2 

What Is the Best Way to Assess and Classify Head-Injured Patients?

Nelson M. Oyesiku

This type of question does not lend itself to the standard class I, II, or III method of reviewing the literature. Although such methodology is appropriate for evaluating many kinds of studies, investigations of grading schemes do not readily lend themselves to that type of analysis. For the sake of consistency, the studies discussed in this chapter can be considered to report class III data, but the limitations of using such a classification scheme for this topic should be remembered.

BRIEF ANSWER

Background

Traumatic brain injury (TBI) is a major public health problem, affecting more than 500,000 people in the United States each year. Its annual incidence is estimated to be 200 per 100,000 population. The neurologic condition of TBI patients on initial presentation is a key determinant for determining triage, therapy, and prognosis. Methods of clinically assessing the neurologic condition of TBI patients should be uniform, straightforward, and reliable. Advantages of effective assessment tools include reliability of documentation, ease of data access, consistency of evaluation methods, and accuracy of prognostication.

Various assessment systems have been developed for documenting a patient’s status after TBI (Table 2-1). These include the GCS,¹ the Reaction Level Scale,^2,3 the Innsbruck Coma Scale,⁴ the Edinburgh Coma Scale,^5–7 the Comprehensive Level of Consciousness Scale,⁸ the Maryland Coma Scale,⁹ the Glasgow-Liege Scale,^10,11 and the Head Injury Watch Sheet.¹² Some assessment tools undergo periodic refinement through serial iterations and remain in widespread use, whereas others have not become broadly disseminated.

One type of assessment scale is disability-specific. Scales of this type focus on neurologic deficits. They are generally based on the motor and sensory examination, with assignment of numerical values or letter grades. The second type of scale focuses on functional outcome, such as the patient’s ability to transfer, ambulate, or participate in activities of daily living. In general, the first type of scale is used for the acute assessment of patients with TBI, whereas both types are important for the assessment and description of patients who are past the acute stage of their injuries.

The most widely used and most extensively and rigorously evaluated tool for the acute classification and assessment of TBI patients is the GCS.1 In its current form, it is based on eye opening (score of 1 to 4), verbal responsiveness (1 to 5), and motor responsiveness (1 to 6). The scale was originally designed as three separate subscales. Subsequently, the sum of the three scores was used as a measure of a patient’s overall responsiveness. Common usage classifies patients with a GCS score of 3 to 8 as having severe injuries, those with a score of 9 to 12 as moderate injuries, and those with a score of 13 to 15 as mild injuries. The probability of a poor outcome increases with a decreasing GCS. It must be remembered, however, that different patients with the same total GCS score can have significantly different probabilities of death due to differences in the components making up that score.¹³

Observations in the early stages after injury define the depth and duration of coma. In combination with clinical features such as a patient’s age and degree of brainstem function, these observations have been used to predict outcome.¹⁴ However, to ensure reliability, the GCS should be measured in a standardized way and must be obtained through interaction with the patient. The assessment should be made only after pulmonary and hemodynamic resuscitation and only after pharmacologic sedation or paralytic agents have been metabolized.¹⁵

Studies of interobserver reliability of the GCS have documented overall consistency of assessment among medical, nursing, and paramedical personnel, but the distinction between “abnormal” and “normal” flexion movements has been reported to be a major source of disagreement.¹⁶ Braakman et al¹⁷ found a high inter-observer agreement between experienced and nonexperienced observers in assessment of the motor response. In a study of intensive care unit nurses with varying degrees of experience in the use of the GCS, Ingersoll and Leyden¹⁸ found that the rate of interobserver agreement in determination of the summed GCS score in TBI patients was 100% in straightforward cases, but this rate decreased when more complex patients (e.g., those with a tracheostomy) were assessed. Rowley and Fielding^19,20 found mean interobserver agreement rates of 88.5 to 98%. Although higher agreement rates were seen among trained and experienced practitioners, interobserver reliability coefficients were good within each group tested (groups were assigned based on experience). These results suggest that accuracy (or inaccuracy) of GCS determinations is more of a problem than consistency among different observers. The GCS is also useful and reliable in comparing patients from different countries.²¹

Jagger et al²² found that the motor response score alone was a slightly better predictor of outcome than the GCS sum score. Prasad²³ assessed the utility of the GCS by reviewing published articles describing and using the scale. The GCS was found to have good sensibility and reliability (intraclass correlation coefficient of 0.8 to 1 for trained users). It also has a well-established cross-sectional construct validity. Its predictive validity in traumatic coma, when combined with age and brainstem reflexes, is good (sensitivity of 79 to 97%; specificity of 84 to 97%). This review concluded that, although the scale is an established discriminative instrument, its validity as a predictive and evaluative instrument has not yet been studied adequately.

OTHER SCALES

The scale that later came to be known among Japanese researchers as the Edinburgh Coma Scale was described in the Japanese literature in 1973. The initial version (E1 CS) was modified in 1983 to become the E2 CS.^5–7 It has not gained widespread acceptance.

The Comprehensive Level of Consciousness Scale was proposed primarily as a research tool.⁸ It consists of eight subscales: posture (scored 0–4), eye position at rest (0–6), spontaneous eye opening (0–4), general motor functioning (0–6), abnormal ocular movements (0–6), pupillary light reflexes (0–7), general responsiveness (0–8), and best communicative effort (0–7). This scale is said to be internally more consistent than the GCS. Although its interrater and test-retest reliability are similar to those of the GCS, it is not suitable for routine bedside use because of its detail.

