Further experience in the management of severe head injury

Full access

✓ A prospective and consecutive series of 225 patients with severe head injury who were managed in a uniform way was analyzed to relate outcome to several clinical variables. Good recovery or moderate disability were achieved by 56% of the patients, 10% remained severely disabled or vegetative, and 34% died. Factors important in predicting a poor outcome included the presence of an intracranial hematoma, increasing age, abnormal motor responses, impaired or absent eye movements or pupil light reflexes, early hypotension, hypoxemia or hypercarbia, and elevation of intracranial pressure over 20 mm Hg despite artificial ventilation. Most of these predictive factors were assessed on admission, but a subset of 158 patients was identified in whom coma was present on admission and was known to have persisted at least until the following day. Although the mortality in this subset (40%) was higher than in the total series, it was lower than in several comparable reported series of patients with severe head injury. Predictive correlations were equally strong in the entire series and in the subset of 158 patients with coma. A plea is made for inclusion in the definition of “severe head injury” of all patients who do not obey commands or utter recognizable words on admission to the hospital after early resuscitation.

Abstract

✓ A prospective and consecutive series of 225 patients with severe head injury who were managed in a uniform way was analyzed to relate outcome to several clinical variables. Good recovery or moderate disability were achieved by 56% of the patients, 10% remained severely disabled or vegetative, and 34% died. Factors important in predicting a poor outcome included the presence of an intracranial hematoma, increasing age, abnormal motor responses, impaired or absent eye movements or pupil light reflexes, early hypotension, hypoxemia or hypercarbia, and elevation of intracranial pressure over 20 mm Hg despite artificial ventilation. Most of these predictive factors were assessed on admission, but a subset of 158 patients was identified in whom coma was present on admission and was known to have persisted at least until the following day. Although the mortality in this subset (40%) was higher than in the total series, it was lower than in several comparable reported series of patients with severe head injury. Predictive correlations were equally strong in the entire series and in the subset of 158 patients with coma. A plea is made for inclusion in the definition of “severe head injury” of all patients who do not obey commands or utter recognizable words on admission to the hospital after early resuscitation.

In 1977, we reported on the outcome of 160 patients treated for severe head injury using a uniform and intensive management protocol.1 There appeared to be a reduction in mortality without an increase in severe morbidity compared with results reported from some other centers.11,17,18,21 A management regimen of immediate diagnosis and surgical decompression of mass lesions, artificial ventilation, and intracranial pressure (ICP) monitoring was strongly advocated.1,15

The relationship between several neurological and physiological variables and subsequent outcome was examined in these 160 patients. Our principal conclusions were: 1) Comatose patients with intracranial mass lesions requiring surgical decompression (40% of the series) were older, neurologically worse, more often had raised ICP, and had a poorer outcome than patients who did not require surgery. 2) Mortality rose with age, mainly from the medical complications of prolonged coma. 3) Bilateral absence of the pupil light reflex and impaired or absent oculocephalic responses predicted a poor outcome in all patients; abnormal or absent motor responses were significantly correlated with a poor outcome only in patients requiring surgical decompression. 4) A few patients with the adverse combination of absent pupil light response, impaired oculocephalic response, and abnormal motor responses did well. All of these patients had undergone surgical decompression. 5) Elevated ICP was signicantly related to a poorer outcome. This was true both for patients with mass lesions and for those with diffuse brain injury, although the latter less frequently exhibited high ICP. 6) Half of all deaths in this series were associated with severely elevated ICP which could not be controlled despite hyperventilation, cerebrospinal fluid (CSF) drainage, and intravenous mannitol.

Some criticisms can be leveled at our claim for reduced mortality.8,13,14 The criteria for admission of patients to our study may have permitted inclusion of less severely injured patients, with a presumably better outcome, than the series with which we compared our results. Entry to our series was determined by failure to obey commands or utter recognizable words immediately upon resuscitation of the patient in the emergency room, and did not include the status of eye opening to stimuli. The criteria for the large series of head-injury patients from three different countries (1000 cases) analyzed and reported by Jennett, et al.,12 were: no eye opening to painful stimuli, no verbal response, and failure to obey commands for at least 6 hours following injury or subsequent deterioration.12 (Henceforth in this paper, the data from that multinational group of collaborating centers will be referred to as the International Data Bank: IDB.)

After careful survey of the results of head-injury treatment spanning 50 years, Langfitt14 concluded in 1978 that there was still no conclusive evidence that mortality in comparable cases had been reduced by newer or more aggressive therapies. In doing so he, in effect, issued a challenge to the neurosurgical community to study and classify their head-injury patients in such a way that true comparability of data could be obtained. Since 1976, we have collected prospective data on patients with severe head injury as part of the program of a head-injury clinical research center. Data have been collected in a form that is compatible both with our previous study and with the IDB. Having obtained a complete follow-up review in a greater number of patients than the previous study, we can compare results both with our previous study and with the IDB, and test the validity of our previous conclusions concerning the significance for outcome of neurological status and raised ICP. Assessment of head-injured patients sooner than 6 hours from injury or deterioration and inclusion of patients who have eye opening to painful stimuli have been said to weaken prognostic correlations.10,12 The IDB contains no such patients, and this remains an untested hypothesis. The present study provides an opportunity to test it.

