British Journal of Neurosurgely (1992) 6, 549-557
ORIGINAL ARTICLE
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
Outcome prediction in early management of severe head injury: an experience in Malaysia BENEDICT M. SELLADURAI,* RAMASAMY JAYAKUMARJ YAN Y. TAN* & HENG C. LOWS
*Department of Surge y, tDepartment of Radiology and $Department of Preventive Medicine, School of Medical Sciences, Hospital Universiti Sains Malaysia, Kubang Kerian, Kelantan, Malaysia
Abstract The outcome of 109 patients with severe head injury was studied in relation to clinical and computed tomographic (CT) criteria on admission, after resuscitation. Age, Glasgow Coma Score (GCS) and state of pupils strongly correlated with outcome. The presence of hypothalamic disturbances, hypoxia and hypotension were associated with an adverse outcome. The C T indicators associated with poor outcome were perimesencephalic cistern (PMC) obliteration, subarachnoid haemorrhage, diffuse axonal injury and acute subdural haematoma. The prognostic value of midline shift and mass effect were influenced by concomitant presence of diffuse brain injury. For the subset of patients aged t20 years, with GCS 6-8 and patent P M C (n=21), 71.4% correct predictions were made for a good outcome. For the subset of patients aged >20 years, with GCS 3-5 and partial or complete obliteration of PMC (n=28), 89.3% correct predictions were made for a poor outcome. Key words: Compured tomography, outcome prediction, severe head injury.
Introduction Outcome prediction is of great importance in early management of severe head injury. In a country such as Malaysia’ availability of trained staff and intensive care facilities are restricted. In these circumstances it is even more important to identify patients with severe head injury who have the potential for a good functional outcome, who would greatly benefit from such restricted resources. A series of comprehensive studies,I-l2 have demonstrated the prognostic value of clinical features and computerized tomography (CT) scan findings in severe head injury. These studies have been exclusively conducted in developed countries under nearly ideal conditions. The aims of this study are (a) to outline the outcome profile of
severe head injury in our experience, in relation to established clinical and C T criteria evident on admission (after resuscitation) and (b) to determine if these criteria could be employed to predict outcome for certain subsets of patients with severe head injury.
Materials and methods Patient population This study was conducted on 109 consecutive patients with severe head injury admitted to Hospital Universiti Sains Malaysia’ from April 1989 to April 1991. During the same period, 946 patients with head injuries of all grades of severity were admitted to the same institution.
549
550
Benedict M . Selladurai et al.
Severe head injury was defined as a Glasgow Coma Score (GCS) of 8 or less at or within 48 h of injury. Patients with clinical evidence of brain death were excluded from the study.
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
Data collection Clinical and C T criteria considered for this study are shown in Table I. In all patients CT scans were performed immediately following cardiopulmonary stabilization. All C T scans were performed on a Phillips 359 Tomoscan with a 256 x 256 matrix. T h e plane of orientation was parallel to the orbitomeatal line. A total of nine slices each with a 9 mm thickness were made. Intravenous contrast agents were not given. T h e following guidelines were used in the abstraction of data from CT scans: midline shift was assessed using the septum pellucidum as a midline structure; perimesencephalic cisterns (PMC) were considered partially obliterated if they were only visible as thin hypodense slits or only visible on one side; diffuse brain swelling was considered present
if the ventricles were small and compressed, with obliteration of sulci over the hemispheres; diffuse axonal injury was considered present in scans showing small eccentrically placed haemorrhages in the deep white matter, corpus callosum ganglionic region, corticomedullarj junction, brainstem or in the ventricular system; haemorrhagic contusions and intracerebra1 haematomata were considered as a single entity.
Assessment of outcome Outcome was assessed 6 months after injurj using the Glasgow Outcome Scale." Death, persistent vegetative state and severe disabilitl were considered a poor outcome. Good recovery and moderate disability were considered a good outcome.
Results T h e age range of the patient population in thif study was 9 months to 75 years with a mear age of 28.5 years. T h e GCS was 3-5 in 37
TABLE I. Clinical and C T criteria considered for this study 1. Clinical criteria (on admission, after resuscitation)
Age GCS State of pupils Hypothalamic disturbance (hyperpyrexia, sweating, hypertension) Hypoxia ( p O , t 6 5 mm torr) Hypotension (systolic blood pressure(90 mmHg) Lucid interval 2.
Criteria in initial C T scan (a) Non-focal pathological changes State of perimesencephalic cisterns CT-visible subarachnoid blood Diffuse axonal injury Midline shift Mass effect Diffuse brain swelling (b)
Focal lesions Extracerebral lesions Extradural haematoma Acute subdural haematoma Intracerebral lesions Single contusion/intracerebral haematoma Multiple contusions/intracerebralhaematomata
(c)
Normal C T scans
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
Outcome prediction in severe head injury (34%) patients and pupillary abnormalities were evident in 46 patients (unilateral 23.8%, bilateral 18.3%). The clinical characteristics of the patient population in this study were compared with those from a similar study conducted in a developed country5 (Fig. la). Most clinical characteristics were similar in the two groups of patients except that the injury-admission interval was significantly longer in this study. The outcome for the entire patient population in this study was: good recovery, n=29 (26.6%); moderate disability, n = 12 ( 1 1 .O%); severe disability, n= 16 (14.7%); persistent vegetative state, n = 8 (7.3%); death, n =44 (40.4%).
