- Email: [email protected]
The Westmead head injury project: Outcome prediction in acute subdural haematoma
1 [~-.1 I ~ - ~ t ~ l ~" .....
- ...........-- ......... ......
-. . . . . . . . . . . . . . . . . .
The Westmead head injury project: outcome prediction in acute subdural haematoma
R. J. C o o k MB BS FRACS * t M. R. F e a r n s i d e MS FRACS* P. M c D o u g a l l , RN* R. J. M c N e i l MSc* tDepartment o£Neurosurgery, Westmead Hospital, Westmead NSW *Department of Neurosurgery, Sir Charles Gairdner Hospital, Perth WA,Australia
A prospective two year study o f a consecutive sample o f patients with an acute subdural haematoma who were admitted to Westmead Hospital, New South Wales, Australia was undertaken. There were 103 patients with an acute subdural haematoma admitted in the period. Twenty-four o f these scored 9 or greater on the Glasgow Coma Scale (GCS) and of these all made a functional recovery, i.e. Glasgow Outcome Scale (GOS 1 or 2). The remaining 79 patients scored 8 or less o n admission and of these 30% made a functional recovery. O f the 70% remaining, 4% were moderately or severely disabled (GOS 3 or 4) while 66% died (GOS 5). Age, hypoxia, hypotension, response to intracranial pressure control and two CT scan features, midline shift as measured from the septum peUucidum and cerebral oedema, were all significant in predicting outcome. Time from injury to treatment, initial pupil response, lucid interval and compression o f brainstem cisterns on CT scans statistically failed to predict outcome. The data were analysed using logistic regression which showed age and midline shift to predict death or disability with an accuracy o f 80% at twelve months after the injury (sensitivity 58%, specificity 89%). Journal of Clinical Neuroscience 1996 3 (2): 143-148
© 1996 Pearson Professional Ltd
Keywords: Acute subdural haematoma, Prediction of outcome, Head injury
Introduction Acute subdural h a e m a t o m a remains a significant cause of death and disability in the community. The aim of this prospective study was to analyse which clinical and CT parameters were significant in predicting outcome. All parameters which might affect o u t c o m e were subjected to logistic regression analysis to construct a m o d e l of prediction. This model was then applied prospectively to test its accuracy. Data collection was from Westmead Hospital which is a designated neurosurgical and trauma centre attached to the University of Sydney, Australia. The prospective application was a c o h o r t of patients studied at the Sir Charles Gairdner Hospital, Perth, Australia.
Clinical m a t e r i a l a n d m e t h o d s ...... The definition of an acute subdural h a e m a t o m a (ASDH) is n o t consistent in the literature. We defined patients presenting to Westmead either primarily or secondarily within 24 h of their head injury with a subdural h a e m a t o m a as acute. O t h e r authors a,z have used presentation within three days for acute and three to 24 days for subacute as their definition. The data was analysed using a protocol based on the US trauma c o m a databank. ~ Clinical variables were
mechanism of injury, age, sex, hypoxia (defined as an arterial PaO2 of <60 m m H g ) , hypotension (defined as a systolic blood pressure <90 m m H g systolic) and time from injury to treatment. The Glasgow C o m a Score was the first recorded score on arrival at Westmead prior to resuscitation. Initial pupillary responses were graded as n o r m a l if both pupils r e s p o n d e d to light; an abnormal response was scored if one or both pupils failed to respond to light. Mean intracranial pressure level and response to intracranial pressure control were also analysed. CT variables analysed included the presence of cerebral o e d e m a which was assessed by a n e u r o s u r g e o n and neuroradiologist. Cerebral o e d e m a was present if any three of the following four factors were present: compression o f brain stem cisterns, loss of grey/white differentiation, compression of the ventricular system and obliteration of the sulcal/gyral pattern. The a m o u n t of midline shift was measured from the septum pellucidum and four categories were used: category 0 for no shift, category 1 for <5 m m midline shift, category 2 for 5-15 m m midline shift and category 3 for >15 m m o f midline shift. O t h e r CT parameters included the presence of subarachnoid blood, intraventricular haemorrhage, other intracranial h a e m a t o m a and the presence and size of an
J. Clin. Neuroscience Volume 3 Number 2 April 1996
143
Clinical studies
Acute subdural haematoma
associated brain contusion. A contusion was defined as a hyperdense area less than 1 cm diameter and a haematoma if the diameter exceeded 1 cm. O u t c o m e was assessed at one year after the head injury using the Glasgow O u t c o m e Scale. 4 GOS 1 and 2 were grouped together and t e r m e d functional recovery (FR), while GOS 3-5 were g r o u p e d together and t e r m e d death or disability (DD). Patients presenting in coma were managed according to a standard protocol with intensive care admission, intubafion, ventilation, narcotic analgesia and muscle relaxation. Anficonvulsants were used prophylactically in all patients. Initial assessm e n t was p e r f o r m e d by a multi-disciplinary trauma response team. A C T scan was p e r f o r m e d on all patients when cardiovascular stability was achieved and within 4 h of admission. ASDHs causing midline shift of the septum pellucidum of >5 m m were evacuated at craniotomy. Surgical preference dictated whether the dura was left open or closed and whether the b o n e flap was replaced. T h e majority o f patients had their intracranial pressure (ICP) m o n i t o r e d either post-operatively or electively if the h a e m a t o m a did not require surgical drainage. ICP was managed in the post-operative group via a subdural catheter inserted at operation. In the patients not undergoing craniotomy a variety of methods were employed to monitor ICP including subdural catheter, ventricular catheter or a Richmond bolt. ICP was measured continuously and a mean ICP of <20 m m H g was considered as normal. W h e n the mean ICP was >20 m m H g for 15 min continuously the ICP was considered as raised. Raised ICP was managed with hypervenfilation, mannitol, intravenous p h e n o b a r b i t o n e as a bolus injection and refractory cases were treated with a p h e n o b a r b i t o n e infusion. Raised ICP requiring treatment was assessed by a three tier response to that treatment. A good response o c c u r r e d when the ICP r e t u r n e d from an elevated value to normal (<20 m m H g ) and remained there for at least 4 h. A partial response was defined as an ICP which decreased or r e t u r n e d to normal for <4 h. Nonresponsive ICP occurred when there was no change or an increase folowing treatment. Duration o f ventilation was d e t e r m i n e d by clinical response and ICP levels. Statistical analysis was p e r f o r m e d to assess which of the variables were significant in predicting final outcome. The logistic regression analysis was p e r f o r m e d using SPIDA. 5 Students' t-test was used for comparisons of means. Chi-squared tests were used for proportions. A logistic regression model was used to predict the probability of a subject having a death/disability outcome with given characteristics. The model selection enabled the identification o f the characteristics which appeared to be the best predictors o f outcome. T h e probability of death or disability was estimated for 12 months post-injury. For each patient a prognostic index (S) was based on the significant variables (ZK) which can be obtained and expressed as:
These are estimated by maximum likelihood using the LREG p r o c e d u r e of SPIDA. After S is obtained, the probability of death/disability (PD) for a given patient is indicated as: 1
P D - - 1 + e -s To test the classification accuracy of the predictive model, the probability scores are assigned to FR or DD categories at 99 probability partition values ranging from 1-99%. Thus a 2 x 2 contingency table of observed versus predicted outcomes was obtained for each probability partition value, and the corresponding sensitivity and specificity calculated. A partition value was then selected which gave the best overall accuracy o f predicted numbers of DD outcomes versus observed DD outcomes. The predicted n u m b e r of deaths/disability in each group was then estimated and c o m p a r e d with the observed number.
