Injury (1992) 23,(7). 471-474 Rinkdin Great Brituin
471
Death in hospital after head injury without transfer to a neurosurgical unit: who, when, and why? D. Gentleman, B. Jennett and R. MacMillan Department
of Neurosurgery,
Institute
of Neurological
Sciences, Southern
Most studies of hospital dab after head injury have been in patients transferred to neurosurgical units (NSU), but over 90 per cent of hospitalized head-injured pafients are not transferredandsomeof these die. TO assess the t$ecfiveness of triuge of seriously head-injured patients in Glasgow, we studied 2 70patients who died after head injury in any of the six Glasgow general hospitals during 1979-1988 and who were not transferred to the regional NSU. The proportion of fatal cases of head injury who had nof been to theNSUfellfrom 69percent in 1971-1975 to 45per cent in1979-l 988. Most of k untransferredpatients were elderly, and most died from irremediable injuries or complications.Although 31 (11 per cent) had a significant intracranial haematoma, only seven of these might have been salvaged by neurosurgical intervention. Seven other patients diedfrom potentially preventable extracranial injuries or complications. lh5e findings wed that a relatively satkfacto y level of triage of seriowly head-injured patienfs has been achieved, by promoting effecfive communication befwezn neurosurgeons and other specialisk, and by a continuous programme of audit and education.
Introduction Death after head injury may occur before the patient reaches hospital, or in the general hospital to which he is first taken, or after secondary transfer to the NSU. We have reported that avoidable adverse factors contributed to at least one-third of deaths which occurred after -transfer to the Glasgow NSU in the early 1970s (Rose et al., 1977). There is also evidence that few deaths before arrival at the first hospital can be prevented (Jennett and Carlin, 1978). Little is known about a third group of head-injury fatalities - those who survive long enough to be admitted to a general hospital, but who die there without having been transferred to a NSU - as mentioned in a study of trauma deaths in general (Anderson et al., 1988). Neurosurgeons in the UK see relatively few of the many head-injured patients admitted to general hospitals, but there is marked regional variation. Only I per cent of head-injured patients admitted to Merseyside general hospitals in the mid 1970s were transferred to the regional NSU in Liverpool (Jeffreys and Jones, 1981), compared with 5 per cent for the UK as a whole, and 30 per cent or more in Edinburgh (Jennett et al., 1979). In Glasgow the rate was about 5 per cent until 1978 when, in the light of 4-years’ experience of CT scanning, a new policy was devised to encourage the transfer to the NSU of more patients considered at risk of remediable complications, in particular intracranial haematoma. The introduction of this policy led c 1992Butterworth-Heinemann 0020-1383/92/070471-04
Ltd
General Hospital, Glasgow,
UK
to a doubling of the number of head-injured patients transferred each year to the Glasgow NSU, and it has remained at this higher level since (Bryden and Jennett, 1983). More haematomas were then detected and treated, and there was an improvement in the mortality and morbidity rates for patients operated on for this complication (Teasdale et al., 1982). The experience gained from implementing this policy formed the basis of guidelines agreed by a group of British neurosurgeons for adults (A Group of Neurosurgeons, 1984). which have recently been updated and extended to include children (Teasdale et al., 1990). The present paper describes an investigation of headinjured patients who died in general hospitals in Glasgow, without having been transferred to the NSU, during the 10 years which followed the introduction of the new transfer policy. It was designed to discover whether some of these patients might have benefited from transfer to the NSU, and to allow comparison with an unpublished study carried out before the introduction of the more liberal transfer policy.