Outcome in relation to clinical criteria (Fig. 1b) An inverse relationship was evident between age and a good outcome. The risk of a poor outcome nearly doubled in patients over 50 years as compared with those below 20 years. A significant correlation was noted between the GCS, state of pupils and a poor outcome (p 10 mm. However, for the entire patient population, the degree of midline shift did not prove to be of predictive significance (p=0.39). Of the 43 patients with a midline shift(10 mm, 23 showed evidence of diffuse axonal injury and 10 had bilateral haemorrhagic lesions. Mass effect. No relationship was evident between presence of mass effect and a poor outcome (p=0.16). In this study, the volume of mass lesions was not directly measured from the C T scans. Lipper et al.'* used the number of slices a haemorrhagic lesion was seen in the scan as an index of the volume of a mass lesion. Using this criterion, the outcome was related to the size of mass lesion in 60 patients. No predictive significance was evident (p =0.28). Of the 40 patients who did not demonstrate a focal mass effect, 14 showed obliteration of PMC, six showed CT-visible subarachnoid blood and 10 showed evidence of diffuse axonal injury. Diffuse brain swelling. The presence of diffuse brain swelling did not indicate a poor outcome ( p = O . 1 1 ) . Of the 23 patients who had hypoxia and or hypotension on admission, 21 showed evidence of diffuse brain swelling.
Focal lesions (Fig. 3a) Acute subdural haematoma (acute SDH).
552
Benedict M . Selladurai et al. 14 100-
go
80 -
-
0,
60-
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
a 40-
20-
0interval
Pupillary abnormality
major inluries
interval c6 hours
100-
80 -
-3
-
-
60-
0
E
2u' 40a" 20-
0-
c 2 0 2&50
>50
6 8
3-5
None Unilal Bilat
Absent Present
Absent Presenl
Age in ye.3rs
Glasgow coma
Pupillary
Hypothalamic
Hypoxia or shock
score
abnormality
disturbance
p 002
p 00002
p 00002
pa02
p 002
FIG.1. (a) Comparison of clinical characteristics on admission of patients in this study ( u),with those of a study by Van Good recovery; 0, Dongen et al.' (B). (b) Outcome of patients in this study in relation to clinical characteristics. 0, moderate disability; UU, severe disability; vegetative; B, death.
m,
Acute SDH proved to be the focal lesion with the greatest potential for a poor outcome (p=0.0062). A high incidence of co-existing lesions were evident in the 17 patients with acute SDH (haemorrhagic contusions/intracerebral haematomata in nine patients, diffuse axonal injury in four patients). All 15 patients with acute SDH and a poor outcome had evidence of PMC obliteration, nine had unilateral hemisphere swelling. T h e interval from injury to surgery was more than 4 h in 11 patients. Extradural haematoma (EDH). Of the eight
patients with EDH who had a poor outcome, four had associated intrinsic haemorrhagic lesions, and one had evidence of diffuse axonal injury. Two patients who presented in a poor neurological state after rapid deterioration showed extradural haematomas of heterogenous density on CT suggesting ongoing bleeding. Haemorrhagic contusion/intracerebral haematoma. T h e outcome was only slightly better in patients with a single intrinsic haemorrhagic lesion as compared with those with multiple such lesions. There was a significant incidence
Outcome prediction in severe head injury
(32)
(77)
n n
-
80
(78)
553
(31)
nn
-
v)
I
C
ar
+
60-
+
0
Iu
m
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
2 40-
$ a 20
-
0-
Patent Partial ComDlete Obliteration
Absent Present
State of perimessencephalic
CT visible
Cisterns
Subarachnoid blood
p=o 0001
p-0 0002
II I I
I t f
II II I
4 m r n 5-10mm >lOrnm
.,
Diffuse axonal injury p=O 0017
nn n
(b)
80-H
Absent Present
I
I
I 1
I
Absent
I
It
I
I
Absent Present
Midline shift
Mass effect
Diffuse brain swelling
p=o 11
p p 0 16
p=0.39
FIG.2. (a, b) Outcome o f patients in this study in relation t o non-focal pathological changes on i n i t i a l CT scan. Outcome of patients in this study in relation t o clinical characteristics. 0, G o o d recovery; 0, moderate disability; LII], severe disability;
69, vegetative;
death.
of associated lesions in the 22 patients with single intrinsic haemorrhagic lesions who had a poor outcome (diffuse axonal injury 4, acute SDH 4 and EDH 3).
Normal CT scans. A good outcome was evident in six out of the eight (75%) patients who had normal C T scans. Both patients who had a poor outcome with a normal CT scan were of
554
Benedict M. Selladurai et al.
