41 

The first step in answering the question, What is the outcome of patients with mild, moderate, and severe head injury? is consideration of what different people mean by outcome, and in particular by the term good outcome. To neurosurgeons, a good outcome might mean survival or a return of the ability to care for oneself, whereas to the patient’s family it might mean a return to premorbid status. To an insurer it might mean a rapid end to the need for ongoing payments. The original International classification of Impairments, Disabilities, and Handicaps (ICIDH)1 looked at outcome from four perspectives: pathology, impairment, disability, and handicap.

Pathology

Most neurosurgeons view outcome from the perspective of pathology. This approach focuses on clinical findings, such as a subdural hematoma, and their effect on a patient’s outcome. This view of outcome is often what dominates discussions with family and other providers early in a patient’s hospital course. During this time period, a focus of this type is often essential for a patient’s survival. Such an emphasis, however, does not address the main concerns of families and loved ones about return to premorbid status and future lifestyle.

Impairment

The impairment view of outcome looks at the functional consequences of the underlying pathology, such as weakness, personality change, or aphasia. To patients and their families, impairments are the primary focus of concern.

Ideally, evidence-based studies of outcome would explore the impact of mild, moderate, and severe head injury on impairment, disability, and handicap. However, such studies are very time-consuming and expensive. For this reason, most outcomes research focuses on much simpler end points, such as death, or on simple scoring systems such as the Glasgow Outcome Scale2 (GOS) or the Disability Rating Scale.³

Prognostication

Prognostic Variables

Many characteristics can be used as prognostic variables to predict outcome from head injury. Patient age, findings on computed tomography (CT) scan, findings on neurologic examination, the presence of fixed and dilated pupils, and the presence of hypotension have all been used to build models for the prediction of outcome from head injury. All of these models, however, are derived from studies of very specifically defined patient populations, from which attempts are often made to formulate generalizable, evidence-based recommendations.⁴

Application of the results of such studies to particular patients is often hampered by the fact that head-injured patients are not a homogeneous group. Choi et al⁵ have demonstrated that predictors are population-specific; different predictors of outcome will have varying significance in different patient populations (class II data). For example, the important predictors of outcome for a 40-year-old are likely to be different from those for a 10-year-old, and patients with fixed and dilated pupils have a different order of importance for predictors of outcome than patients with normal pupils. Before clinicians use any outcome studies in their clinical practices, they need to be sure that they appreciate how their particular patients are both different from and similar to those in the outcomes studies that they are reviewing.

Glasgow Coma Scale

In reality, the question, What is the outcome of patients with mild, moderate and severe head injury? is simply asking what is known about the Glasgow Coma Scale (GCS) as a predictor of outcome from TBI, because the categories of mild, moderate, and severe head injury are usually based on the GCS score.

Without question, the GCS has become an extremely powerful tool for the initial and early assessment of TBI patients. It is easy to learn, widely disseminated, and extensively validated. However, clinicians should be aware that it may have important limitations. Many patients who present to emergency rooms cannot have their GCS score calculated. In one study, a full GCS score was not obtainable in 44% of patients, and 28% could not have their motor score assessed (class III data).⁶ The principal reasons for such incompleteness are intubation in the field, which prevents assessment of the verbal score, and the early use of paralytics and sedatives in the emergency department. In addition, techniques for assigning the GCS may vary widely from institution to institution, potentially resulting in difficulties transferring results from one institution to another.

Furthermore, it appears that the predictive value of the GCS changes based on when in the patient’s treatment course it is assessed. Using the initial GCS score as a basis for prediction, one study found that 68.6% of head-injured patients predicted to have a good outcome and 76.5% of those predicted to have a poor outcome actually had those outcomes at 1 year (class II data).⁷ The predictive accuracy went up if a later GCS was used (obtained a mean of 7 days after injury), with 80.6% of those predicted to have a good outcome actually achieving such an outcome at 1 year. The ability to predict poor outcomes remained essentially unchanged at 78.6%.

Even when the best efforts are made to apply evidence-based data to prediction of outcome from TBI, the results can be discouraging. Current methods often overestimate the likelihood of poor outcomes and underestimate the probability of good outcomes. Kaufmann et al8 investigated the accuracy of an experienced neurosurgeon in using overall clinical assessment and initial CT scan results to predict outcome on the first day of admission (class III data). The predictive accuracy was only 56%. Because the “false pessimism rate” may be higher if predictions are based on assessments that are obtained very early in a patient’s treatment, providers should be leery of making definitive predictions immediately after a patient is admitted to the hospital.

Thus, even in the best possible circumstances, outcome prediction is quite difficult in the head-injured population, especially early in a patient’s course. These observations are disturbing in light of an observation made by Murray et al⁹ in 1993 that physicians, when provided with models of outcome prediction after head injury, alter their care in response to these models (class III data). Use of therapies considered beneficial to the head-injured patient increased for those patients predicted to have a good outcome and decreased for those patients predicted to have a poor outcome.

Literature Review

Severe Head Injury

In 1977 Jennett et al¹⁰ reported a 50% mortality rate from head injury in three countries (class II data). By 1987, the Traumatic Coma Data Bank (TCDB) in the United States was reporting a 36% mortality rate (class II data).¹¹ Although some would disagree, most people accept this difference as evidence of interval improvement in the treatment of TBI. These studies enrolled patients with severe TBI; that is, a GCS score of 3 to 8. Some investigators group patients differently. Narayan et al¹² examined the relationship between admission GCS score and outcome and found that a GCS score of 3 to 5 was associated with a poor outcome 77% of the time. A GCS score of 6 to 8 was associated with a poor outcome 26% of the time (class II data).

In terms of GOS categories, outcome of TCDB patients was as follows: 36% mortality, 5% vegetative state, 15% severe disability, 15 to 20% moderate disability, and 25% good recovery (class II data).¹¹ For specific GCS scores, mortality rates in the TCDB were 78.4% for an initial GCS of 3, 55.9% for a GCS of 4, and 40.2% for a GCS of 5 (class II data).¹¹ Patients with GCS scores of 3, 4, and 5 had good outcomes 4.1%, 6.3%, and 12.2% of the time, respectively (class II data).¹¹ These results are consistent with those of other studies, although further evidence suggests that particular types of pathology can affect outcomes. For example, one study showed worse outcomes at each GCS level when an acute subdural hematoma was present (class III data).¹³