The Head Injury Watch Sheet was described in 1978. It consists of assessment of consciousness and mental status (scored 0–12), motor response to pain (0–12), and pupillary reactivity (0–3 for each eye) for a maximum score of 30.¹² It has not seen widespread use.

The Reaction Level Scale (RLS85) is an eight-part scale for assessment of patient responsiveness. The RLS85 is easily learned and is based on the same concepts as the GCS, but the separate responses are directly weighted together in one scale that has eight categories. The RLS85 has good interobserver agreement^2,3; in fact, one study found that the RLS85 had better interobserver agreement than the GCS sum score.²

The Innsbruck Coma Scale was introduced in 1981.4 The scale scores eight items. A scale of 0 to 3 is used for reaction to acoustic stimuli, reaction to pain, body posture, eye opening, pupil size, pupil response to light, and position and movement of eyeballs; oral automatisms are scored on a 0 to 2 scale. This scale allows for highly accurate prediction of death in patients with scores of 0 or 1, even during the initial posttraumatic examination.

The Maryland Coma Scale was described in 1981.⁹ Its components include eye opening; orientation; pupil, corneal, and caloric reflexes and grimace; type of stimulus required; verbal response; leg motor response; and arm motor response. It contains the three variables of the GCS, excludes responses that cannot be evaluated, and provides more information concerning the status of brainstem reflexes and motor lateralization.

In 1982, the Glasgow-Liege Scale, which combines the GCS with the quantified analysis of five brainstem reflexes, was developed.^10,11 Combining these two parameters in a single scale improves the precision of prognosis, especially for those TBI patients with initial and complete loss of consciousness (LOC).¹⁰

Gibson and Stephenson²⁴ devised the Leeds prognostic scale to identify severely injured patients whose death could be predicted with certainty within 12 hours of admission. They used several variables: age, unreactive pupils, intracranial pressure, systolic blood pressure, GCS score, presence of extracranial injuries, and presence of high-density lesions on computed tomography (CT) scan. They weighted each of these to reflect their influence on mortality. Scores ranged from 0 (best) to 24 (worst). None of the patients with a score higher than 11 survived. Based on these results and on calculated probabilities of survival, they proposed withdrawal of active treatment when a patient’s score exceeded 13. However, when Feldman et al²⁵ tested the reliability of the Leeds scale in patients with TBI, they found that the scale failed to predict mortality with 100% accuracy. Their findings suggested that the Leeds prediction model is not infallible and that it should be applied cautiously when making decisions about the early termination of care in severely head-injured patients.

The Clinical Neurologic Assessment Tool is a 21-item instrument assessing response to verbal and tactile stimulation, ability to follow commands, muscle tone, body position, movement, chewing, and yawning in TBI patients.²⁶ This scale was developed to detect subtle changes in neurologic status. Its reliability has been shown to be high, and concurrent validity testing with the GCS has indicated a strong positive correlation. In one study, scores correctly classified 95.1% of patient observations into their respective GCS categories, leading the authors to conclude that this instrument is valid, convenient, and easily scored and that it captures subtle changes in the patient with TBI.

The Oxford scale, described in 1980, measures the duration of posttraumatic amnesia (PTA), which is defined as a failure of continuous memory.²⁷ This scale builds on the observations of Russell,²⁸ who proposed duration of LOC as a useful index of the severity of brain injury. The Oxford scale uses simple memory tests to categorize PTA by duration, from less than 10 minutes to more than 4 weeks. An extension of the Oxford scale is the Westmead PTA Scale.²⁹ Described in 1986, it is a more standardized measurement of the duration of PTA following TBI. It has a high degree of interrater reliability, takes about 3 minutes to administer, and requires a minimum of training.³⁰ As a predictor of neuropsychological outcome, duration of PTA as determined by the Westmead PTA Scale was reported to have predictive superiority over admission GCS score and duration of coma as determined by the GCS.³⁰

Multiple classification systems have been devised to sort mild TBI by severity. In 1991, the American Congress of Rehabilitation Medicine defined mild TBI as an injury with at least one of the following: (1) any period of LOC lasting less than 30 minutes, with a GCS score of 13 to 15 after the period of LOC; (2) any loss of memory for events immediately before or after the accident, with duration of PTA of <24 hours; (3) any alteration in mental state at the time of the accident; (4) focal neurologic deficit that may or may not be transient.³¹ In addition, various classification systems have been proposed to evaluate concussion in athletes after TBI. One of the better known is that proposed by Cantu: grade I, or mild, has no LOC and duration of PTA <30 minutes; grade II, or moderate, has LOC <5 minutes or PTA >30 minutes but <24 hours; and grade III, or severe, has LOC >5 minutes or PTA >24 hours.³² A modification of this scheme adds a grade 0 for the athlete with only delayed onset of headache and difficulty concentrating (but without symptoms immediately after injury) and omits LOC from the definitions of grades I and II. Ommaya and Gennarelli³³ described syndromes of cerebral concussion based on experimental and clinical observation, with grades I to III involving no LOC and grades IV to VI associated with LOC. Gennarelli differentiates mild concussion with no LOC from a classic concussion with LOC. Mild concussion is divided into several levels of increasing severity of injury. Classic concussion is defined as an injury characterized by coma of up to 6 hours in duration.³⁴