The hospitals, the methods and the patients The regional NSU in Glasgow provides the only neurosurgical service for a population of 2.7 million in the west of Scotland. It accepts head-injured patients only after they have initially been dealt with by other clinical services (e.g. accident and emergency, general surgery, orthopaedics) in some 20 general hospitals in the region. More than 500 head-injured patients are transferred each year to this NSU, one-half of them from the six teaching hospitals in Glasgow (including the Children’s Hospital). To exclude any possible influence of distance on the decision to transfer, this study was confined to these six hospitals, which are between 10 and 30 min by road from the NSU. National statistics were used to identify patients who had died in these hospitals during the 10 years 1979-1988, in whom any of the ICD-9 codes for head injury had been recorded (in any position). There were 1060 such cases: 662 of them had died at the NSU or after return to their referring hospital, which in 336 cases was one of the six Glasgow hospitals. Of the 398 Glasgow patients who had not been to the NSU, the case notes of 52 could not be traced and another 76 cases had been miscoded as head injuries (mainly cases of spontaneous intracranial haemorrhage or facial injury). The remaining 270 patients with available records who had died in Glasgow hospitals after a head injury (but
Injury: the British Journal of Accident Surgery (1992) Vol. 23/No. 7
472
the head injury alone) without being transferred to the NSU formed the basis of this study. The 270 case records were scrutinized by one of the authors (DC). Data abstracted included severity of initial brain damage (assessed by evidence of talking after injury, the Glasgow Coma Scale Score (Teasdale and JenneH, 1974) on admission and the pupil response). The severity of extracranial injuries was assessed using the Injury Severity Score (Baker et al., 1974); also recorded were secondary factors that might have contributed to death such as intracranial haematoma or systemic hypoxia or hypotension. The Regional Procurator Fiscal or the hospital pathologists concerned kindly provided details of autopsy in 189 (70 per cent) of the 270 cases, including 112 (92 per cent) of the 122 patients aged under 70 years; only 19 patients had no post-mortem examination. not necessarily
from
Results The patients The mean age of the 270 patients was 62 years (median 72 years, range 2 months-101 years). There were 175 males and 95 females. Road traffic accidents accounted for 49 per cent and falls for 44 per cent. The median survival for all cases was 42 h, and 64 cases (24 per cent) died within 6 h of injury. Causes of death (TableI) Primary brain damage was judged to be the cause of death in 116 patients (43 per cent). In 82 (71 per cent) of these there was no motor response to painful stimuli, and/or dilated unresponsive pupils when first seen at hospital. Intractable hypotension (36 cases) and respiratory or cardiac arrest (20 cases) were common in this group. Their median survival was only 18 h, with 42 (36 per cent) of them dead within 6 h of injury. None of these 116 deaths was considered preventable. An acute intracranial haematoma was considered to be the main cause of death in 31 patients (11 per cent), but nine of them had presented to hospital only after a delay of 12 h or more. Although the median age was 73 years, four were under 40 years of age. Most (21) of the 31 had an intracerebral haematoma, and only one had an isolated extradural haematoma. This latter patient lay unnoticed for many hours, and was moribund on reaching hospital. Of the 31 patients, 12 (39 per cent) had definitely talked after injury and another four may have done. In three cases the finding of a haematoma at post-mortem had been a surprise. The severity of the head injury had been underestimated in at least 15, seven of whom were aged under 70 years. However, in no more than seven of the 31 cases was it considered that death might have been preventable, taking into account age and the severity of injury (Tuble II). Spontaneous cerebrovascular accidents caused the injuries by falling and the deaths of 13 patients (5 per cent), all of whom were over 70 years of age. None of these 13 deaths was judged to be preventable. Extracranial injuries or complications caused 105 (40 per cent) of the 270 deaths, but only seven of these were judged to be potentially preventable. All seven were aged under 70 years, and their head injury was mild or moderate only (Tub/e 111).Exsanguination from multiple injuries occurred in 28 patients, 20 (71 per cent) of them within 24 h of injury. Of the 28 patients, 13 had a Glasgow Coma Score of 12-15 on
Table I. Principal cause of death of 270 patients who died after a head injury during 1979-1988 and were not transferred to z NW Intracranial factors Primary brain damage Intracranial haematoma Cerebrovascular accident Extracranial factors Extracranial blood loss Pneumonia Major chest injury Pelvic/limb fractures Traumatic quadriplegia Cerebrovascular accident Cor pulmonale Motor neurone disease Kyphoscoliosis End-stage alcoholism No predisposing factor Acute cardiac events/pulmonary embolism Various complications of injury Infection (other than chest) Renal failure Fat embolism Gastric bleeding Long immobilization Coincidental malignant disease Anoxic brain damage from cardiac arrest Cause remains obscure
160 116 31 13 106 28 46 15 6 : 4 1 1 2 14 21 6 2 1 1 1 1 3 1 5
Total
270
TableII. ‘Saveable’ and ‘unsaveable’ patients who died from intracranial haematoma Potentially savable (N= 7)
Unsavable (N=24)
6 (86%)
6 (25%)’
6 (86%)
12 (50%)’
: (86%) 1 (14%)
1 ( 4%): 23 (96%) 0
Age under 70 years Best Glasgow Coma Score after injury over 8 Type of haematoma Extradural lntradural Both
‘Best Glasgow Coma Score was 6 in one case, 3 in the remainder. ‘Aged from 73 to 86 years. :Long delay before hospital; moribund on arrival.
