after injury, the model demonstrated an increase in specificity to 73%. In a later report, Lang et al⁴ reanalyzed 1066 patients from the same database and included hypotension as an independent predictor in their model (class II data). They also concluded that the 24-hour evaluation was more accurate, but the sensitivity of their model (as defined above) was only 89.6%, and the specificity was 76%.