-
80
(0 +
p
60
0
-
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
0
al
5c
40
8
ti
20
0
Extra dural haematoma
Single ICHiContusion
Multiple ICHiContusion
Acute suhdural haey?toma ’*
i
100
-
80
u) +
a,
60
0
c 0 al
cn
2
plOmm D=o
003
.,
FIG.3. (a) Outcome in relation to type of focal lesion seen an initial CT scan. (b) Outcome in relation to three groups of patients: (1) those with subarachnoid blood, cistern obliteration and midline shift >10 mm; (2) those with all other abnormalities; (3) those with normal CT scans. Outcome of patients in this study in relation to clinical characteristics. 0, Good recovery; J,moderate disability; UO, severe disability; B, vegetative; death.
the GCS 3-5 range with one patient having hypoxia and hypotension on admission. The outcome was then considered for three
categories of patients: (1) those with PMC obliteration and/or CT-visible subarachnoid blood and/or a midline shift of >10 mm: (2)
Outcome prediction in severe head injury
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
those with all other abnormalities and ( 3 ) those with normal C T scans (Fig. 3b). Patients in category (1) had a significant risk of a poor outcome (p=0.0003). There was no significant difference in mortality between categories ( 2 ) and ( 3 ) , although there was a higher incidence of functional incapacitation in the former category.
Prediction of outcome The principles used for prediction of outcome were based on the prediction tree model of Choi et al.,'O where the patient population was divided into smaller subsets using recursive splitting, based on critical levels of different prognostic variables. However, a tree structure was not compiled for the entire patient population in this study in view of the rather small sample size. Instead the criteria of age, GCS and state of PMC were used to identify two subsets of patients for prediction of good outcome and poor outcome, respectively: (1 ) the subset aged t 2 0 years, with GCS 6-8 and without obliteration of MCP ( n = 2 1 ) , for whom a good outcome was predicted; ( 2 ) the subset aged >20 years, with GCS 3-5 and with partial or complete obliteration of PMC ( n = 2 8 ) , for whom a prediction of poor outcome was made. On comparison with actuarial outcome, correct predictions were 71.4% for category (1) and 89.3% for category ( 2 ) , respectively. Discussion
A major goal of current studies in severe head injury has been to define criteria that are of value in predicting outcome. Prediction of outcome based on early clinical and C T findings has its drawbacks. The pathology of severe head injury is heterogeneous and its outcome depends on a complex interaction of a multitude of factors, some of which may not be in evidence in early stages of injury. Rapid transfer by trained personnel and high-quality intensive care are hallmarks of care for severe head injury in developed countries. In a developing country such facilities are either
555
lacking or are of limited availability. There was no significant difference in the clinical characteristics of the patient population in this study, when compared with that in a similar study performed in a developed country5 except that there was a significant delay between injury and the initial C T in this study. Not withstanding the above-mentioned limitations, early prediction in severe head injury is even more important in a developing country to make optimal use of limited resources available. The predictive power of established clinical criteria of age, GCS, and state of pupils was well illustrated in this study. In addition, evidence of a hypothalamic disturbance, hypoxia and hypotension on admission also indicated an adverse outcome. The role of C T features in prediction of outcome in severe head injury has been extensively investigated in previous studies.2-7~9~11-19 At least three previous s t u d i e ~ ~ * ~ - ~ conclusively demonstrated the predictive potential of PMC obliteration in severe head injury. In our study this feature showed a significant association with poor outcome. Toutant et al.' demonstrated the value of PMC in the first few hours after injury, especially in prediction of potentially lethal events in patients in the higher GCS range of 6-8. Our experience demonstrated the ability of this feature to identify patients in the GCS 6-8 range, in whom urgent evacuation of intracranial mass lesion was warranted. The study of the National Institutes of Health Trauma Coma Data Bank9 (NIH study), was the first large CT-based study to conclusively demonstrate the prognostic value of CT-visible subarachnoid blood, which was evident in 40% of patients in that study. In our patient population, this feature was evident in 29% of patients and was definitely correlated with a poor outcome. In our experience, patients who had evidence of diffuse axonal injury (DAI) in association with PMC obliteration, had a uniformly poor outcome, a feature also highlighted in previous studi e ~ . ' ~ DAI . ' ~ also played a role in the poor outcome of some patients in this study, after surgical evacuation of mass lesions. The NIH
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
556
Benedict M. Selladurai et al.