Table III. Potentially saveable patients who died from extracranial injuries or complications Age Case (years)
Cause of death
1.
16
Blood loss
2. 3.
65 42
Blood loss Blood loss
4.
39
Blood loss
5. 6.
52 57
Pulmonary embolism Pulmonary embolism
7.
11
Anoxic brain damage
Remarks Refused surgery for vena cava tear Unrecognized aortic tear Delayed recognition of haemoperitoneum Delayed recognition of haemothorax Multiple injuries, immobilized; no anticoagulant prophylaxis Cardiac arrest under anaesthesia for manipulation of limb fracture
Gentleman et al.: Death in hospital after head injury arrival at hospital, and 11 became hypotensive only after admission, raising questions about the appropriateness of early management. However, even without the head injury, the Injury Severity Score was high (25-50) in all but three very elderly patients. We judged that only four of these 28 deaths were potentially preventable. Predisposing factors were identified in 32 of the 46 deaths from pneumonia, the other 14 patients being aged between 80 and 97 years. Acute cardiac events caused 17 deaths (6 per cent), only three of them in patients under 70 years of age. Four patients died from massive pulmonary embolism after mild head injuries, six from various systemic complications of injury, three from coincidental malignant disease, and one from anoxic brain damage caused by a cardiac arrest during a genera1 anaesthetic for a minor orthopaedic procedure. The only potentially avoidable deaths in this large group were this last case, and fwo young patients with fatal pulmonary embolism who had not received anticoagulant prophylaxis despite being at high risk from immobilization with multiple injuries. In five patients (2 per cent), the cause of death remained obscure. Although clinical evidence suggested an intracrania1 haematoma in four of the five, the youngest was aged 85 years, and it is unlikely that their deaths were preventable. Referral to the NSU The NSU had been consulted by telephone in 123 cases (46 per cent), one-half of them within 6 h of injury; in 16 of these a neurosurgeon had also visited the patient at the general hospital. Transfer to the NSU in all 123 cases had been considered inappropriate by the neurosurgeon, because of the severity of initial injury, the patient’s advanced age, or both. No patient had a CT scan in a genera1 hospital. A significant haematoma was subsequently found at autopsy in 11 of these patients, but only three were aged under 70 years. Historical
trends
among
untransferred
patients
(TubleIV) An earlier unpublished
study of hospital deaths after head injury in Glasgow during 1971-1975 provides a historical control for this series (Strang, personal communication). A smaller proportion of patients (31 per cent of 388) was transferred during that earlier period, and analysis of the 268 untransferred fatalities showed more patients aged under 70 years, and more who survived for over 24 h, than in the present series. Transferred and untransferred patients during 19791988 (Table V) Far fewer of the transferred patients were elderly, fewer had severe multiple injuries, and fewer died within 24 h of their injury. This suggests that appropriate triage was being carried out.