Results
The 103 patients who were entered in the study were divided into two groups. Group 1 comprised 79 patients with an initial GCS of 8 or less and Group 2 comprised 24 who had a GCS o f 9-15 at presentation. In this latter group 15 were male and 9 were female. All patients achieved a GOS of 1 or 2 at 6 months (22 with 1 and two with 2). Ten patients were treated surgically with a craniotomy and removal of the ASDH, while 14 were treated non-surgically with neurological observation, fluid restriction and osmotic agents. Eighteen patients were discharged from hospital within 14 days. The remaining 79 patients with a severe head injury and who were in coma were analysed in detail and all tables refer to this group. Mechanisms of injury
O f the 79 patients in Group 1, 35 were primarily admitted to Westmead Hospital, while 44 were transferred from surrounding base hospitals for neurosurgical care. The majority of patients were involved in m o t o r vehicle collisions, 34/79 (43%), suggesting that most of the haematomas were secondary to high energy impacts (Table 1).
Table 1 Mechanism of Injury
S = B 0 + B1Z a + B2Z ~ + ....... + BKZ K
where Z1, Z~ and Z~ are the predictive variables, and B 0, B 1 and B~ are the corresponding regression coefficients.
144
J. Clin. Neuroscience Volume 3 Number 2 April 1996
No of patients MVA Fall Pedestrian MBA Bicycle Assault Horse Other
34 8 5 5 5 3 3 16
Acute subdural haematoma
Clinical studies
Time from injury to treatment and neurosurgical intervention
Table 2 Outcomes of 74 Group 1 patients Age (Years)
Functional Recovery
Death/ Disability
0- 9 10-19 20 - 29 30 - 39 40 - 49 50 - 59 60 - 69 70 - 79 80 - 89
1 8 9 3 1 2 0 0 0
4 5 14 10 3 2 6 9 2
Time from injury to treatment at Westmead was assessed using two groups: group A's time from injury to treatm e n t was less than four hours and group B was greater than 4 h. 33% of patients were functional survivors in the <4 h group, while 21% o f patients were functional survivors in the >4 h group. T h e r e was no statistical difference between these groups.
Pupillary response Age and sex distribution Fifty-five patients (70%) were male, while 24 were female. The ages ranged from 2 to 89 years (mean 37 years). As age advanced o u t c o m e became worse with no functional survival after age 60 (Table 2).
Hypoxia / Hypotension Nine patients were confirmed hypoxic on admission via arterial blood gas. It was suspected (ambulance reports, etc) that a further 24 patients may have had an hypoxic episode prior to their admission to Westmead. These two groups were c o m b i n e d and o u t c o m e was assessed in terms of FR and DD. T h e presence of hypoxia as defined above was significant in predicting o u t c o m e (P < 0.005, Chi-squared test for i n d e p e n d e n c e ) . Hypotension was confirmed in 20 patients prior to resuscitation. T h e presence of hypotension was significant in predicting a DD o u t c o m e (P < 0.01, Chi-squared test for i n d e p e n d e n c e ) (Tables 3 and 4).
Lucid interval
21% of patients who had an abnormal pupillary response achieved a functional recovery c o m p a r e d with 41% o f patients who had equal and reactive pupils. This result is not statistically significant.
CT scan analysis (i) Midline shift (Table 5) shows the a m o u n t of midline shift as measured from the septum pellucidum. With increasing midline shift the o u t c o m e became worse. All 12 patients with >15 m m o f shift either died or were severely disabled. Midline shift was significant in determining outcome (P < 0.013, Chisquared test for i n d e p e n d e n c e ) . (ii) Cerebral o e d e m a (Table 6) occurred in 43 of 79 patients and was significant in predicting o u t c o m e (P< 0.013, Chi-squared test for i n d e p e n d e n c e ) . (iii) O t h e r CT parameters. The presence of brain stem cistern compression, site of haematoma, subarachnoid haemorrhage, intraventricular haemorrhage, intracerebral haemorrhage, extradural haemorrhage or brain contusions did not reach statistical significance.
A lucid interval was identified in 12 patients, however it did not help to differentiate outcome.
Table 5 Shift of midline on CT None
Table 3 The influence of hypoxia on o u t c o m e Hypoxia Present Absent
Functional Recovery
Death/ Disability
4 20
29 26
Functional Recovery
Death/ Disability
6 10 8 0
4 12 27 12
None < 5 mm 5-15 mm > 15 mm
P < 0.005 (Chi-square)
P < 0.005 (Chi-square)
Table 4 The influence of hypotension on outcome Hypotension
Functional Recovery
Death/ Disability
Present Absent
2 22
18 37
Table 6 Cerebral oedema on CT Functional Recovery
Death/ Disability
8 16
35 20
Present Absent
P < 0.01 (Chi-square)
P < 0.013 (Chi-square)
J. Clin, Neuroscience
Volume 3
Number
2 April 1996
145
Clinical studies
Acute subdural haematoma
I n t r a c r a n i a l pressure
1.o
ICP was m o n i t o r e d in 56 patients. T h e trend was as m e a n ICP rose for the o u t c o m e to b e c o m e worse, but this was n o t statistically significant.
~.m "~ o.8tD
r-Response
to treatment
of raised ICP
~ 0.6-
T h e three groups of response to t r e a t m e n t o f raised ICP did differentiate the o u t c o m e groups FR and DD (P < 0.05, Chi-squared test for i n d e p e n d e n c e ) . All eight patients who showed no response to persistently raised ICP achieved a DD o u t c o m e while a g o o d response to t r e a t m e n t (36 patients) d e m o n s t r a t e d a 61% DD outcome. Barbiturate c o m a was instituted in ten cases to control ICP and four o f these patients m a d e a functional recovery (Table 7).
rn 0.5O 0.4-
s~ o A-
0-2-
0
i
I
3
4
5
6
7
I 3
I 4
I 5
I 6
7
I
-5-5-4-3-2-1
0
1
2
S Score Fig. 1 Percentage chance of d e a t h or disability.