study9 demonstrated midline shift to be a strong predictor of poor outcome. In our experience, a majority of patients with a midline shift > l O mm had a poor outcome. However, midline shift did not prove to be of statistical significance for the entire patient population. Clifton et a1.2 and Ross et al.” demonstrated the presence of smaller degrees of midline shift in patients with diffuse brain injury, who had a poor outcome. Similarly, in our study a significant proportion of patients with midline shift (10 mm had evidence of diffuse axonal injury and bilateral haemorrhagic lesions. An association between presence of a focal mass effect and poor outcome has been demonstrated in some previous stud~ ~ s . ~ JHowever, *J’ the NIH study’ and a study by Tabbador et a1.I4 did not demonstrate such a correlation, as has also been our experience. In this study, a significant incidence of predictors of poor outcome such as PMC obliteration, CT-visible subarachnoid blood diffuse axonal injury were evident in patients without focal mass effect. In addition, the mitigation of effects of mass lesions by early surgical removal may also account for the diminished prognostic value. The NIH study9 demonstrated an association between hypoxia, hypotension and diffuse brain swelling, a feature also shown in our study. Choi et al.lo used the prediction tree technique to identify subsets of patients with severe head injury, using critical levels of several prognostic criteria. This provided for minimal variability within a given subset with consideration given to interaction of different prognostic criteria. Furthermore, this methodology was more easily applicable than the sophisticated statistical models such as logistic regression and discriminant analysis. Hence we considered it more suitable for use in conditions such as exist in Malaysia. In view of the relatively small size of the patient population in this study a complete prediction tree was not compiled. Instead we restricted our predictions to two subsets of patients selected using critical levels of the most significant prognostic criteria of age, GCS and state of PMC. The mediction rates achieved for the two subsets of
patients are similar to those Choi et al.’” achieved for a much larger patient population managed under more sophisticated conditions. In the final analysis, it has to be conceded that prediction of outcome in severe head injury will remain a difficult and subjective exercise. It can only supplement and not replace clinical judgement.
Conclusions A study of outcome in 109 patients with severe head injury correlated to clinical and C T criteria on admission indicates that: (1) age, GCS and state of pupils remain the predominant clinical indicators of outcome; (2) obliteration of perimesencephalic cisterns, subarachnoid haemorrhage, diffuse axonal injury and acute subdural haematoma are CT features strongly related to a poor outcome; (3) presence of diffuse brain injury could influence prognostic value of midline shift and focal mass effect; (4) modest success was achieved with use of the prediction tree technique for two subsets of patients selected using criteria of age, GCS and state of PMC. Address for correspondence: Benedict M. Selladurai, Department of Surgery, School of Medical Sciences, 15990 Kota Bharu, Kelantan, Malaysia.
References Jennett B, Teasdale G, Braakman R, et al. Predicting outcome in individual patients after severe head injury. Lancet 1976; 1:1031-4. Clifton GL, Grossman RG, Makela ME, et al. Neurological course and correlated computer tomography findings after severe head injury. J Neurosurg 1980, 52:611-24. Narayan RK, Greenberg RP, Miller JD, et al. Improved confidence of outcome prediction in severe head injury. A comparative analysis of clinical examination, multimodality evoked potentials, CT scanning and intracranial pressure. J Neurosurg 1981; 54:751-62. Genarelli TA, Speilman GM, Langfitt TW, et al. Influence of type of intracranial lesion on outcome of severe head injury. J Neurosurg 1982; 56:26-32. Van Dongen KJ, Braakman R, Gelpke GJ. The prognostic value of computerized tomography in
Outcome prediction in severe head injury
6
7
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
8 9
10
11 12
comatose head-injured patients. J Neurosurg 1983; 59~951-7. Lobato RD, Cordobes F, Rivas JJ, et al. Outcome of severe head injury related to the type of intracranial lesion-a computerized tomography study. J Neurosurg 1983; 59:762-74. Toutant SM, Klauber MR, Marshall LF, et al. Absent or compressed basal cisterns on first C T scan: ominous predictors of outcome in severe head injury. J Neurosurg 1984; 61591-4. Choi SC, Narayan RK, Anderson RL. Enhanced specificity of prognosis in severe head injury. J Neurosurg 1988; 69:381-5. Eisenberg HM, Gary Jr HE, Aldrich EF, et al. Initial C T findings in 753 patients with severe head injury-a report from the NIH Traumatic Coma Data Bank. J Neurosurg 1990; 73:688-98. Choi SC, Muizelaar IP, Barnes T Y , et al. Prediction tree for severely head-injured patients. J Neurosurg 1991; 75~251-5. Jennet B, Bond M. Assessment of outcome after severe brain damage. A practical scale. Lancet 1975; 1:480-4. Lipper MH, Kishore PRS, Enas GG, et al. Computed tomography in the prediction of outcome in head injury. AIR 1985; 144:483-6.
557
13 Ross DA, Olsen WL, Ross AM, et al. Brain shift, level of consciousness and restoration of consciousness in patients with acute intracranial haematoma. J Neurosurg 1989; 71:498-502. 14 Tabbador K, Danziger A, Wisoff HS. Estimation of intracranial pressure by C T scan in closed head trauma. Surg Neurol 1982; 18:212-15. 15 Kishore PRS, Lipper MH, Becker DP, et al. Significance of C T in head injury: correlation with intracranial pressure. AJNR 1981; 2:307-11. 16 Ropper AH. Lateral displacement of brain and level of consciousness in patients with acute hemispherical mass. N Engl J Med 1986; 314:953-8. 17 Miller JD, Gudeman SK, Kishore PRS, et al. C T scan, ICP and early neurological evaluation in prognosis of severe head injury. Acta Neurochir (Wein) 1979; 28:86-8. 18 Genarelli TA, Thibault LE, Adams JH, et al. Diffuse axonal injury and traumatic coma in the primate. Ann Neurol 1982; 12:564-74. 19 Levi L, Guilburd JN, Lemberger A, et al. Diffuse axonal injury: analysis of 100 patient with radiological signs. Neurosurgery 1990; 27:429-32.