Discussion Only 31 per cent of hospital deaths after head injury in Glasgow in 1971-1975 were in patients who had been transferred to the NSU, compared with 55 per cent during the 10 years after a more liberal transfer policy was introduced in 1978. Although death is a common outcome after severe head injury, and is often the result of overwhelming primary injury, the concept of ‘avoidable deaths’, is useful in auditing
473
Table IV. Head injury deaths in Glasgow hospitals; 1971-1975 and 1979-1988 (Glasgow patients only) 1971-1975
1979-1988
388 129’ (31%) 268 (69%)
606 336’ (55%) 270 (45%)
170 180
121 148
Total deaths (ommitting patients with missing case records) N (%) transferred to NSU N (%) not transferred to NSU Of those not transferred: N (%) aged under 70 years N (%) survived more than 24 h
(66%) (67%)
(45%) (55%)
*About one-half of these had been transferred to the NSU from outside Glasgow.
Table V. Transferred and untransferred fatal cases during 1979-1988 (Glasgow patients only)
Age distribution (years) Under 70 70-79 80-89 90 and over No motor response to pain, dilated unreactive pupils, or spontaneous apnoea on first arrival in hospital Multiple injuries Severe multiple injuries (three or more body regions) Survival less than 24 h
Transferred (N = 336)
Not transferred (N=270)
293 (87%) 36 (11%) 7 (2%) 0 (0%)
121 (45%) 60 (22%) 76 (28%) 13 (5%)
128 (38%) 108 (32%)
104 (38%) 141 (52%)
27 (8%) 77 (23%)
62 (23%) 118 (44%)
the effectiveness of policies for dealing with injured patients (Reilly et al., 1975; Rose et al., 1977; JenneH and Carlin, 1978; Phair et al., 1991). Deciding whether or not a particular death might have been averted by better management is not an exact science, but relies upon a value judgement to distinguish between what might have been achieved under ideal circumstances and what could reasonably have been expected in practical terms. For example, one-third of the patients who died from intracranial haematoma in this series reached hospital only after a delay of several hours, during which some who had been talking deteriorated info coma. Some of these deaths might well have been prevented had they attended hospital sooner, but it would be unreasonable to regard them as preventable by better medical care. Age is an important determinant of outcome after severe head injury or multiple trauma (Teasdale et al., 1979). Operating for intracranial haematoma over age 70 years gives uniformly poor results, unless the patient has been talking after injury, is operated upon before deterioration into coma, and has only an extracerebral clot. Age therefore plays a legitimate part in triage, and it is no surprise that during 1979-1988 there were many more elderly patients among the untransferred cases than among those who went to the NSU. The second major determinant of outcome after head injury is the severity of primary brain damage. At any age, the finding on arrival at hospital of no motor response, dilated unreactive pupils, or spontaneous apnoea is highly predictive of a fatal outcome. The third major influence on outcome - and the only one which can be corrected - is the severity of secondary
474
Injury: the British Journal of Accident Surgery (1992) Vol. 23/No. 7
hypoxic-ischaemic insults to the brain and other organs. This caused 14 deaths in patients aged under 70 years which might have been averted by different management after admission to hospital. These cases include a boy who had a cardiac arrest under anaesthesia for manipulation of a forearm fracture, two fatal cases of pulmonary embolism in multiply-injured men who were not given prophylactic anticoagulation, and four who died from unheated concealed thoracic or abdominal blood loss (one was a teenage motorcyclist who refused life-saving abdominal surgery despite warnings of the likely consequences). This leaves only seven patients under 70 years of age who had not had overwhelming primary injury to the brain or other organs, and who died from unsuspected intracranial haematomas. In the period during which these seven potentially salvageable cases were not transferred, some 3000 head-injured patients were transferred from the same hospitals, and some 600 of these had operations for significant haematomas, with an overall mortality of around 30 per cent. The system for identifying patients who should be transferred therefore appears to be working reasonably effectively. Delayed referral of patients is known to be a major contributor to mortality, as well as to permanent disability in some survivors (Mendelow et al., 197% Seelig et al., 1981). In the west of Scotland much work has gone into developing and refining a policy fo encourage early referral of appropriate patients to the NSU, and during 1979-1988 many fewer potentially salvageable patients were not transferred than in 1971-1975. Maintaining this policy has required considerable and continuing effort in education and publicity as junior staff change in the various referring hospitals. Despite these efforts, a few patients who might benefit from neurosurgical services are still not transferred, and it is a maHer for concern that avoidable death from secondary brain damage caused by compressive intracranial haematomas still occurs in the same city as a regional NSU. There is an urgent need for more systematic audit of head injury management, both in general hospitals and in NSUs nationwide. This could establish whether policies or practices need to be reviewed to minimize avoidable death and disability. It is important that consulfants in accident departments and those in charge of primary surgical wards should liaise regularly with their colleagues in the regional NSU, who in turn must be prepared to agree locally appropriate criteria about which patients call for consultation with the NSU, for possible transfer. Transfer should only happen after consultation, and after adequate steps have been taken to minimize adverse incidents en route that might add to the brain damage (Gentleman and Jennett, 1990). As most head-injured patients are initially managed by doctors in training, who change at frequent intervals, it is important that they be taught and know what is expected of them.