T a b l e 7 Response t o t r e a t m e n t of raised ICP
Functional Recovery
No response Partial response Good response
0 7 14
Death/ Disability
8 5 22
P < 0.05 (Chi-square)
1.0 t~
0.8-
C3 t-
O.6-
£3 ~_ 0 . 5 -
Prediction of o u t c o m e
0
i
>~ 0 . 4 -
..--
Age, sex, time f r o m injury to t r e a t m e n t at a neurosurgical centre, pupillary response, hypoxia, hypotension, brain stem cistern compression, shift size and cerebral o e d e m a were f u r t h e r analysed using logistic regression against the two o u t c o m e categories FR a n d DD. Using a backward elimination selection p r o c e d u r e the variables which comprised the final m o d e l were age and shift. T h e m o d e l comprises an ASDH score of "S" where S = - 2.481 + 0.044 x age + 1.284 x shift category. Figure 1 is a n o m o g r a m curve of"S" versus probability of death or disability. T h e above m o d e l was most accurate w h e n using a partition value of 0.5 and this resulted in an overall accuracy of 80% with sensitivity of 58% and specificity of 89%. An example o f the use of this m o d e l would be a patient with an ASDH aged 40 years with a shift of 10 m m (categ o r y 2). T h e "S" score w h e n calculated is 1.85 (- 2.481 + 40 x 0.044 + 2 x 1.284). T h e c o r r e s p o n d i n g percentage chance of death or disability is therefore 0.86 f r o m the n o m o g r a m curve (Fig. 1). At the completion o f this study the above m o d e l was tested on 22 consecutive patients with an ASDH GCS8 or less, following h e a d injury who were admitted to Westmead Hospital and to Sir Charles Gairdner and the Royal Perth Hospitals. T h e y constituted a similar sample in respect to age, sex, m e c h a n i s m of injury a n d clinical features to those analysed above. T h e o u t c o m e of these patients was assessed at six months. Figure 2 is a n o m o g r a m curve with the "S" score for the 22 patients indicated by a dot. Two patients were predicted inaccurately by the m o d e l (cross dot). These patients were predicted
146
i
.Q -~ 0.2--
0
I
~
-6
-5
-4
-3
-2
-1
0
I 1
I 2
S Score Fig. 2 O u t c o m e assessed at six months.
by the m o d e l to make a DD o u t c o m e at six months, however the actual o u t c o m e was an FR. In this small prospective test of the m o d e l it was accurate at predicting 20 of 22 patients, i.e. 91%.
Discussion T h e mortality rate of ASDH is high, 6-15 in a majority of series ranging between 48% 7 a n d 75%. 11 O u r series mortality for the 103 patients was 50%; for the 79 patients in c o m a (GCS 8 or less) it was 66%. Only 4% of the patients presenting in c o m a r e m a i n e d severely disabled while 30% were functional. This implies that m o d e r n managem e n t of ASDH does not p r o d u c e large n u m b e r s of severely disabled p e o p l e as a secondary effect of reducing the n u m b e r o f fatalities. Stone et a115 r e p o r t 27% of his 128 patients as functional survivors, a similar percentage to o u r series, i For those patients with a severe h e a d injury (GCS 8 or less) age was a significant d e t e r m i n a n t of o u t c o m e with
J. Clin. Neuroscience Volume 3 Number 2 April 1996
Acute subdural haematoma
Clinical studies
no functional survival greater than 60 years. Howard et a116 analysed age in relation to ASDH; they concluded that 66% of y o u n g patients (18-40 years) m a d e a functional recovery c o m p a r e d with a 9% functional recovery in the elderly (age > 65 years). T h e m e c h a n i s m by which age has such an effect on o u t c o m e is unknown, 17but suggestions include a p o o r regenerative capacity of the older brain and a predisposition to develop a m o r e lethal injury, x6 In the m o d e l predicting o u t c o m e by Klun, TM age was a factor with a sharp rise in mortality for age > 50 years. O u r predictive m o d e l used age as o n e of the key variables in d e t e r m i n i n g outcome. Hypoxia and hypotension were significant factors in d e t e r m i n i n g outcome. T h e diagnosis of hypoxia implies a p r i m a r y (hypoventilation secondary to brain stem injury) or secondary event (airway or chest injury). Whatever the underlying cause the prognosis was p o o r with only four functional recoveries in a g r o u p of 33 patients (12%). Hypotension similarly caused a significant grave prognosis with only 2 of our 20 patients in this g r o u p making a functional recovery. This factor was rated by Stening et a119 as being a significant factor in the o u t c o m e o f a review of ASDH in New South Wales, Australia, in the 1970s. T h e larger the midline shift, as m e a s u r e d f r o m midline to s e p t u m pellucidum, the worse the outcome. T h e degree of shift is d e t e r m i n e d by the volume of the ASDH plus the swelling of the underlying h e m i s p h e r e . It has b e e n suggested that the h a e m a t o m a itself is an epiphen o m e n o n with the underlying brain injury d e t e r m i n i n g outcome. 2° T h e m e a s u r e m e n t o f brain shift encompasses b o t h of these factors. Marshall et al 2 r e p o r t e d that a greater midline shift was associated with a worse recovery. Ross et a122 d e m o n s t r a t e d that a trend towards high ICP was seen with a shift > 5 m m . Becker et a123 concluded that the mortality in patients with 10 m m or greater midline shift was 53% a n d this c o m p a r e d with 25% for those patients with less than 10 m m o f shift. We f o u n d that midline shift was significant when d e t e r m i n i n g o u t c o m e a n d was such a strong predictor that it was included in the m o d e l for prediction. Cerebral o e d e m a was significant in d e t e r m i n i n g outc o m e (P< 0.013, Chi-squared test for i n d e p e n d e n c e ) . O f the four factors used to assess cerebral oedema, individually they were n o t significant in predicting o u t c o m e but when used in c o m b i n a t i o n they achieved a significance (any 3 of the 4). ICP was studied in two ways. Firstly, the m e a n ICP category in which the patient spent most of the postoperative time was c o m p a r e d with outcome. As ICP rose patients t e n d e d to have worse outcomes, b u t the numbers were too small to achieve statistical significance. We also graded the patients as to their response to ICP control; this proved to be significant in d e t e r m i n i n g outcome. Miller et a122 h a d similar ICP response categories in their review of the significance of ICP m o n i t o r i n g in h e a d injury (including but not limited to ASDH) a n d in their n o n - r e s p o n d i n g group; as with ours, all the patients died or were severely disabled. Seelig et a124 concluded that o u t c o m e was substantially influenced by ICP; 79% of
their patients who had a functional recovery had a postoperative ICP of less than 20 m m H g . Seelig et a124 in their study concluded that a delay f r o m injury to operation was the factor of greatest therapeutic i m p o r t a n c e in ASDH. Patients who u n d e r w e n t surgery in the first four hours h a d a 30% mortality comp a r e d with 90% in those who h a d surgery after four hours. O u r study did not confirm this finding, with 66% of patients achieving a DD o u t c o m e if t r e a t m e n t was instituted within four hours of injury, and 79% achieving a DD o u t c o m e if treated m o r e than four hours following injury. This difference was not statistically significant. Klun et all8 r e p o r t e d 330 cases o f ASDH and failed to achieve lower mortality by early operation a n d similarly Wilberger et al25 c o n f i r m e d the absolute timing for the removal of an ASDH to be u n i m p o r t a n t in terms of this four h o u r surgical window. T h e two categories we used for pupil response failed to help us predict o u t c o m e with 21% of patients achieving an FR with pupillary signs on admission c o m p a r e d with 40% achieving an FR with equal a n d reactive pupils. T h e pupillary response r e c o r d e d was at the initial examination at Westmead and a p e r c e n t a g e of patients may have c h a n g e d their initial pupillary signs once adequately resuscitated. T h e literature has widely s u p p o r t e d pupillary change as being significant in d e t e r m i n i n g outcome. J a m i e s o n et al's study 26 o f 553 ASDH patients gave a mortality rate of 41% for o n e fixed pupil and 85% for two fixed pupils. Similarly Stening et al's 19 mortality for fLxed dilated pupils was 93% as o p p o s e d to 58% for normally reactive pupils. Lucid interval did not influence o u t c o m e for FR a n d DD and this result c o m p a r e s favourably with Stening et aP 9 who also f o u n d no relationship.