Clinical Material and Methods

This study is based on 225 consecutive patients with severe head injury who were seen at the Medical College of Virginia Hospitals between May, 1976, and March, 1980. Criteria for admission to this series were inability to obey commands or utter recognizable words following injury and after appropriate cardiopulmonary resuscitation in the emergency room. The only patients excluded from the series were those with gunshot wounds of the head or those who, on arrival, already fulfilled the criteria of brain death, including apnea. All patients with spontaneous respiration, however poor, were included.

Approximately half of the patients were admitted directly from the scene of the accident. The remainder were referred from another hospital. The majority of patients were seen by a neurosurgeon soon after injury: 20% within 1 hour, 35% within 2 hours, and 70% within 6 hours. The mean delay between injury and examination by our neurosurgical staff was 3 hours.

Patients underwent cranial computerized tomography (CT) scanning as soon as it was possible to move them from the emergency room, and after plain films of head and neck had been obtained. In patients who could not be moved, twist drill ventriculography was performed in the emergency room (four cases). Patients with intracranial mass lesions associated with significant brain shift (>5 mm actual midline displacement), including both extra- and intracerebral lesions, received intravenous mannitol solution (1 gm/kg body weight) and were taken for immediate decompressive craniotomy. Contused brain was excised when necessary and the bone flap was replaced if possible. All patients were managed in a neurosurgical intensive care unit by artificial ventilation and continuous monitoring of ICP by means of a ventricular catheter or subarachnoid screw. Full details of the management protocol have been described elsewhere.1,15

Subsequent neurological evaluation was carried out by a participating neurologist and/or the neurosurgical head-injury fellow on postinjury Days 1, 4, and 14, and in survivors at 3, 6, and 12 months. The neurological findings were recorded on coded data forms and transferred into a computer data bank. Information on sequential CT scans, evoked potential studies, and ICP monitoring was also reduced by appropriate investigators, coded, and transferred to the data bank.

The neurological status of our patients was first recorded as soon as possible after injury. This evaluation was necessary to compare our current results with those of our previous study, and to assess the prognostic effectiveness of such early information compared with data obtained later. In this paper, unless otherwise specified, initial neurological data were obtained in the emergency room or, in the case of patients who deteriorated later, in the neurosurgical intensive care unit.

We have also tried to identify from our patients those who had no eye opening, no verbal response, and did not obey commands (E1, V2, M5, or less on the Glasgow Coma Scale24) for at least 6 hours. This proved to be extremely difficult because, at the 6-hour mark, patients had an endotracheal tube in place, and many were under the effects of anesthesia or muscle relaxants. Twenty-eight percent of our patients who were not obeying commands on admission had eye opening to painful stimuli, and 26% had eye opening to painful stimuli on the following day. However, only five of these were the same patients. In other words, a quarter of the patients deteriorated from eye opening to none and a similar proportion improved in the first 24 hours. Eight patients died within 6 hours of admission. These are included in our series. From this series of 225 patients, we identified 158 (70%) who were known to be at or below the E1-V2-M5 level on admission, and were still at or below this level on examination the next day (6 to 24 hours later). Although this must be regarded as a conservative estimate, these 158 patients are henceforth referred to as the IDB-compatible group, being at least as severe as the cases reported by the participants in the IDB.

The outcome of all patients was assessed at 3 months, 6 months, or 1 year after injury using the Glasgow Outcome Scale.9 All survivors were followed up by personal interviews or, in the case of those patients who had moved out of the area, by telephone conversation with the patients and their next of kin. A complete follow-up survey of all patients admitted during the study period was obtained. We were able to assess outcome at 3 months, 6 months, and 1 year in 128 patients. In comparing outcomes in the same patients between 3 months and 1 year, there was no change in status in 103 cases (80%), improvement in status by one or two points on the Glasgow Outcome Scale in 23 cases (18%), and regression by one point in two cases. In reporting results in 149 survivors, we have therefore used the latest available assessment: 1 year in 102 cases, 6 months in 25 cases, and 3 months in 22 cases.

Relationships between neurological signs, physiological measurements, and outcome were assessed by the chi-square test.

Results
Diagnostic Groups

In our previous series, the most widely used diagnostic technique was air ventriculography. Patients were broadly divided into those with 5-mm or more of midline shift, who were deemed to have mass lesions requiring decompressive surgery, and the remainder who were simply classified as having “diffuse brain injury.” A wider classification could be made in the present series based upon immediate CT scanning in virtually all patients. As before, patients are grouped principally into those who were subjected to decompressive craniotomy and unoperated cases. Ninety-three patients (41%) required craniotomy, about the same proportion as in the earlier series (40%). The lesions in these cases included 18 epidural hematomas, 55 acute subdural hematomas with or without intracerebral hemorrhage and contusion, and 20 intracerebral hematomas with swelling sufficient to cause more than 5-mm calculated midline shift. Decompressive surgery was not necessary in 132 patients (59%); 31 of these had hemorrhagic intracerebral lesions on the CT scan, but without significant brain shift. The remaining 101 patients, classified as having diffuse brain injury, variously had a normal CT scan, small or collapsed ventricles, or areas of lucency in the brain with no, or less than 5 mm actual midline shift.