Acknowledgements The authors gratefully acknowledge the help they received from many individuals. Dr John Bryden and MS Barbara Boyd of the Management Information Services Division of Greater Glasgow Health Board provided information to identify patients for the study. The Medical Records Officers of Glasgow Royal Infirmary, the Western Infirmary/Gartnavel General Hospital, the Victoria Infirmary, Stobhill Hospital, the Southern General Hospital, and Royal Hospital for Sick Children helped in the location of
hospital case records, and the consultants in these hospitals gave permission for the records of their deceased patients to be examined. The Crown Office, the Procurator Fiscals of Glasgow and Strathkelvin, Dumbarton, and Greenock, and the forensic pathologisb who carried out autopsies on these patients allowed access to their records. Dr Ian Strang collected the data on 1971-1975 head injury fatalities which were used as a comparison with this study. Dr Lilian Murray PhD provided data on head-injured patients transferred fo the NSU during 1979-1988. Miss Anne Semple prepared the final drafts ot’ the manuscript. This study was part of an investigation into avoidable mortality and morbidity after head injury funded by the Scottish Home and Health Department.
References A Group of Neurosurgeons (1984) Guidelines for initial management after head injury in adults. Br. Med. j. 288,983. Anderson I. D., Woodford M., De Dombal F. T. et al. (1988) Retrospective study of 100 deaths from injury in England and Wales. Br. Med. 1. 296, 1305. Baker S. P., O’Neill B., Haddon W. et al. (1974) The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care. 1. Trauma 14, 187. Bryden J. S. and Jennett B. (1983) Neurosurgical resources and transfer policies for head injuries. Br. Med. 1. 286, 1791. Gentleman D. and Jennett B. (1990) Audit of unconscious head injured patients transfeerred to a neurosurgical unit. Lancet 1, 330. Jeffreys R. V. and Jones J. J. (1981) Avoidable factors contributing to the death of head injury patients in general hospitals in Mersey Region. Luncef 2, 459. Jennett B. and Carlin J. (1978) Preventable mortality and morbidity after head injury. I+ry 10,31. Jennett B., Murray A., Carlin J. et al. (1979) Head injury in three Scottish neurosurgical units. Br. Med. J 2, 955. Mendelow A. D., Karmi M. Z., Paul K. S. et al. (1979) Extradural haematoma effect of delayed treatment. Br. Med. 1. fi), 1240. Phair I. C., Barton D. J. Barnes M. R. et al. (1991) Deaths following trauma: an audit of performance. Ann. R. Co/l. Swg. Engl. 73,53. Reilly P. L., Adams J. H., Graham D. I. et al. (1975) Patients with head injury who talk and die. Lance/ 2,375. Rose J., Valtonen S. and Jennett B. (1977) Avoidable factors contributing to death after head injury. Br. Med. J 2,615. Seelig J. M., Becker D. P., Miller J. D. et al. (1981) Traumatic acute subdural haematoma; major mortality reduction in comatose patients treated within 4 hours. N. Engl. J. Med. 304, 15 11. Teasdale G., Galbraith S., Murray L. et al. (1982) Management of traumatic intracranial haematoma. Br. Med. J 285, 1695.