Conclusions Based on logistic regression analysis o f factors which may have influenced o u t c o m e in ASDH, two factors - age and shift of the septum pellucidum - were used to construct a m o d e l to predict the probability of achieving a DD outcome. T h e accuracy of this m o d e l was 80% and it successfully predicted the o u t c o m e o f a f u r t h e r 20 out o f 22 patients (91%) when used prospectively. Attention is drawn to the factors of hypoxia and hypotension which are often easily correctable in the m a n a g e m e n t of a patient with an ASDH. An elderly p a t e n t (> 60 years) in c o m a with a traumatic ASDH, irrespective of the treatment, is m o s t likely to be either d e a d or severely disabled at 12 m o n t h s post-injury. Received 11 January 1994 Accepted for publication 19 May 1995
Correspondence and offprint requests: R.J. Cook Northshore Medical Centre 66 Pacific Highway St Leonards NSW 2065, Australia Phone: (02) 43%4125 Fax: (02) 437-4147
J. Clin. Neuroscience Volume 3 Number 2 April 1996
147
Clinical studies References 1. Jones NR, Blumbergs PC, NorthJB. Acute subdural haematomas: aetiology, pathology and outcome. Aust NZ J Surg 1986; 56:907-915. 2. Marshall LF, Toole BM, Bowers SA. The National Traumatic Coma Data Bank, Part 2. Patients who talk and deteriorate: implications for treatment. J Neurosurg 1983; 59:285-288. 3. The Traumatic Coma Data Bank: Design, methods and baseline characteristics.J Neurosurg 1991; 75: (Suppl. S8-S13). 4. Jennett B, Bond M. Assessment of outcome after severe brain damage. A practical scale. Lancet 1975; 1:480-484. 5. Hamey L, Leung O, McNeil D. SPIDA (Statistical Package for Interactive Data Analysis), Users manual. MacQuarie Press 1985; Sydney, Australia. 6. GalbraithJG. Subdural haematoma- acute and chronic. A reappraisal. Clin Neurosurg 1982; 29:24-31. 7. Fell DA, Fitzgerald MD, Moiel RH, et al. Acute subdural haematoma - review of 144 cases.J Neurosurg 1975; 42: 3742. 8. Shenkin HA. Acute subdural haematoma. J Neurosurg 1982; 57:254-257. 9. Yui CP, Mann KS, Ong GB. Acute subdural haematoma Better results with neurosurgeons than general surgeons. Injury 1983; 14:489-492. 10. Haselberger K, Pucher R, Aver LM. Prognosis after acute subdural and epidural haemorrhage. Acta Neurochir 1988; 90:111-116. 11. Richards T, HoffJ. Factors affecting survival from acute subdural haematoma. Surgery 1974; 75:253-258. 12. Shigemori M, Syojimi K, Nakayama K, et al. The outcome from acute subdural haematoma following decompressive hemi-craniectomy. Acta Neurochir. 1980; 54:61-69. 13. HernesniemiJ. Outcome following acute subdural haematoma. Acta Neurochir 1970; 49:191-198. 14. Talalla IA, Morin MA. Acute traumatic subdural haematoma: A review of one hundred consecutive cases J Trauma 1971; 11:771-777.
148
Acute subdural haematoma 15. StoneJL, Lowe RJ, Jonasson o. Acute subdural haematoma: Direct admission to a trauma centre yields improved results. J Trauma 1986; 26:445-450. 16. Howard MA III, Gross AS, Daay RG, et al. Acute subdural haematomas: An age dependent clinical entity. J Neurosurg 1989; 71:858-863. 17. HernesniemiJ. Outcome following head injuries in the aged. Acta Neurochir 1979; 49:6%69. 18. Klun B, Fettich M. Factors influencing the outcome in acute subdural haematoma. Acta Neurochir 1984; 71: 171-178. 19. Stening WA, Berry N, Dan NG, et al. Experience with acute subdural haematomas in New South Wales. Aust NZ J Surg 1986; 56:549-556. 20. Casey ATH, Cook RJ, Fearnside MR, et al. Acute subdural haematoma- Predictors of outcome in 214 patients. Neurology Neurosurgery Psychiatry 1993; 56:222-225. Abstract 120th Proceedings SBNS. 21. 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. 22. MillerJD, Becker DP, WardJD, et al. Significance of intracranial hypertension in severe head injury. J Neurosurg 1977; 47:503-516. 23. Becker DP, MillerJD, Ward J, et al. The outcome from severe head injury with early diagnosis and intensive management. J Neurosurg 1977; 47: 491-502. 24. SeeligJM, Becker DP, MillerJD, et al. Traumatic acute subdural haematoma. Major mortality reduction in comatose patients treated within four hours. New EnglJ Med 1981; 304:1511-1518. 25. WilbergerJE, Harris M, Diamond DL. Acute subdural haematoma: Morbidity, mortality and operative timing. J Neurosurg 1991; 74:212-218. 26. Jamieson KG, YellandJDN. Surgically treated traumatic subdural haematomas.J Neurosurg 1979; 37:137-149.