ORIGINAL ARTICLE
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
Outcome prediction in early management of severe head injury: an experience in Malaysia BENEDICT M. SELLADURAI,* RAMASAMY JAYAKUMARJ YAN Y. TAN* & HENG C. LOWS
*Department of Surge y, tDepartment of Radiology and $Department of Preventive Medicine, School of Medical Sciences, Hospital Universiti Sains Malaysia, Kubang Kerian, Kelantan, Malaysia
Abstract The outcome of 109 patients with severe head injury was studied in relation to clinical and computed tomographic (CT) criteria on admission, after resuscitation. Age, Glasgow Coma Score (GCS) and state of pupils strongly correlated with outcome. The presence of hypothalamic disturbances, hypoxia and hypotension were associated with an adverse outcome. The C T indicators associated with poor outcome were perimesencephalic cistern (PMC) obliteration, subarachnoid haemorrhage, diffuse axonal injury and acute subdural haematoma. The prognostic value of midline shift and mass effect were influenced by concomitant presence of diffuse brain injury. For the subset of patients aged t20 years, with GCS 6-8 and patent P M C (n=21), 71.4% correct predictions were made for a good outcome. For the subset of patients aged >20 years, with GCS 3-5 and partial or complete obliteration of PMC (n=28), 89.3% correct predictions were made for a poor outcome. Key words: Compured tomography, outcome prediction, severe head injury.
Introduction Outcome prediction is of great importance in early management of severe head injury. In a country such as Malaysia’ availability of trained staff and intensive care facilities are restricted. In these circumstances it is even more important to identify patients with severe head injury who have the potential for a good functional outcome, who would greatly benefit from such restricted resources. A series of comprehensive studies,I-l2 have demonstrated the prognostic value of clinical features and computerized tomography (CT) scan findings in severe head injury. These studies have been exclusively conducted in developed countries under nearly ideal conditions. The aims of this study are (a) to outline the outcome profile of
severe head injury in our experience, in relation to established clinical and C T criteria evident on admission (after resuscitation) and (b) to determine if these criteria could be employed to predict outcome for certain subsets of patients with severe head injury.
Materials and methods Patient population This study was conducted on 109 consecutive patients with severe head injury admitted to Hospital Universiti Sains Malaysia’ from April 1989 to April 1991. During the same period, 946 patients with head injuries of all grades of severity were admitted to the same institution.
549
550
Benedict M . Selladurai et al.
Severe head injury was defined as a Glasgow Coma Score (GCS) of 8 or less at or within 48 h of injury. Patients with clinical evidence of brain death were excluded from the study.
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
Data collection Clinical and C T criteria considered for this study are shown in Table I. In all patients CT scans were performed immediately following cardiopulmonary stabilization. All C T scans were performed on a Phillips 359 Tomoscan with a 256 x 256 matrix. T h e plane of orientation was parallel to the orbitomeatal line. A total of nine slices each with a 9 mm thickness were made. Intravenous contrast agents were not given. T h e following guidelines were used in the abstraction of data from CT scans: midline shift was assessed using the septum pellucidum as a midline structure; perimesencephalic cisterns (PMC) were considered partially obliterated if they were only visible as thin hypodense slits or only visible on one side; diffuse brain swelling was considered present
if the ventricles were small and compressed, with obliteration of sulci over the hemispheres; diffuse axonal injury was considered present in scans showing small eccentrically placed haemorrhages in the deep white matter, corpus callosum ganglionic region, corticomedullarj junction, brainstem or in the ventricular system; haemorrhagic contusions and intracerebra1 haematomata were considered as a single entity.
Assessment of outcome Outcome was assessed 6 months after injurj using the Glasgow Outcome Scale." Death, persistent vegetative state and severe disabilitl were considered a poor outcome. Good recovery and moderate disability were considered a good outcome.
Results T h e age range of the patient population in thif study was 9 months to 75 years with a mear age of 28.5 years. T h e GCS was 3-5 in 37
TABLE I. Clinical and C T criteria considered for this study 1. Clinical criteria (on admission, after resuscitation)
Age GCS State of pupils Hypothalamic disturbance (hyperpyrexia, sweating, hypertension) Hypoxia ( p O , t 6 5 mm torr) Hypotension (systolic blood pressure(90 mmHg) Lucid interval 2.
Criteria in initial C T scan (a) Non-focal pathological changes State of perimesencephalic cisterns CT-visible subarachnoid blood Diffuse axonal injury Midline shift Mass effect Diffuse brain swelling (b)
Focal lesions Extracerebral lesions Extradural haematoma Acute subdural haematoma Intracerebral lesions Single contusion/intracerebral haematoma Multiple contusions/intracerebralhaematomata
(c)
Normal C T scans
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
Outcome prediction in severe head injury (34%) patients and pupillary abnormalities were evident in 46 patients (unilateral 23.8%, bilateral 18.3%). The clinical characteristics of the patient population in this study were compared with those from a similar study conducted in a developed country5 (Fig. la). Most clinical characteristics were similar in the two groups of patients except that the injury-admission interval was significantly longer in this study. The outcome for the entire patient population in this study was: good recovery, n=29 (26.6%); moderate disability, n = 12 ( 1 1 .O%); severe disability, n= 16 (14.7%); persistent vegetative state, n = 8 (7.3%); death, n =44 (40.4%).