Teasdale G. and Jennett B. (1974) Assessment impaired consciousness. Lancet 2, 81.
of coma and
Teasdale G.. Murray G., Anderson E. et al. (1990) Risks of acute traumatic intracranial haematoma in children and adults: implications for managing head injuries. Br. Med. J. 300, 363. Teasdale G., Skene A., Parker L. et al. (1979) Age, severity and outcome of head injury. A&a Neurocbir. Suppl. 28, 140.
Paper accepted 24 February
1992.
Requests for reprints shouM be addressed to: Mr D. Gentleman,
Consultant Neurosurgeon, Department of Neurosurgery, Dundee Royal Infirmary, Dundee DDl 9ND, UK.
471
Death in hospital after head injury without transfer to a neurosurgical unit: who, when, and why? D. Gentleman, B. Jennett and R. MacMillan Department
of Neurosurgery,
Institute
of Neurological
Sciences, Southern
Most studies of hospital dab after head injury have been in patients transferred to neurosurgical units (NSU), but over 90 per cent of hospitalized head-injured pafients are not transferredandsomeof these die. TO assess the t$ecfiveness of triuge of seriously head-injured patients in Glasgow, we studied 2 70patients who died after head injury in any of the six Glasgow general hospitals during 1979-1988 and who were not transferred to the regional NSU. The proportion of fatal cases of head injury who had nof been to theNSUfellfrom 69percent in 1971-1975 to 45per cent in1979-l 988. Most of k untransferredpatients were elderly, and most died from irremediable injuries or complications.Although 31 (11 per cent) had a significant intracranial haematoma, only seven of these might have been salvaged by neurosurgical intervention. Seven other patients diedfrom potentially preventable extracranial injuries or complications. lh5e findings wed that a relatively satkfacto y level of triage of seriowly head-injured patienfs has been achieved, by promoting effecfive communication befwezn neurosurgeons and other specialisk, and by a continuous programme of audit and education.
Introduction Death after head injury may occur before the patient reaches hospital, or in the general hospital to which he is first taken, or after secondary transfer to the NSU. We have reported that avoidable adverse factors contributed to at least one-third of deaths which occurred after -transfer to the Glasgow NSU in the early 1970s (Rose et al., 1977). There is also evidence that few deaths before arrival at the first hospital can be prevented (Jennett and Carlin, 1978). Little is known about a third group of head-injury fatalities - those who survive long enough to be admitted to a general hospital, but who die there without having been transferred to a NSU - as mentioned in a study of trauma deaths in general (Anderson et al., 1988). Neurosurgeons in the UK see relatively few of the many head-injured patients admitted to general hospitals, but there is marked regional variation. Only I per cent of head-injured patients admitted to Merseyside general hospitals in the mid 1970s were transferred to the regional NSU in Liverpool (Jeffreys and Jones, 1981), compared with 5 per cent for the UK as a whole, and 30 per cent or more in Edinburgh (Jennett et al., 1979). In Glasgow the rate was about 5 per cent until 1978 when, in the light of 4-years’ experience of CT scanning, a new policy was devised to encourage the transfer to the NSU of more patients considered at risk of remediable complications, in particular intracranial haematoma. The introduction of this policy led c 1992Butterworth-Heinemann 0020-1383/92/070471-04
Ltd
General Hospital, Glasgow,
UK
to a doubling of the number of head-injured patients transferred each year to the Glasgow NSU, and it has remained at this higher level since (Bryden and Jennett, 1983). More haematomas were then detected and treated, and there was an improvement in the mortality and morbidity rates for patients operated on for this complication (Teasdale et al., 1982). The experience gained from implementing this policy formed the basis of guidelines agreed by a group of British neurosurgeons for adults (A Group of Neurosurgeons, 1984). which have recently been updated and extended to include children (Teasdale et al., 1990). The present paper describes an investigation of headinjured patients who died in general hospitals in Glasgow, without having been transferred to the NSU, during the 10 years which followed the introduction of the new transfer policy. It was designed to discover whether some of these patients might have benefited from transfer to the NSU, and to allow comparison with an unpublished study carried out before the introduction of the more liberal transfer policy.