J. Clin. Neuroscience Volume 3 Number 2 April 1996
- ...........-- ......... ......
-. . . . . . . . . . . . . . . . . .
The Westmead head injury project: outcome prediction in acute subdural haematoma
R. J. C o o k MB BS FRACS * t M. R. F e a r n s i d e MS FRACS* P. M c D o u g a l l , RN* R. J. M c N e i l MSc* tDepartment o£Neurosurgery, Westmead Hospital, Westmead NSW *Department of Neurosurgery, Sir Charles Gairdner Hospital, Perth WA,Australia
A prospective two year study o f a consecutive sample o f patients with an acute subdural haematoma who were admitted to Westmead Hospital, New South Wales, Australia was undertaken. There were 103 patients with an acute subdural haematoma admitted in the period. Twenty-four o f these scored 9 or greater on the Glasgow Coma Scale (GCS) and of these all made a functional recovery, i.e. Glasgow Outcome Scale (GOS 1 or 2). The remaining 79 patients scored 8 or less o n admission and of these 30% made a functional recovery. O f the 70% remaining, 4% were moderately or severely disabled (GOS 3 or 4) while 66% died (GOS 5). Age, hypoxia, hypotension, response to intracranial pressure control and two CT scan features, midline shift as measured from the septum peUucidum and cerebral oedema, were all significant in predicting outcome. Time from injury to treatment, initial pupil response, lucid interval and compression o f brainstem cisterns on CT scans statistically failed to predict outcome. The data were analysed using logistic regression which showed age and midline shift to predict death or disability with an accuracy o f 80% at twelve months after the injury (sensitivity 58%, specificity 89%). Journal of Clinical Neuroscience 1996 3 (2): 143-148
© 1996 Pearson Professional Ltd
Keywords: Acute subdural haematoma, Prediction of outcome, Head injury
Introduction Acute subdural h a e m a t o m a remains a significant cause of death and disability in the community. The aim of this prospective study was to analyse which clinical and CT parameters were significant in predicting outcome. All parameters which might affect o u t c o m e were subjected to logistic regression analysis to construct a m o d e l of prediction. This model was then applied prospectively to test its accuracy. Data collection was from Westmead Hospital which is a designated neurosurgical and trauma centre attached to the University of Sydney, Australia. The prospective application was a c o h o r t of patients studied at the Sir Charles Gairdner Hospital, Perth, Australia.
Clinical m a t e r i a l a n d m e t h o d s ...... The definition of an acute subdural h a e m a t o m a (ASDH) is n o t consistent in the literature. We defined patients presenting to Westmead either primarily or secondarily within 24 h of their head injury with a subdural h a e m a t o m a as acute. O t h e r authors a,z have used presentation within three days for acute and three to 24 days for subacute as their definition. The data was analysed using a protocol based on the US trauma c o m a databank. ~ Clinical variables were
mechanism of injury, age, sex, hypoxia (defined as an arterial PaO2 of <60 m m H g ) , hypotension (defined as a systolic blood pressure <90 m m H g systolic) and time from injury to treatment. The Glasgow C o m a Score was the first recorded score on arrival at Westmead prior to resuscitation. Initial pupillary responses were graded as n o r m a l if both pupils r e s p o n d e d to light; an abnormal response was scored if one or both pupils failed to respond to light. Mean intracranial pressure level and response to intracranial pressure control were also analysed. CT variables analysed included the presence of cerebral o e d e m a which was assessed by a n e u r o s u r g e o n and neuroradiologist. Cerebral o e d e m a was present if any three of the following four factors were present: compression o f brain stem cisterns, loss of grey/white differentiation, compression of the ventricular system and obliteration of the sulcal/gyral pattern. The a m o u n t of midline shift was measured from the septum pellucidum and four categories were used: category 0 for no shift, category 1 for <5 m m midline shift, category 2 for 5-15 m m midline shift and category 3 for >15 m m o f midline shift. O t h e r CT parameters included the presence of subarachnoid blood, intraventricular haemorrhage, other intracranial h a e m a t o m a and the presence and size of an
J. Clin. Neuroscience Volume 3 Number 2 April 1996
143
Clinical studies
Acute subdural haematoma
associated brain contusion. A contusion was defined as a hyperdense area less than 1 cm diameter and a haematoma if the diameter exceeded 1 cm. O u t c o m e was assessed at one year after the head injury using the Glasgow O u t c o m e Scale. 4 GOS 1 and 2 were grouped together and t e r m e d functional recovery (FR), while GOS 3-5 were g r o u p e d together and t e r m e d death or disability (DD). Patients presenting in coma were managed according to a standard protocol with intensive care admission, intubafion, ventilation, narcotic analgesia and muscle relaxation. Anficonvulsants were used prophylactically in all patients. Initial assessm e n t was p e r f o r m e d by a multi-disciplinary trauma response team. A C T scan was p e r f o r m e d on all patients when cardiovascular stability was achieved and within 4 h of admission. ASDHs causing midline shift of the septum pellucidum of >5 m m were evacuated at craniotomy. Surgical preference dictated whether the dura was left open or closed and whether the b o n e flap was replaced. T h e majority o f patients had their intracranial pressure (ICP) m o n i t o r e d either post-operatively or electively if the h a e m a t o m a did not require surgical drainage. ICP was managed in the post-operative group via a subdural catheter inserted at operation. In the patients not undergoing craniotomy a variety of methods were employed to monitor ICP including subdural catheter, ventricular catheter or a Richmond bolt. ICP was measured continuously and a mean ICP of <20 m m H g was considered as normal. W h e n the mean ICP was >20 m m H g for 15 min continuously the ICP was considered as raised. Raised ICP was managed with hypervenfilation, mannitol, intravenous p h e n o b a r b i t o n e as a bolus injection and refractory cases were treated with a p h e n o b a r b i t o n e infusion. Raised ICP requiring treatment was assessed by a three tier response to that treatment. A good response o c c u r r e d when the ICP r e t u r n e d from an elevated value to normal (<20 m m H g ) and remained there for at least 4 h. A partial response was defined as an ICP which decreased or r e t u r n e d to normal for <4 h. Nonresponsive ICP occurred when there was no change or an increase folowing treatment. Duration o f ventilation was d e t e r m i n e d by clinical response and ICP levels. Statistical analysis was p e r f o r m e d to assess which of the variables were significant in predicting final outcome. The logistic regression analysis was p e r f o r m e d using SPIDA. 5 Students' t-test was used for comparisons of means. Chi-squared tests were used for proportions. A logistic regression model was used to predict the probability of a subject having a death/disability outcome with given characteristics. The model selection enabled the identification o f the characteristics which appeared to be the best predictors o f outcome. T h e probability of death or disability was estimated for 12 months post-injury. For each patient a prognostic index (S) was based on the significant variables (ZK) which can be obtained and expressed as:
These are estimated by maximum likelihood using the LREG p r o c e d u r e of SPIDA. After S is obtained, the probability of death/disability (PD) for a given patient is indicated as: 1
P D - - 1 + e -s To test the classification accuracy of the predictive model, the probability scores are assigned to FR or DD categories at 99 probability partition values ranging from 1-99%. Thus a 2 x 2 contingency table of observed versus predicted outcomes was obtained for each probability partition value, and the corresponding sensitivity and specificity calculated. A partition value was then selected which gave the best overall accuracy o f predicted numbers of DD outcomes versus observed DD outcomes. The predicted n u m b e r of deaths/disability in each group was then estimated and c o m p a r e d with the observed number.