Age Distribution

The majority of the 225 patients in this series were male (76%); a similar male preponderance was seen in our earlier series (79%). The average age in this series was 31 years (range 2 to 89 years). This is an average increase of 4 years compared with our previous series, due to an increase in the proportion of patients over 40 years old (Fig. 1). The average age of the 158 IDB-compatible patients was also 31 years (range 2 to 84 years). Patients in the surgical group were older (mean age 37 years, range 7 to 84 years) than the nonsurgical group (mean age 27 years, range 2 to 89 years) (p < 0.01).

Fig. 1.
Fig. 1.

Comparison of the age distribution of patients in the present (1976–1979) series with patients in our previous (1972–1976) series. n = number of cases.

Clinical Status on Admission

The clinical status on admission to this study is shown in Table 1. The incidence of abnormal neurological signs is shown in patients in both the surgical and nonsurgical groups. The findings in the IDB-compatible cases within each group are also shown. Not only were the patients in the surgical group older, but they also demonstrated a significantly higher incidence of abnormal neurological signs; 43% of the surgical group had abnormal motor responses (abnormal flexion, extension, or nil) as compared with 20% of the nonsurgical group (p < 0.001). Impaired or absent oculocephalic responses were seen in 56% of the surgical group and 31% of the nonsurgical group, while the incidence of bilaterally unreactive pupils was 38% and 13% in the two groups, respectively (p < 0.001).

TABLE 1

Clinical status on admission to the study*

FactorTotal Series (225 cases)IDB-Compatible Series (158 cases)
Surgical CasesNonsurgical CasesSurgical CasesNonsurgical Cases   
no. of cases931327088
average age (yrs)37273725
abnormal motor response (%)43204526
bilateral fixed pupils (%)38134317
impaired or absent oculocephalic response (%)563155NS38

IDB-compatible series = patients compatible with the International Data Bank, see text. Abnormal motor response includes abnormal flexion, extension, or nil. NS = not significant.

Difference significant, p < 0.001.

Difference significant, p < 0.01.

Of the 93 patients in the surgical group, 70 (74%) were IDB-compatible, while 88 (67%) of the 132 patients in the nonsurgical group were IDB-compatible. The incidence of abnormal neurological signs in these subgroups also reflected the greater neurological severity of patients in the surgical group.

Multiple Injury, Early Physiological Insult, and Cause of Injury

Major systemic injuries were present in addition to the presenting head injury in 109 patients (48%). In 24% of patients there were two or more additional injuries. Limb injuries were most common (31% of the whole series), followed by chest (28%), abdominal (17%), and spinal injury (6%).

One or more of the systemic insults of hypoxemia (PaO2 < 60 mm Hg), hypercarbia (PaCO2 > 45 mm Hg), anemia (hematocrit < 30%), and arterial hypotension (systemic blood pressure < 90 mm Hg) were observed on admission in 100 patients (44% of the series) (Table 2).14 Hypoxemia was most common in 78 patients (37%), followed by hypotension in 34 (16%), anemia in 21 (10%), and hypercarbia in 18 (8%).

TABLE 2

Systemic insults observed on admission*

Systemic InsultsTotal CasesAssociated With MVA (%)Poor Outcome (%)
No.Percent   
arterial hypoxemia (pO2 < 60 mm Hg)78377659
arterial hypotension (SBP < 90 mm Hg)34169165
anemia (hematocrit < 30%)21107762
arterial hypercarbia (pCO2 > 45 mm Hg1887278
none117527335

MVA = motor-vehicle accident; SBP = systemic blood pressure. Poor outcome includes severe disability, vegetative state, and death.

Difference from the 35% “no insult” figure is significant, p < 0.01.

Overall, 75% of injuries were caused by motor-vehicle accidents and 25% by blows, falls, and miscellaneous causes. Multiple injuries were predominantly (90%) associated with motor-vehicle accidents (Table 2). There was a significantly lower incidence of intracranial hematoma in comatose patients whose head injury was caused by motor-vehicle accidents (29%) as compared with other causes of injury (75%) (p < 0.001). Arterial hypotension was mainly seen (91%) in patients who had been involved in motor-vehicle accidents, and exclusively seen in patients with multiple injury. Early hypoxemia was equally common in both classes of head injury, that is, in 41% of patients who had had motor-vehicle accidents and in 37% of patients who had suffered other forms of head injury.

Outcome from Injury

The outcome of patients in the different diagnostic groups is shown in Table 3. Overall, 45% of the 225 patients made a good recovery, 11% were moderately disabled but able to care for themselves, 7% remain severely disabled requiring the help of others, 3% were vegetative, and 34% died. Patients in the surgical group had a higher mortality (55%) than patients in the nonsurgical group (18%) (p < 0.001). The high mortality in the surgical group is confined mainly to patients with acute subdural (65%) and intracerebral hematomas (58%). Within the nonsurgical group, mortality, vegetative survival, and severe disability were more frequent in patients with cerebral contusion (36%) than in patients with diffuse brain injury (27%).