The hospitals, the methods and the patients The regional NSU in Glasgow provides the only neurosurgical service for a population of 2.7 million in the west of Scotland. It accepts head-injured patients only after they have initially been dealt with by other clinical services (e.g. accident and emergency, general surgery, orthopaedics) in some 20 general hospitals in the region. More than 500 head-injured patients are transferred each year to this NSU, one-half of them from the six teaching hospitals in Glasgow (including the Children’s Hospital). To exclude any possible influence of distance on the decision to transfer, this study was confined to these six hospitals, which are between 10 and 30 min by road from the NSU. National statistics were used to identify patients who had died in these hospitals during the 10 years 1979-1988, in whom any of the ICD-9 codes for head injury had been recorded (in any position). There were 1060 such cases: 662 of them had died at the NSU or after return to their referring hospital, which in 336 cases was one of the six Glasgow hospitals. Of the 398 Glasgow patients who had not been to the NSU, the case notes of 52 could not be traced and another 76 cases had been miscoded as head injuries (mainly cases of spontaneous intracranial haemorrhage or facial injury). The remaining 270 patients with available records who had died in Glasgow hospitals after a head injury (but
Injury: the British Journal of Accident Surgery (1992) Vol. 23/No. 7
472
the head injury alone) without being transferred to the NSU formed the basis of this study. The 270 case records were scrutinized by one of the authors (DC). Data abstracted included severity of initial brain damage (assessed by evidence of talking after injury, the Glasgow Coma Scale Score (Teasdale and JenneH, 1974) on admission and the pupil response). The severity of extracranial injuries was assessed using the Injury Severity Score (Baker et al., 1974); also recorded were secondary factors that might have contributed to death such as intracranial haematoma or systemic hypoxia or hypotension. The Regional Procurator Fiscal or the hospital pathologists concerned kindly provided details of autopsy in 189 (70 per cent) of the 270 cases, including 112 (92 per cent) of the 122 patients aged under 70 years; only 19 patients had no post-mortem examination. not necessarily
from
Results The patients The mean age of the 270 patients was 62 years (median 72 years, range 2 months-101 years). There were 175 males and 95 females. Road traffic accidents accounted for 49 per cent and falls for 44 per cent. The median survival for all cases was 42 h, and 64 cases (24 per cent) died within 6 h of injury. Causes of death (TableI) Primary brain damage was judged to be the cause of death in 116 patients (43 per cent). In 82 (71 per cent) of these there was no motor response to painful stimuli, and/or dilated unresponsive pupils when first seen at hospital. Intractable hypotension (36 cases) and respiratory or cardiac arrest (20 cases) were common in this group. Their median survival was only 18 h, with 42 (36 per cent) of them dead within 6 h of injury. None of these 116 deaths was considered preventable. An acute intracranial haematoma was considered to be the main cause of death in 31 patients (11 per cent), but nine of them had presented to hospital only after a delay of 12 h or more. Although the median age was 73 years, four were under 40 years of age. Most (21) of the 31 had an intracerebral haematoma, and only one had an isolated extradural haematoma. This latter patient lay unnoticed for many hours, and was moribund on reaching hospital. Of the 31 patients, 12 (39 per cent) had definitely talked after injury and another four may have done. In three cases the finding of a haematoma at post-mortem had been a surprise. The severity of the head injury had been underestimated in at least 15, seven of whom were aged under 70 years. However, in no more than seven of the 31 cases was it considered that death might have been preventable, taking into account age and the severity of injury (Tuble II). Spontaneous cerebrovascular accidents caused the injuries by falling and the deaths of 13 patients (5 per cent), all of whom were over 70 years of age. None of these 13 deaths was judged to be preventable. Extracranial injuries or complications caused 105 (40 per cent) of the 270 deaths, but only seven of these were judged to be potentially preventable. All seven were aged under 70 years, and their head injury was mild or moderate only (Tub/e 111).Exsanguination from multiple injuries occurred in 28 patients, 20 (71 per cent) of them within 24 h of injury. Of the 28 patients, 13 had a Glasgow Coma Score of 12-15 on