Results
The 103 patients who were entered in the study were divided into two groups. Group 1 comprised 79 patients with an initial GCS of 8 or less and Group 2 comprised 24 who had a GCS o f 9-15 at presentation. In this latter group 15 were male and 9 were female. All patients achieved a GOS of 1 or 2 at 6 months (22 with 1 and two with 2). Ten patients were treated surgically with a craniotomy and removal of the ASDH, while 14 were treated non-surgically with neurological observation, fluid restriction and osmotic agents. Eighteen patients were discharged from hospital within 14 days. The remaining 79 patients with a severe head injury and who were in coma were analysed in detail and all tables refer to this group. Mechanisms of injury
O f the 79 patients in Group 1, 35 were primarily admitted to Westmead Hospital, while 44 were transferred from surrounding base hospitals for neurosurgical care. The majority of patients were involved in m o t o r vehicle collisions, 34/79 (43%), suggesting that most of the haematomas were secondary to high energy impacts (Table 1).
Table 1 Mechanism of Injury
S = B 0 + B1Z a + B2Z ~ + ....... + BKZ K
where Z1, Z~ and Z~ are the predictive variables, and B 0, B 1 and B~ are the corresponding regression coefficients.
144
J. Clin. Neuroscience Volume 3 Number 2 April 1996
No of patients MVA Fall Pedestrian MBA Bicycle Assault Horse Other
34 8 5 5 5 3 3 16
Acute subdural haematoma
Clinical studies
Time from injury to treatment and neurosurgical intervention
Table 2 Outcomes of 74 Group 1 patients Age (Years)
Functional Recovery
Death/ Disability
0- 9 10-19 20 - 29 30 - 39 40 - 49 50 - 59 60 - 69 70 - 79 80 - 89
1 8 9 3 1 2 0 0 0
4 5 14 10 3 2 6 9 2
Time from injury to treatment at Westmead was assessed using two groups: group A's time from injury to treatm e n t was less than four hours and group B was greater than 4 h. 33% of patients were functional survivors in the <4 h group, while 21% o f patients were functional survivors in the >4 h group. T h e r e was no statistical difference between these groups.
Pupillary response Age and sex distribution Fifty-five patients (70%) were male, while 24 were female. The ages ranged from 2 to 89 years (mean 37 years). As age advanced o u t c o m e became worse with no functional survival after age 60 (Table 2).
Hypoxia / Hypotension Nine patients were confirmed hypoxic on admission via arterial blood gas. It was suspected (ambulance reports, etc) that a further 24 patients may have had an hypoxic episode prior to their admission to Westmead. These two groups were c o m b i n e d and o u t c o m e was assessed in terms of FR and DD. T h e presence of hypoxia as defined above was significant in predicting o u t c o m e (P < 0.005, Chi-squared test for i n d e p e n d e n c e ) . Hypotension was confirmed in 20 patients prior to resuscitation. T h e presence of hypotension was significant in predicting a DD o u t c o m e (P < 0.01, Chi-squared test for i n d e p e n d e n c e ) (Tables 3 and 4).
Lucid interval
21% of patients who had an abnormal pupillary response achieved a functional recovery c o m p a r e d with 41% o f patients who had equal and reactive pupils. This result is not statistically significant.
CT scan analysis (i) Midline shift (Table 5) shows the a m o u n t of midline shift as measured from the septum pellucidum. With increasing midline shift the o u t c o m e became worse. All 12 patients with >15 m m o f shift either died or were severely disabled. Midline shift was significant in determining outcome (P < 0.013, Chisquared test for i n d e p e n d e n c e ) . (ii) Cerebral o e d e m a (Table 6) occurred in 43 of 79 patients and was significant in predicting o u t c o m e (P< 0.013, Chi-squared test for i n d e p e n d e n c e ) . (iii) O t h e r CT parameters. The presence of brain stem cistern compression, site of haematoma, subarachnoid haemorrhage, intraventricular haemorrhage, intracerebral haemorrhage, extradural haemorrhage or brain contusions did not reach statistical significance.
A lucid interval was identified in 12 patients, however it did not help to differentiate outcome.
Table 5 Shift of midline on CT None
Table 3 The influence of hypoxia on o u t c o m e Hypoxia Present Absent
Functional Recovery
Death/ Disability
4 20
29 26
Functional Recovery
Death/ Disability
6 10 8 0
4 12 27 12
None < 5 mm 5-15 mm > 15 mm
P < 0.005 (Chi-square)
P < 0.005 (Chi-square)
Table 4 The influence of hypotension on outcome Hypotension
Functional Recovery
Death/ Disability
Present Absent
2 22
18 37
Table 6 Cerebral oedema on CT Functional Recovery
Death/ Disability
8 16
35 20
Present Absent
P < 0.01 (Chi-square)
P < 0.013 (Chi-square)
J. Clin, Neuroscience
Volume 3
Number
2 April 1996
145
Clinical studies
Acute subdural haematoma
I n t r a c r a n i a l pressure
1.o
ICP was m o n i t o r e d in 56 patients. T h e trend was as m e a n ICP rose for the o u t c o m e to b e c o m e worse, but this was n o t statistically significant.
~.m "~ o.8tD
r-Response
to treatment
of raised ICP
~ 0.6-
T h e three groups of response to t r e a t m e n t o f raised ICP did differentiate the o u t c o m e groups FR and DD (P < 0.05, Chi-squared test for i n d e p e n d e n c e ) . All eight patients who showed no response to persistently raised ICP achieved a DD o u t c o m e while a g o o d response to t r e a t m e n t (36 patients) d e m o n s t r a t e d a 61% DD outcome. Barbiturate c o m a was instituted in ten cases to control ICP and four o f these patients m a d e a functional recovery (Table 7).
rn 0.5O 0.4-
s~ o A-
0-2-
0
i
I
3
4
5
6
7
I 3
I 4
I 5
I 6
7
I
-5-5-4-3-2-1
0
1
2
S Score Fig. 1 Percentage chance of d e a t h or disability.