Outcome in relation to clinical criteria (Fig. 1b) An inverse relationship was evident between age and a good outcome. The risk of a poor outcome nearly doubled in patients over 50 years as compared with those below 20 years. A significant correlation was noted between the GCS, state of pupils and a poor outcome (p 10 mm. However, for the entire patient population, the degree of midline shift did not prove to be of predictive significance (p=0.39). Of the 43 patients with a midline shift(10 mm, 23 showed evidence of diffuse axonal injury and 10 had bilateral haemorrhagic lesions. Mass effect. No relationship was evident between presence of mass effect and a poor outcome (p=0.16). In this study, the volume of mass lesions was not directly measured from the C T scans. Lipper et al.'* used the number of slices a haemorrhagic lesion was seen in the scan as an index of the volume of a mass lesion. Using this criterion, the outcome was related to the size of mass lesion in 60 patients. No predictive significance was evident (p =0.28). Of the 40 patients who did not demonstrate a focal mass effect, 14 showed obliteration of PMC, six showed CT-visible subarachnoid blood and 10 showed evidence of diffuse axonal injury. Diffuse brain swelling. The presence of diffuse brain swelling did not indicate a poor outcome ( p = O . 1 1 ) . Of the 23 patients who had hypoxia and or hypotension on admission, 21 showed evidence of diffuse brain swelling.
Focal lesions (Fig. 3a) Acute subdural haematoma (acute SDH).
552
Benedict M . Selladurai et al. 14 100-
go
80 -
-
0,
60-
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
a 40-
20-
0interval
Pupillary abnormality
major inluries
interval c6 hours
100-
80 -
-3
-
-
60-
0
E
2u' 40a" 20-
0-
c 2 0 2&50
>50
6 8
3-5
None Unilal Bilat
Absent Present
Absent Presenl
Age in ye.3rs
Glasgow coma
Pupillary
Hypothalamic
Hypoxia or shock
score
abnormality
disturbance
p 002
p 00002
p 00002
pa02
p 002
FIG.1. (a) Comparison of clinical characteristics on admission of patients in this study ( u),with those of a study by Van Good recovery; 0, Dongen et al.' (B). (b) Outcome of patients in this study in relation to clinical characteristics. 0, moderate disability; UU, severe disability; vegetative; B, death.
m,
Acute SDH proved to be the focal lesion with the greatest potential for a poor outcome (p=0.0062). A high incidence of co-existing lesions were evident in the 17 patients with acute SDH (haemorrhagic contusions/intracerebral haematomata in nine patients, diffuse axonal injury in four patients). All 15 patients with acute SDH and a poor outcome had evidence of PMC obliteration, nine had unilateral hemisphere swelling. T h e interval from injury to surgery was more than 4 h in 11 patients. Extradural haematoma (EDH). Of the eight
patients with EDH who had a poor outcome, four had associated intrinsic haemorrhagic lesions, and one had evidence of diffuse axonal injury. Two patients who presented in a poor neurological state after rapid deterioration showed extradural haematomas of heterogenous density on CT suggesting ongoing bleeding. Haemorrhagic contusion/intracerebral haematoma. T h e outcome was only slightly better in patients with a single intrinsic haemorrhagic lesion as compared with those with multiple such lesions. There was a significant incidence
Outcome prediction in severe head injury
(32)
(77)
n n
-
80
(78)
553
(31)
nn
-
v)
I
C
ar
+
60-
+
0
Iu
m
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
2 40-
$ a 20
-
0-
Patent Partial ComDlete Obliteration
Absent Present
State of perimessencephalic
CT visible
Cisterns
Subarachnoid blood
p=o 0001
p-0 0002
II I I
I t f
II II I
4 m r n 5-10mm >lOrnm
.,
Diffuse axonal injury p=O 0017
nn n
(b)
80-H
Absent Present
I
I
I 1
I
Absent
I
It
I
I
Absent Present
Midline shift
Mass effect
Diffuse brain swelling
p=o 11
p p 0 16
p=0.39
FIG.2. (a, b) Outcome o f patients in this study in relation t o non-focal pathological changes on i n i t i a l CT scan. Outcome of patients in this study in relation t o clinical characteristics. 0, G o o d recovery; 0, moderate disability; LII], severe disability;
69, vegetative;
death.
of associated lesions in the 22 patients with single intrinsic haemorrhagic lesions who had a poor outcome (diffuse axonal injury 4, acute SDH 4 and EDH 3).
Normal CT scans. A good outcome was evident in six out of the eight (75%) patients who had normal C T scans. Both patients who had a poor outcome with a normal CT scan were of
554
Benedict M. Selladurai et al.