TABLE 3

Outcome in different diagnostic groups*

DiagnosisNo. of CasesGR/MD (%)SD/Veg (%)Dead (%)
epidural hematoma1867528
subdural hematoma5526965
intracerebral mass20261658
all surgical lesions93351055
cerebral contusion (unoperated)31641323
diffuse injury101731017
all nonsurgical lesions132711118
total cases225561034

GR = good recovery; MD = moderate disability; SD = severe disability; veg = vegetative state.

Incidence of Increased Intracranial Pressure

In this study “increased ICP” refers to a persistent elevation of mean ICP over 20 mm Hg during the period of continuous monitoring of ICP in the neurosurgical intensive care unit. The duration of ICP monitoring was generally 3 to 5 days, longer only if ICP remained substantially elevated. (In our previous experience the infection risk increased after 3 days20). When ICP remained above a mean level of 25 mm Hg for more than 15 minutes, attempts were made to reduce it by hyperventilation to an arterial pCO2 of 20 mm Hg, by controlled CSF drainage against a positive pressure of 20 to 25 cm H2O, or by mannitol solution (0.5 or 1.0 gm/kg) as an intravenous bolus. All patients received corticosteroids from the time of admission (dexamethasone 4 mg every 6 hours, or methylprednisolone 40 mg every 6 hours). In 20 patients, a trial of high-dose steroid therapy (methylprednisolone 1000 mg/day) was carried out, but no effect on ICP or outcome was observed.7 These patients are, therefore, included with the total group. Three ICP categories are recognized in this study: “normal ICP” where pressure remained below 20 mm Hg throughout, “elevated ICP” where it rose above 20 mm Hg, and “uncontrollable ICP” where the pressure rose above 20 mm Hg but could not be reduced by the measures described above (Table 4).

TABLE 4

Incidence of raised intracranial pressure (ICP) in different diagnostic classes

DiagnosisNo. of CasesNormal ICP (%)All Raised ICP (%)Uncontrollable ICP (%)
epidural hematoma17356512
subdural hematoma48336737
intracerebral mass16128831
all surgical lesions81307031
cerebral contusions (unoperated)2730706
diffuse injury8867334
all nonsurgical lesions11558425
total cases196465416

In the 196 patients in whom satisfactory ICP data could be obtained, 46% had normal ICP throughout the monitoring period, and 54% had elevated ICP. This finding compares with 40% elevated ICP in our previous series. About one-third of the patients with raised ICP, or 16% of the whole series, had uncontrollable intracranial hypertension progressing to cerebral nonperfusion and death in all but one case (15% previously).

In the present series the greatest incidence of raised ICP occurred in the surgical group of cases (70%). In nearly half of these cases the intracranial hypertension could not be controlled by conventional measures. This high postoperative incidence was evident in all surgical subgroups, including epidural, subdural, and intracerebral hematomas.

Within the nonsurgical group of patients there was a considerable difference in the incidence of raised ICP. Of 27 monitored patients with cerebral contusions (dense lesion on CT) that were not surgically treated, 70% had elevated ICP, but in only 6% was the pressure elevation not controllable by routine measures. In 88 monitored patients with diffuse brain injury, only 29% had elevated ICP, and 4% had uncontrollably elevated ICP. The number of additional patients with elevated ICP who might have developed more serious problems had their mild intracranial hypertension not been detected and treated cannot be known.

Outcome Related to Age

Mortality of patients in this series increased steadily with age, from 19% in the group aged 0 to 20 years to 71% in those aged 61 to 90 years (Table 5). The same relationship was clearly observable in the subgroup of IDB-compatible cases, where mortality ranged from 26% in the group aged 0 to 20 years to 87% in those aged 61 to 90 years. Mirroring the increase in mortality with increasing age was the decreasing frequency of good recovery or moderate disability. The incidence of severely disabled and vegetative patients was not related to age.

TABLE 5

Outcome related to age*

Age (yrs)Total SeriesIDB-Compatible Series
No.GR/MD (%)SD/Veg (%)Dead (%)No.GR/MD (%)SD/Veg (%)Dead (%) 
0–207870111950581626
21–408256103465521137
41–604844124435311257
61–901762371812187

IDB-compatible series = patients compatible with the International Data Bank, see text. GR = good recovery; MD = moderate disability; SD = severe disability; veg = vegetative state.

The relationship between increasing age and mortality was more evident in patients in the nonsurgical group. This is partly because mortality of patients with surgical mass lesions was high throughout, ranging from a low of 54% in the group aged 21 to 40 years to 80% in those aged 61 to 90 years. In patients with nonsurgical lesions there was a clear progression of mortality, from a low of 2% in the group aged 0 to 20 years to 57% in those aged 61 to 90 years. This accentuation of age-related mortality in nonsurgical patients was evident also in the subgroup of IDB-compatible patients.

Outcome Related to Neurological Status

Motor Responses. In this study, we recognized three abnormal forms of motor responses to painful stimuli: abnormal flexion (decortication), extension (decerebration), and no motor response (flaccid). Crucial to this type of analysis is confidence that we were able to detect reliably the difference between normal and abnormal flexion in response to pain. Paired motor examinations were carried out by two pairs of examiners on 62 and 132 occasions. The agreement rate in defining motor responses, including distinguishing normal from abnormal flexion, was 87% for one pair of observers, and 78% for the other.