T a b l e 7 Response t o t r e a t m e n t of raised ICP
Functional Recovery
No response Partial response Good response
0 7 14
Death/ Disability
8 5 22
P < 0.05 (Chi-square)
1.0 t~
0.8-
C3 t-
O.6-
£3 ~_ 0 . 5 -
Prediction of o u t c o m e
0
i
>~ 0 . 4 -
..--
Age, sex, time f r o m injury to t r e a t m e n t at a neurosurgical centre, pupillary response, hypoxia, hypotension, brain stem cistern compression, shift size and cerebral o e d e m a were f u r t h e r analysed using logistic regression against the two o u t c o m e categories FR a n d DD. Using a backward elimination selection p r o c e d u r e the variables which comprised the final m o d e l were age and shift. T h e m o d e l comprises an ASDH score of "S" where S = - 2.481 + 0.044 x age + 1.284 x shift category. Figure 1 is a n o m o g r a m curve of"S" versus probability of death or disability. T h e above m o d e l was most accurate w h e n using a partition value of 0.5 and this resulted in an overall accuracy of 80% with sensitivity of 58% and specificity of 89%. An example o f the use of this m o d e l would be a patient with an ASDH aged 40 years with a shift of 10 m m (categ o r y 2). T h e "S" score w h e n calculated is 1.85 (- 2.481 + 40 x 0.044 + 2 x 1.284). T h e c o r r e s p o n d i n g percentage chance of death or disability is therefore 0.86 f r o m the n o m o g r a m curve (Fig. 1). At the completion o f this study the above m o d e l was tested on 22 consecutive patients with an ASDH GCS8 or less, following h e a d injury who were admitted to Westmead Hospital and to Sir Charles Gairdner and the Royal Perth Hospitals. T h e y constituted a similar sample in respect to age, sex, m e c h a n i s m of injury a n d clinical features to those analysed above. T h e o u t c o m e of these patients was assessed at six months. Figure 2 is a n o m o g r a m curve with the "S" score for the 22 patients indicated by a dot. Two patients were predicted inaccurately by the m o d e l (cross dot). These patients were predicted
146
i
.Q -~ 0.2--
0
I
~
-6
-5
-4
-3
-2
-1
0
I 1
I 2
S Score Fig. 2 O u t c o m e assessed at six months.
by the m o d e l to make a DD o u t c o m e at six months, however the actual o u t c o m e was an FR. In this small prospective test of the m o d e l it was accurate at predicting 20 of 22 patients, i.e. 91%.
Discussion T h e mortality rate of ASDH is high, 6-15 in a majority of series ranging between 48% 7 a n d 75%. 11 O u r series mortality for the 103 patients was 50%; for the 79 patients in c o m a (GCS 8 or less) it was 66%. Only 4% of the patients presenting in c o m a r e m a i n e d severely disabled while 30% were functional. This implies that m o d e r n managem e n t of ASDH does not p r o d u c e large n u m b e r s of severely disabled p e o p l e as a secondary effect of reducing the n u m b e r o f fatalities. Stone et a115 r e p o r t 27% of his 128 patients as functional survivors, a similar percentage to o u r series, i For those patients with a severe h e a d injury (GCS 8 or less) age was a significant d e t e r m i n a n t of o u t c o m e with
J. Clin. Neuroscience Volume 3 Number 2 April 1996
Acute subdural haematoma
Clinical studies
no functional survival greater than 60 years. Howard et a116 analysed age in relation to ASDH; they concluded that 66% of y o u n g patients (18-40 years) m a d e a functional recovery c o m p a r e d with a 9% functional recovery in the elderly (age > 65 years). T h e m e c h a n i s m by which age has such an effect on o u t c o m e is unknown, 17but suggestions include a p o o r regenerative capacity of the older brain and a predisposition to develop a m o r e lethal injury, x6 In the m o d e l predicting o u t c o m e by Klun, TM age was a factor with a sharp rise in mortality for age > 50 years. O u r predictive m o d e l used age as o n e of the key variables in d e t e r m i n i n g outcome. Hypoxia and hypotension were significant factors in d e t e r m i n i n g outcome. T h e diagnosis of hypoxia implies a p r i m a r y (hypoventilation secondary to brain stem injury) or secondary event (airway or chest injury). Whatever the underlying cause the prognosis was p o o r with only four functional recoveries in a g r o u p of 33 patients (12%). Hypotension similarly caused a significant grave prognosis with only 2 of our 20 patients in this g r o u p making a functional recovery. This factor was rated by Stening et a119 as being a significant factor in the o u t c o m e o f a review of ASDH in New South Wales, Australia, in the 1970s. T h e larger the midline shift, as m e a s u r e d f r o m midline to s e p t u m pellucidum, the worse the outcome. T h e degree of shift is d e t e r m i n e d by the volume of the ASDH plus the swelling of the underlying h e m i s p h e r e . It has b e e n suggested that the h a e m a t o m a itself is an epiphen o m e n o n with the underlying brain injury d e t e r m i n i n g outcome. 2° T h e m e a s u r e m e n t o f brain shift encompasses b o t h of these factors. Marshall et al 2 r e p o r t e d that a greater midline shift was associated with a worse recovery. Ross et a122 d e m o n s t r a t e d that a trend towards high ICP was seen with a shift > 5 m m . Becker et a123 concluded that the mortality in patients with 10 m m or greater midline shift was 53% a n d this c o m p a r e d with 25% for those patients with less than 10 m m o f shift. We f o u n d that midline shift was significant when d e t e r m i n i n g o u t c o m e a n d was such a strong predictor that it was included in the m o d e l for prediction. Cerebral o e d e m a was significant in d e t e r m i n i n g outc o m e (P< 0.013, Chi-squared test for i n d e p e n d e n c e ) . O f the four factors used to assess cerebral oedema, individually they were n o t significant in predicting o u t c o m e but when used in c o m b i n a t i o n they achieved a significance (any 3 of the 4). ICP was studied in two ways. Firstly, the m e a n ICP category in which the patient spent most of the postoperative time was c o m p a r e d with outcome. As ICP rose patients t e n d e d to have worse outcomes, b u t the numbers were too small to achieve statistical significance. We also graded the patients as to their response to ICP control; this proved to be significant in d e t e r m i n i n g outcome. Miller et a122 h a d similar ICP response categories in their review of the significance of ICP m o n i t o r i n g in h e a d injury (including but not limited to ASDH) a n d in their n o n - r e s p o n d i n g group; as with ours, all the patients died or were severely disabled. Seelig et a124 concluded that o u t c o m e was substantially influenced by ICP; 79% of
their patients who had a functional recovery had a postoperative ICP of less than 20 m m H g . Seelig et a124 in their study concluded that a delay f r o m injury to operation was the factor of greatest therapeutic i m p o r t a n c e in ASDH. Patients who u n d e r w e n t surgery in the first four hours h a d a 30% mortality comp a r e d with 90% in those who h a d surgery after four hours. O u r study did not confirm this finding, with 66% of patients achieving a DD o u t c o m e if t r e a t m e n t was instituted within four hours of injury, and 79% achieving a DD o u t c o m e if treated m o r e than four hours following injury. This difference was not statistically significant. Klun et all8 r e p o r t e d 330 cases o f ASDH and failed to achieve lower mortality by early operation a n d similarly Wilberger et al25 c o n f i r m e d the absolute timing for the removal of an ASDH to be u n i m p o r t a n t in terms of this four h o u r surgical window. T h e two categories we used for pupil response failed to help us predict o u t c o m e with 21% of patients achieving an FR with pupillary signs on admission c o m p a r e d with 40% achieving an FR with equal a n d reactive pupils. T h e pupillary response r e c o r d e d was at the initial examination at Westmead and a p e r c e n t a g e of patients may have c h a n g e d their initial pupillary signs once adequately resuscitated. T h e literature has widely s u p p o r t e d pupillary change as being significant in d e t e r m i n i n g outcome. J a m i e s o n et al's study 26 o f 553 ASDH patients gave a mortality rate of 41% for o n e fixed pupil and 85% for two fixed pupils. Similarly Stening et al's 19 mortality for fLxed dilated pupils was 93% as o p p o s e d to 58% for normally reactive pupils. Lucid interval did not influence o u t c o m e for FR a n d DD and this result c o m p a r e s favourably with Stening et aP 9 who also f o u n d no relationship.