-
80
(0 +
p
60
0
-
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
0
al
5c
40
8
ti
20
0
Extra dural haematoma
Single ICHiContusion
Multiple ICHiContusion
Acute suhdural haey?toma ’*
i
100
-
80
u) +
a,
60
0
c 0 al
cn
2
plOmm D=o
003
.,
FIG.3. (a) Outcome in relation to type of focal lesion seen an initial CT scan. (b) Outcome in relation to three groups of patients: (1) those with subarachnoid blood, cistern obliteration and midline shift >10 mm; (2) those with all other abnormalities; (3) those with normal CT scans. Outcome of patients in this study in relation to clinical characteristics. 0, Good recovery; J,moderate disability; UO, severe disability; B, vegetative; death.
the GCS 3-5 range with one patient having hypoxia and hypotension on admission. The outcome was then considered for three
categories of patients: (1) those with PMC obliteration and/or CT-visible subarachnoid blood and/or a midline shift of >10 mm: (2)
Outcome prediction in severe head injury
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
those with all other abnormalities and ( 3 ) those with normal C T scans (Fig. 3b). Patients in category (1) had a significant risk of a poor outcome (p=0.0003). There was no significant difference in mortality between categories ( 2 ) and ( 3 ) , although there was a higher incidence of functional incapacitation in the former category.
Prediction of outcome The principles used for prediction of outcome were based on the prediction tree model of Choi et al.,'O where the patient population was divided into smaller subsets using recursive splitting, based on critical levels of different prognostic variables. However, a tree structure was not compiled for the entire patient population in this study in view of the rather small sample size. Instead the criteria of age, GCS and state of PMC were used to identify two subsets of patients for prediction of good outcome and poor outcome, respectively: (1 ) the subset aged t 2 0 years, with GCS 6-8 and without obliteration of MCP ( n = 2 1 ) , for whom a good outcome was predicted; ( 2 ) the subset aged >20 years, with GCS 3-5 and with partial or complete obliteration of PMC ( n = 2 8 ) , for whom a prediction of poor outcome was made. On comparison with actuarial outcome, correct predictions were 71.4% for category (1) and 89.3% for category ( 2 ) , respectively. Discussion
A major goal of current studies in severe head injury has been to define criteria that are of value in predicting outcome. Prediction of outcome based on early clinical and C T findings has its drawbacks. The pathology of severe head injury is heterogeneous and its outcome depends on a complex interaction of a multitude of factors, some of which may not be in evidence in early stages of injury. Rapid transfer by trained personnel and high-quality intensive care are hallmarks of care for severe head injury in developed countries. In a developing country such facilities are either
555
lacking or are of limited availability. There was no significant difference in the clinical characteristics of the patient population in this study, when compared with that in a similar study performed in a developed country5 except that there was a significant delay between injury and the initial C T in this study. Not withstanding the above-mentioned limitations, early prediction in severe head injury is even more important in a developing country to make optimal use of limited resources available. The predictive power of established clinical criteria of age, GCS, and state of pupils was well illustrated in this study. In addition, evidence of a hypothalamic disturbance, hypoxia and hypotension on admission also indicated an adverse outcome. The role of C T features in prediction of outcome in severe head injury has been extensively investigated in previous studies.2-7~9~11-19 At least three previous s t u d i e ~ ~ * ~ - ~ conclusively demonstrated the predictive potential of PMC obliteration in severe head injury. In our study this feature showed a significant association with poor outcome. Toutant et al.' demonstrated the value of PMC in the first few hours after injury, especially in prediction of potentially lethal events in patients in the higher GCS range of 6-8. Our experience demonstrated the ability of this feature to identify patients in the GCS 6-8 range, in whom urgent evacuation of intracranial mass lesion was warranted. The study of the National Institutes of Health Trauma Coma Data Bank9 (NIH study), was the first large CT-based study to conclusively demonstrate the prognostic value of CT-visible subarachnoid blood, which was evident in 40% of patients in that study. In our patient population, this feature was evident in 29% of patients and was definitely correlated with a poor outcome. In our experience, patients who had evidence of diffuse axonal injury (DAI) in association with PMC obliteration, had a uniformly poor outcome, a feature also highlighted in previous studi e ~ . ' ~ DAI . ' ~ also played a role in the poor outcome of some patients in this study, after surgical evacuation of mass lesions. The NIH
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
556
Benedict M. Selladurai et al.
study9 demonstrated midline shift to be a strong predictor of poor outcome. In our experience, a majority of patients with a midline shift > l O mm had a poor outcome. However, midline shift did not prove to be of statistical significance for the entire patient population. Clifton et a1.2 and Ross et al.” demonstrated the presence of smaller degrees of midline shift in patients with diffuse brain injury, who had a poor outcome. Similarly, in our study a significant proportion of patients with midline shift (10 mm had evidence of diffuse axonal injury and bilateral haemorrhagic lesions. An association between presence of a focal mass effect and poor outcome has been demonstrated in some previous stud~ ~ s . ~ JHowever, *J’ the NIH study’ and a study by Tabbador et a1.I4 did not demonstrate such a correlation, as has also been our experience. In this study, a significant incidence of predictors of poor outcome such as PMC obliteration, CT-visible subarachnoid blood diffuse axonal injury were evident in patients without focal mass effect. In addition, the mitigation of effects of mass lesions by early surgical removal may also account for the diminished prognostic value. The NIH study9 demonstrated an association between hypoxia, hypotension and diffuse brain swelling, a feature also shown in our study. Choi et al.lo used the prediction tree technique to identify subsets of patients with severe head injury, using critical levels of several prognostic criteria. This provided for minimal variability within a given subset with consideration given to interaction of different prognostic criteria. Furthermore, this methodology was more easily applicable than the sophisticated statistical models such as logistic regression and discriminant analysis. Hence we considered it more suitable for use in conditions such as exist in Malaysia. In view of the relatively small size of the patient population in this study a complete prediction tree was not compiled. Instead we restricted our predictions to two subsets of patients selected using critical levels of the most significant prognostic criteria of age, GCS and state of PMC. The mediction rates achieved for the two subsets of
patients are similar to those Choi et al.’” achieved for a much larger patient population managed under more sophisticated conditions. In the final analysis, it has to be conceded that prediction of outcome in severe head injury will remain a difficult and subjective exercise. It can only supplement and not replace clinical judgement.