The presence of any one of the abnormal motor responses was associated with a sharp increase in mortality. Patients with normal motor responses on admission (localizing or normal flexor withdrawal) had 21% mortality, whereas patients with abnormal flexion had 41% mortality, patients with extensor responses 67% mortality, and patients with no motor response 76% mortality (p < 0.001) (Table 6). As we found previously, the correlation between abnormal motor responses and a poor outcome was less strong in patients with surgically treated intracranial mass lesions (p < 0.01) than in patients in the nonsurgical group (p < 0.001). This is mainly because of higher mortality in patients with surgical mass lesions, even when the motor response was localizing or when there was normal flexor withdrawal. The difference between surgical and nonsurgical patients was present both for the entire series and for IDB-compatible patients.

TABLE 6

Mortality related to motor response

Best Motor ResponseTotal SeriesIDB-Compatible Series
 SurgicalNonsurgicalSurgicalNonsurgical
No.% DeadNo.% DeadNo.% DeadNo.% Dead 
localize/withdrawal5343105103855669
abnormal flexor933850650850
extensor2176125018781060
nil1090757786560
 total93551321869648921

Pupil Light Response and Oculocephalic Response. Correlations between bilateral absence of the pupil response to light, impaired or absent oculocephalic responses, and a poor outcome were present in both surgical (p < 0.01) and nonsurgical patients (p < 0.001) in both the full series and the IDB-compatible cases (Table 7).

TABLE 7

Outcome related to pupil light response and oculocephalic response*

ResponseWhole SeriesIDB-Compatible Series
No. of CasesGR/MD (%)SD/Veg (%)Dead (%)No. of CasesGR/MD (%)SD/Veg (%)Dead (%) 
Pupil Light Response
cases with surgical lesions
 present585094139361054
 bilaterally absent3511127730101377
cases with nonsurgical lesions
 present11378111173771211
 bilaterally absent1718117115131374
Oculocephalic Response
cases with surgical lesions
 normal3750123829382646
 impaired/absent472210683415976
cases with nonsurgical lesions
 normal788171246741214
 impaired/absent3948183430481537

IDB-compatible series = patients compatible with the International Data Bank, see text. GR = good recovery; MD = moderate disability; SD = severe disability; veg = vegetative state.

There are significant associations between the presence of various abnormal neurological signs, that is, between abnormal motor responses and impaired pupil reaction to light (p < 0.001), between abnormal motor responses and abnormal eye movement reflexes (p < 0.001), and between abnormal motor responses and elevated ICP (p < 0.01).

Combined Adverse Signs. Of the patients with surgically treated intracranial mass lesions, 24 showed the combination of an abnormal motor response as the best motor response, impaired or absent oculocephalic responses, and bilaterally absent pupil light responses. Of these 24 patients, 17 (71%) died and seven survived, including four who recovered to independent status. Nine patients in the nonsurgical group had this combination of three adverse signs; eight (89%) died and the remaining patient remains severely disabled. Even when the observations were confined to the IDB-compatible patients, three of the 19 patients in the surgical group with the combination of adverse signs made satisfactory recoveries (good/moderately disabled), whereas none of the eight patients in the nonsurgical group with this combination of signs made such a recovery.

Outcome and Glasgow Coma Score Recorded Early and Late

The relationship between the Glasgow Coma Scale (GCS) score, being the sum of the scores for eye opening, motor, and verbal responses, and final outcome was examined using GCS scores recorded in the emergency room and after 24 hours (Table 8). In the entire group of 225 patients there was a strong correlation between decreasing GCS score and increasing mortality (p < 0.001), which was equally sound whether the observation was made in the emergency room after resuscitation or 24 hours from the time of admission. The correlation between GCS score and mortality was equally close in both the full group and the IDB-compatible group of patients.

TABLE 8

Outcome related to Glasgow Coma Scale score recorded on admission and after 24 hours*

Glasgow Coma Scale ScoreWhole SeriesIDB-Compatible Series
No. of CasesGR/MD (%)SD/Veg (%)Dead (%)No. of CasesGR/MD (%)SD/Veg (%)Dead (%) 
on admission
 3–44716137139131374
 5–710859113086521335
 8–1568798133376915
after 24 hours
 3–46211167365201961
 5–7100667277765827
 8–1563845111675619

IDB-compatible series = patients compatible with the International Data Bank, see text. GR = good recovery; MD = moderate disability; SD = severe disability; veg = vegetative state.

Systemic Insults and Outcome

There was an association between poor outcome (severely disabled, vegetative, or dead) and hypoxemia (p < 0.01), arterial hypotension (p < 0.01), and hypercarbia (p < 0.01) (Table 2). The association between anemia and a poor outcome was not significant.

Despite the association between multiple injury and some systemic insults, there was not a significant relationship between the presence of multiple injuries and outcome, nor was there a significant relationship between cause of injury and outcome.

Intracranial Pressure and Outcome

There was a significant correlation between raised ICP and a poor outcome, namely, severely disabled, vegetative, or dead (p < 0.001) (Table 9). The highest proportion of patients who remained severely disabled or vegetative was seen in those patients who had raised ICP that could be controlled by hyperventilation, CSF drainage, and/or mannitol. Virtually all patients in whom ICP could not be brought under control died. Severe intracranial hypertension was the main cause of death in such cases, rising inexorably to the level of arterial pressure and terminating in brain death.