Conclusions Based on logistic regression analysis o f factors which may have influenced o u t c o m e in ASDH, two factors - age and shift of the septum pellucidum - were used to construct a m o d e l to predict the probability of achieving a DD outcome. T h e accuracy of this m o d e l was 80% and it successfully predicted the o u t c o m e o f a f u r t h e r 20 out o f 22 patients (91%) when used prospectively. Attention is drawn to the factors of hypoxia and hypotension which are often easily correctable in the m a n a g e m e n t of a patient with an ASDH. An elderly p a t e n t (> 60 years) in c o m a with a traumatic ASDH, irrespective of the treatment, is m o s t likely to be either d e a d or severely disabled at 12 m o n t h s post-injury. Received 11 January 1994 Accepted for publication 19 May 1995
Correspondence and offprint requests: R.J. Cook Northshore Medical Centre 66 Pacific Highway St Leonards NSW 2065, Australia Phone: (02) 43%4125 Fax: (02) 437-4147
J. Clin. Neuroscience Volume 3 Number 2 April 1996
147
Clinical studies References 1. Jones NR, Blumbergs PC, NorthJB. Acute subdural haematomas: aetiology, pathology and outcome. Aust NZ J Surg 1986; 56:907-915. 2. Marshall LF, Toole BM, Bowers SA. The National Traumatic Coma Data Bank, Part 2. Patients who talk and deteriorate: implications for treatment. J Neurosurg 1983; 59:285-288. 3. The Traumatic Coma Data Bank: Design, methods and baseline characteristics.J Neurosurg 1991; 75: (Suppl. S8-S13). 4. Jennett B, Bond M. Assessment of outcome after severe brain damage. A practical scale. Lancet 1975; 1:480-484. 5. Hamey L, Leung O, McNeil D. SPIDA (Statistical Package for Interactive Data Analysis), Users manual. MacQuarie Press 1985; Sydney, Australia. 6. GalbraithJG. Subdural haematoma- acute and chronic. A reappraisal. Clin Neurosurg 1982; 29:24-31. 7. Fell DA, Fitzgerald MD, Moiel RH, et al. Acute subdural haematoma - review of 144 cases.J Neurosurg 1975; 42: 3742. 8. Shenkin HA. Acute subdural haematoma. J Neurosurg 1982; 57:254-257. 9. Yui CP, Mann KS, Ong GB. Acute subdural haematoma Better results with neurosurgeons than general surgeons. Injury 1983; 14:489-492. 10. Haselberger K, Pucher R, Aver LM. Prognosis after acute subdural and epidural haemorrhage. Acta Neurochir 1988; 90:111-116. 11. Richards T, HoffJ. Factors affecting survival from acute subdural haematoma. Surgery 1974; 75:253-258. 12. Shigemori M, Syojimi K, Nakayama K, et al. The outcome from acute subdural haematoma following decompressive hemi-craniectomy. Acta Neurochir. 1980; 54:61-69. 13. HernesniemiJ. Outcome following acute subdural haematoma. Acta Neurochir 1970; 49:191-198. 14. Talalla IA, Morin MA. Acute traumatic subdural haematoma: A review of one hundred consecutive cases J Trauma 1971; 11:771-777.
148
Acute subdural haematoma 15. StoneJL, Lowe RJ, Jonasson o. Acute subdural haematoma: Direct admission to a trauma centre yields improved results. J Trauma 1986; 26:445-450. 16. Howard MA III, Gross AS, Daay RG, et al. Acute subdural haematomas: An age dependent clinical entity. J Neurosurg 1989; 71:858-863. 17. HernesniemiJ. Outcome following head injuries in the aged. Acta Neurochir 1979; 49:6%69. 18. Klun B, Fettich M. Factors influencing the outcome in acute subdural haematoma. Acta Neurochir 1984; 71: 171-178. 19. Stening WA, Berry N, Dan NG, et al. Experience with acute subdural haematomas in New South Wales. Aust NZ J Surg 1986; 56:549-556. 20. Casey ATH, Cook RJ, Fearnside MR, et al. Acute subdural haematoma- Predictors of outcome in 214 patients. Neurology Neurosurgery Psychiatry 1993; 56:222-225. Abstract 120th Proceedings SBNS. 21. 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. 22. MillerJD, Becker DP, WardJD, et al. Significance of intracranial hypertension in severe head injury. J Neurosurg 1977; 47:503-516. 23. Becker DP, MillerJD, Ward J, et al. The outcome from severe head injury with early diagnosis and intensive management. J Neurosurg 1977; 47: 491-502. 24. SeeligJM, Becker DP, MillerJD, et al. Traumatic acute subdural haematoma. Major mortality reduction in comatose patients treated within four hours. New EnglJ Med 1981; 304:1511-1518. 25. WilbergerJE, Harris M, Diamond DL. Acute subdural haematoma: Morbidity, mortality and operative timing. J Neurosurg 1991; 74:212-218. 26. Jamieson KG, YellandJDN. Surgically treated traumatic subdural haematomas.J Neurosurg 1979; 37:137-149.
J. Clin. Neuroscience Volume 3 Number 2 April 1996