Conclusions A study of outcome in 109 patients with severe head injury correlated to clinical and C T criteria on admission indicates that: (1) age, GCS and state of pupils remain the predominant clinical indicators of outcome; (2) obliteration of perimesencephalic cisterns, subarachnoid haemorrhage, diffuse axonal injury and acute subdural haematoma are CT features strongly related to a poor outcome; (3) presence of diffuse brain injury could influence prognostic value of midline shift and focal mass effect; (4) modest success was achieved with use of the prediction tree technique for two subsets of patients selected using criteria of age, GCS and state of PMC. Address for correspondence: Benedict M. Selladurai, Department of Surgery, School of Medical Sciences, 15990 Kota Bharu, Kelantan, Malaysia.
References Jennett B, Teasdale G, Braakman R, et al. Predicting outcome in individual patients after severe head injury. Lancet 1976; 1:1031-4. Clifton GL, Grossman RG, Makela ME, et al. Neurological course and correlated computer tomography findings after severe head injury. J Neurosurg 1980, 52:611-24. Narayan RK, Greenberg RP, Miller JD, et al. Improved confidence of outcome prediction in severe head injury. A comparative analysis of clinical examination, multimodality evoked potentials, CT scanning and intracranial pressure. J Neurosurg 1981; 54:751-62. Genarelli TA, Speilman GM, Langfitt TW, et al. Influence of type of intracranial lesion on outcome of severe head injury. J Neurosurg 1982; 56:26-32. Van Dongen KJ, Braakman R, Gelpke GJ. The prognostic value of computerized tomography in
Outcome prediction in severe head injury
6
7
Br J Neurosurg Downloaded from informahealthcare.com by Chulalongkorn University on 01/07/15 For personal use only.
8 9
10
11 12
comatose head-injured patients. J Neurosurg 1983; 59~951-7. Lobato RD, Cordobes F, Rivas JJ, et al. Outcome of severe head injury related to the type of intracranial lesion-a computerized tomography study. J Neurosurg 1983; 59:762-74. Toutant SM, Klauber MR, Marshall LF, et al. Absent or compressed basal cisterns on first C T scan: ominous predictors of outcome in severe head injury. J Neurosurg 1984; 61591-4. Choi SC, Narayan RK, Anderson RL. Enhanced specificity of prognosis in severe head injury. J Neurosurg 1988; 69:381-5. Eisenberg HM, Gary Jr HE, Aldrich EF, et al. Initial C T findings in 753 patients with severe head injury-a report from the NIH Traumatic Coma Data Bank. J Neurosurg 1990; 73:688-98. Choi SC, Muizelaar IP, Barnes T Y , et al. Prediction tree for severely head-injured patients. J Neurosurg 1991; 75~251-5. Jennet B, Bond M. Assessment of outcome after severe brain damage. A practical scale. Lancet 1975; 1:480-4. Lipper MH, Kishore PRS, Enas GG, et al. Computed tomography in the prediction of outcome in head injury. AIR 1985; 144:483-6.
557
13 Ross DA, Olsen WL, Ross AM, et al. Brain shift, level of consciousness and restoration of consciousness in patients with acute intracranial haematoma. J Neurosurg 1989; 71:498-502. 14 Tabbador K, Danziger A, Wisoff HS. Estimation of intracranial pressure by C T scan in closed head trauma. Surg Neurol 1982; 18:212-15. 15 Kishore PRS, Lipper MH, Becker DP, et al. Significance of C T in head injury: correlation with intracranial pressure. AJNR 1981; 2:307-11. 16 Ropper AH. Lateral displacement of brain and level of consciousness in patients with acute hemispherical mass. N Engl J Med 1986; 314:953-8. 17 Miller JD, Gudeman SK, Kishore PRS, et al. C T scan, ICP and early neurological evaluation in prognosis of severe head injury. Acta Neurochir (Wein) 1979; 28:86-8. 18 Genarelli TA, Thibault LE, Adams JH, et al. Diffuse axonal injury and traumatic coma in the primate. Ann Neurol 1982; 12:564-74. 19 Levi L, Guilburd JN, Lemberger A, et al. Diffuse axonal injury: analysis of 100 patient with radiological signs. Neurosurgery 1990; 27:429-32.