TABLE 9

Outcome related to intracranial pressure*

ICP StatusNo. of CasesGR/MD (%)SD/Veg (%)Dead (%)
normal (0–20 mm Hg)9174918
raised but reducible74551926
not reducible313392
 total196561233

ICP = intracranial pressure; GR = good recovery; MD = moderate disability; SD = severe disability; veg = vegetative state.

Comparison With Other Series

As compared with our previous series, the outcome in this group was somewhat, but not significantly, poorer, with a small increase in mortality from 30% to 34% (Table 10).1 The mean age of the present group of patients was somewhat higher, but the number of patients requiring surgical decompression and the percentage of patients with abnormal neurological signs was similar. The surgical patients in the present series had a higher incidence of raised ICP and a poorer outcome than in our previous study.

TABLE 10

Comparison of present series with previous and other reported series*

Clinical Features & OutcomeRichmond, 1977Richmond, 1981 (Full Series)Richmond, 1981 (IDB-CompatibleInternational Data Bank
GlasgowNetherlandsLos Angeles    
Clinical Features
no. of cases160225158593239168
average age (yrs)273131353235
surgically treated mass lesions (%)404144542856
increased ICP (%)435462
best motor response extensor/nil (%)342225172021
bilaterally fixed pupils (%)212329192931
impaired or absent reflex eye movements (%)404246453740
Outcome
good recovery/moderate disability (%)605647414131
severe disability/vegetative (%)10101212919
dead (%)303440485050

For a description of the International Data Bank (IDB), see text. ICP = intracranial pressure.

When the clinical features and outcome of the subgroup of our IDB-compatible patients are compared with our total group, the mean age of IDB-compatible patients was the same, but more patients who had mass lesions requiring surgical decompression were decerebrate or flaccid and had signs of abnormal brain-stem function. Correspondingly, the mortality in our IDB-compatible subgroup was higher than in the overall group (40% compared with 34%).

The clinical features of our IDB-compatible group of patients are very similar to those actually recorded by the IDB participants. The mortality rate in our IDB-compatible series (40%) is, however, significantly lower than the 49% mortality rate recorded by the members of the IDB.12 This is the case regardless whether comparisons are made between the Richmond IDB-compatible series and the entire IDB series or the subsets of data from its three constituent reporting centers (p < 0.001).

The issue of comparability cannot be completely settled, however. The mean age of the IDB patients is slightly higher (34 versus 31 years), and the incidence of surgical cases is also higher than in our IDB-compatible series (48% versus 44%). While the cases appear to be comparable in terms of neurological dysfunction, we cannot compare data concerning the incidence of intracranial hypertension, early hypoxemia, or shock, the prognostic importance of which we have shown here. Nevertheless, the results of our study, which are so similar to those recently reported by Bowers and Marshall2 and Clifton and his colleagues,5 encourage us in the belief that a real, albeit modest, reduction in the mortality of severe head injury has been obtained by our policy of rapid transfer, early CT diagnosis, early surgical decompression, artificial ventilation, and ICP monitoring.

Discussion
Prognostic Factors

As the number of patients in whom details of the early neurological status, cranial CT scanning, and ICP monitoring have been correlated with final outcome increases, one can become more confident of the prognostic value of certain clinical factors. The presence of an intracranial mass lesion requiring surgical decompression remains for us one of the most important single factors in determining the outcome of patients, and influences the relationship between neurological status and outcome.22 Age is also of high importance, especially in the nonsurgical group.4 Abnormal motor responses were more common in patients with intracranial mass lesions requiring surgical decompression, but were of more consistently adverse significance in patients who did not require surgery. Bilateral absence of the pupil light response and impaired or absent oculocephalic responses, both signs which we associate with abnormal brain-stem function, were strongly associated with increased mortality in all patient subgroups. We have previously commented on the significant correlation between midline shift and abnormal motor and eye movement responses.1 These factors are mainly of value in predicting death. As Bricolo, et al.,3 have recently observed, the same factors are poor predictors of prolonged disability.

Raised ICP remains an important adverse factor, both for death and disability. Severe intracranial hypertension (more than 40 mm Hg) in patients with head injury appears directly responsible for brain ischemia and severe, if not fatal, neurological dysfunction. The association between moderate intracranial hypertension (20 to 40 mm Hg) and a poor outcome (severe disability and vegetative state) more likely reflects the “extent” of brain damage diffusely scattered in the cerebral hemispheres. In this respect, moderate intracranial hypertension may be considered an indicator of the fractional volume of damaged brain rather than the cause of the damage.

This study has also confirmed the importance of parenchymal lesions of increased density on the CT scan in severely head-injured patients, and their association with raised ICP and a poor outcome.6,23 Computerized tomography provides another way of classifying head injuries that can cut across our principal division into surgical and nonsurgical cases based only on midline shift. We can now distinguish parenchymal from extraparenchymal lesions. Patients with parenchymal lesions of increased density on CT comprise a wide clinical spectrum, ranging from alert to comatose. Since the majority of the comatose patients also have significant midline shift, we feel justified in continuing to stress the importance of midline shift as a means of classifying patients. It is noteworthy, however, that the two principal factors associated with a large percentage of patients surviving in the severely disabled or vegetative state were: moderately raised ICP, and parenchymal lesions of increased density on the CT scan with or without midline shift. Patients with a relatively normal CT scan, and hence no mass lesion, had a fairly low incidence of raised ICP. It could therefore be argued that ICP monitoring is unnecessary in this group. This point will be discussed below.

Some Aspects of Management

For the neurosurgeon who manages patients with severe head injury, the two most troublesome questions today probably are: “which patients, if any, should have ICP monitoring?” and “which patients should have decompressive craniotomy?”

By making it a policy to monitor ICP in all comatose head-injured patients, we have been able to indicate when ICP is likely to be high and when it is likely to be normal. Only in ventilated patients with normal CT scans is the incidence of raised ICP less than 25%. We have also shown a correlation between raised ICP and severe morbidity and mortality. We feel we can safely conclude that ICP is an important variable to measure in head-injured patients. A further question now being asked, however, is “can you show that ICP monitoring results in a lower mortality and thereby justifies the risks of intracranial infection which are entailed?” We cannot answer this question directly, since all of our patients have been monitored. Bowers and Marshall2 found a lower mortality in deeply comatose patients managed with ICP monitoring compared to those managed without. It is doubtful, however, if comparison with other series will conclusively answer this question. A randomized trial with large numbers of comparable patients is necessary, and failure to demonstrate an advantage for ICP monitoring does not prove that this does not exist. One cannot help but feel that to ask for proof of the utility of ICP monitoring in terms of improved outcome is somewhat unfair, analogous to demanding proof that arterial pressure measurement reduces mortality from myocardial infarction. It seems reasonable to suggest from our data, however, that patients with no intracranial mass effect and a relatively normal CT scan should be considered for epidural ICP monitoring because it is safer and because CSF drainage is seldom required for control of ICP in such cases.

Questions about craniotomy can be equally thorny. When CT shows a thick extracerebral hematoma with an appropriate midline shift, the decision to operate is easy. When CT shows a thin extracerebral lesion with a disproportionately large shift, the decision is harder. In such cases we have opted for craniotomy based on experience that the hematoma is frequently larger than appears on CT (presumably because it is partly isodense). Others would opt for conservative management and control of ICP, and we have found that persistent brain swelling remains a problem, as indicated by our high incidence of postoperative intracranial hypertension. Our greatest problem arises in deciding whether or not decompressive craniotomy is indicated in patients who have intraparenchymal lesions of increased density on CT. Such lesions may be intracerebral hematomas, hemorrhagic contusions, or hemorrhagic infarcts. We have chosen craniotomy when such lesions were associated with a more than 5-mm actual midline shift, but the 88% incidence of postoperative intracranial hypertension, which proved fatal in nearly a third of patients, is eloquent testimony to the inadequacy of this solution.

It is only since the advent of CT that we have been able to recognize that for every case of intracerebral “hematoma” with shift there are two without shift. Until the genesis of such lesions is better understood, treatment will be empirical and, in the case of surgical decompression, unsatisfactory. We have remarked elsewhere on the possible role of hypoxemia in the genesis of “delayed intracerebral hematoma” on CT.6 Further analysis of our data is proceeding to see whether early hypoxemia also favors the appearance of early dense CT lesions.

Defining Severe Head Injury

For the purpose of formulating an accurate prognosis, Jennett, et al., have recommended restricting the definition of severe head injury to those patients who have no eye opening, no verbal response, and cannot obey commands for a period of 6 hours or more after injury or deterioration.10–12,24 Inclusion of less severely injured patients, or patients assessed at an earlier stage, might permit inclusion of many patients who have suffered only from alcohol overdose, undiagnosed epilepsy, or milder degrees of trauma, and therefore are more likely to have a favorable outcome from injury. The dilution of a series by less severely injured patients weakens prognostic correlations between neurological signs and outcome. Despite the logic of this argument, we have not found this to be the case. Every correlation between physiological status and outcome was equally valid both in the IDB-compatible patients and the IDB-noncompatible patients that we studied. Although the incidence of poor outcome is significantly higher in IDB-compatible (52%) than in IDB-noncompatible cases (24%), tests for second-order interaction showed the same patterns of correlation. Neurological information gathered from less severely injured patients and at an early stage after head injury, when the patient has been resuscitated in the emergency room, appears to be as useful in developing prognostic algorithms as information that is gathered only from more severe cases some hours later. This does not imply, however, that the neurological status of individual patients does not change in the first 24 hours. It does, of course, and the direction of that change, improvement or deterioration, is a useful predictive factor for those particular patients.

We would make a plea for the inclusion in the definition of “severe head injury” of all patients who are unable to speak and unable to obey commands on admission to the hospital, irrespective of the status of eye opening. Our reason for making this plea is that if observations and correlations are restricted to the most severely injured patients, in whom there is an extremely high mortality, it becomes more difficult to discern the interplay of prognostic factors on outcomes other than death and to distinguish any effect on outcome of different modes of therapy. For example, in a group of patients in whom the mortality rate is in excess of 60%, about half of those deaths are likely to be unavoidable.19 Even a completely safe treatment that reduces the remaining deaths by a third can only have a 10% effect on overall mortality. An impossibly large sample size may be required to demonstrate such a difference reliably. A similar study conducted in a series with a smaller proportion of patients with irremediable brain damage might be more successful in gauging the efficacy of a new treatment. Head injury occurs in a continuum of severity. To limit scientific interest to a subpopulation defined too narrowly may result in a predictable study of uniformity rather than a reflection of the natural history and effects of therapy over the spectrum of this serious neurological disease.

References

Article Information

Address reprint requests to: J. Douglas Miller, M.D., Professor of Neurosurgery, MCV Station Box 508, Richmond, Virginia 23298.

© AANS, except where prohibited by US copyright law.

Headings

Figures

  • View in gallery

    Comparison of the age distribution of patients in the present (1976–1979) series with patients in our previous (1972–1976) series. n = number of cases.

References

1.

Becker DPMiller JDWard JDet al: The outcome from severe head injury with early diagnosis and intensive management. J Neurosurg 47:4915021977J Neurosurg 47:

2.

Bowers SAMarshall LF: Outcome in 200 consecutive cases of severe injury treated in San Diego County: a prospective analysis. Neurosurgery 6:2372421980Neurosurgery 6:

3.

Bricolo ATurazzi SFeinotto G: Prolonged posttraumatic unconsciousness. Therapeutic assets and liabilities. J Neurosurg 52:6256341980J Neurosurg 52:

4.

Carlsson C-Avon Essen CLöfgren J: Factors affecting the clinical course of patients with severe head injuries. Part 1: Influence of biological factors; Part 2: Significance of posttraumatic coma. J Neurosurg 29:2422511968J Neurosurg 29:

5.

Clifton GLGrossman RGMakela MEet al: Neurological course and correlated computerized tomography findings after severe closed head injury. J Neurosurg 52:6116241980J Neurosurg 52:

6.

Gudeman SKKishore PRSMiller JDet al: The genesis and significance of delayed traumatic intracerebral hematoma. Neurosurgery 5:3093131979Neurosurgery 5:

7.

Gudeman SKMiller JDBecker DP: Failure of high-dose steroid therapy to influence intracranial pressure in patients with severe head injury. J Neurosurg 51:3013061979J Neurosurg 51:

8.

Heiden JSWeiss MH: Head injury management. J Neurosurg 48:4914931978 (Letter)J Neurosurg 48:

9.

Jennett BBond M: Assessment of outcome after severe brain damage. A practical scale. Lancet 1:4804841975Lancet 1:

10.

Jennett BTeasdale G: Aspects of coma after severe head injury. Lancetv 1:8788811977Lancet 1:

11.

Jennett BTeasdale GBraakman Ret al: Predicting outcome in individual patients after head injury. Lancet 1:103110341976Lancet 1:

12.

Jennett BTeasdale GBraakman Ret al: Prognosis of patients with severe head injury. Neurosurgery 4:2832891979Neurosurgery 4:

13.

Jennett BTeasdale GGalbraith S: Treatment of head injuries. J Neurosurg 49:9439441978 (Letter)J Neurosurg 49:

14.

Langfitt TW: Measuring the outcome from head injuries. J Neurosurg 48:6736781978Langfitt TW: Measuring the outcome from head injuries. J Neurosurg 48:

15.

Miller JDBecker DPWard JDet al: Significance of intracranial hypertension in severe head injury. J Neurosurg 47:5035161977J Neurosurg 47:

16.

Miller JDSweet RCNarayan Ret al: Early insults to the injured brain. JAMA 240:4394421978JAMA 240:

17.

Pagni CA: The prognosis of head injured patients in a state of coma with decerebrated posture. J Neurosurg Sci 17:2892951973Pagni CA: The prognosis of head injured patients in a state of coma with decerebrated posture. J Neurosurg Sci 17:

18.

Pazzaglia PFrank GFrank Fet al: Clinical course and prognosis of acute post-traumatic coma. J Neurol Neurosurg Psychiatry 38:1491541975J Neurol Neurosurg Psychiatry 38:

19.

Rose JValtonen SJennett B: Avoidable factors contributing to death after head injury. Br Med J 2:6156181977Br Med J 2:

20.

Rosner MJBecker DP: ICP monitoring: complications and associated factors. Clin Neurosurg 23:4945191976Clin Neurosurg 23:

21.

Rossanda MSelenati AVilla Cet al: Role of automatic ventilation in treatment of severe head injuries. J Neurosurg Sci 17:2652701973J Neurosurg Sci 17:

22.

Stablein DMMiller JDChoi SCet al: Statistical methods for determining prognosis in severe head injury. Neurosurgery 6:2432481980Neurosurgery 6:

23.

Sweet RCMiller JDLipper Met al: The significance of bilateral abnormalities on the CT scan in patients with severe head injury. Neurosurgery 3:16211978Neurosurgery 3:

24.

Teasdale GJennett B: Assessment of coma and impaired consciousness. A practical scale. Lancet 2:81841974Lancet 2:

TrendMD

Metrics

Metrics

All Time Past Year Past 30 Days
Abstract Views 5 5 5
Full Text Views 219 219 115
PDF Downloads 63 63 39
EPUB Downloads 0 0 0

PubMed

Google Scholar