Table I. Principal cause of death of 270 patients who died after a head injury during 1979-1988 and were not transferred to z NW Intracranial factors Primary brain damage Intracranial haematoma Cerebrovascular accident Extracranial factors Extracranial blood loss Pneumonia Major chest injury Pelvic/limb fractures Traumatic quadriplegia Cerebrovascular accident Cor pulmonale Motor neurone disease Kyphoscoliosis End-stage alcoholism No predisposing factor Acute cardiac events/pulmonary embolism Various complications of injury Infection (other than chest) Renal failure Fat embolism Gastric bleeding Long immobilization Coincidental malignant disease Anoxic brain damage from cardiac arrest Cause remains obscure
160 116 31 13 106 28 46 15 6 : 4 1 1 2 14 21 6 2 1 1 1 1 3 1 5
Total
270
TableII. ‘Saveable’ and ‘unsaveable’ patients who died from intracranial haematoma Potentially savable (N= 7)
Unsavable (N=24)
6 (86%)
6 (25%)’
6 (86%)
12 (50%)’
: (86%) 1 (14%)
1 ( 4%): 23 (96%) 0
Age under 70 years Best Glasgow Coma Score after injury over 8 Type of haematoma Extradural lntradural Both
‘Best Glasgow Coma Score was 6 in one case, 3 in the remainder. ‘Aged from 73 to 86 years. :Long delay before hospital; moribund on arrival.
Table III. Potentially saveable patients who died from extracranial injuries or complications Age Case (years)
Cause of death
1.
16
Blood loss
2. 3.
65 42
Blood loss Blood loss
4.
39
Blood loss
5. 6.
52 57
Pulmonary embolism Pulmonary embolism
7.
11
Anoxic brain damage
Remarks Refused surgery for vena cava tear Unrecognized aortic tear Delayed recognition of haemoperitoneum Delayed recognition of haemothorax Multiple injuries, immobilized; no anticoagulant prophylaxis Cardiac arrest under anaesthesia for manipulation of limb fracture
Gentleman et al.: Death in hospital after head injury arrival at hospital, and 11 became hypotensive only after admission, raising questions about the appropriateness of early management. However, even without the head injury, the Injury Severity Score was high (25-50) in all but three very elderly patients. We judged that only four of these 28 deaths were potentially preventable. Predisposing factors were identified in 32 of the 46 deaths from pneumonia, the other 14 patients being aged between 80 and 97 years. Acute cardiac events caused 17 deaths (6 per cent), only three of them in patients under 70 years of age. Four patients died from massive pulmonary embolism after mild head injuries, six from various systemic complications of injury, three from coincidental malignant disease, and one from anoxic brain damage caused by a cardiac arrest during a genera1 anaesthetic for a minor orthopaedic procedure. The only potentially avoidable deaths in this large group were this last case, and fwo young patients with fatal pulmonary embolism who had not received anticoagulant prophylaxis despite being at high risk from immobilization with multiple injuries. In five patients (2 per cent), the cause of death remained obscure. Although clinical evidence suggested an intracrania1 haematoma in four of the five, the youngest was aged 85 years, and it is unlikely that their deaths were preventable. Referral to the NSU The NSU had been consulted by telephone in 123 cases (46 per cent), one-half of them within 6 h of injury; in 16 of these a neurosurgeon had also visited the patient at the general hospital. Transfer to the NSU in all 123 cases had been considered inappropriate by the neurosurgeon, because of the severity of initial injury, the patient’s advanced age, or both. No patient had a CT scan in a genera1 hospital. A significant haematoma was subsequently found at autopsy in 11 of these patients, but only three were aged under 70 years. Historical
trends
among
untransferred
patients
(TubleIV) An earlier unpublished
study of hospital deaths after head injury in Glasgow during 1971-1975 provides a historical control for this series (Strang, personal communication). A smaller proportion of patients (31 per cent of 388) was transferred during that earlier period, and analysis of the 268 untransferred fatalities showed more patients aged under 70 years, and more who survived for over 24 h, than in the present series. Transferred and untransferred patients during 19791988 (Table V) Far fewer of the transferred patients were elderly, fewer had severe multiple injuries, and fewer died within 24 h of their injury. This suggests that appropriate triage was being carried out.
Discussion Only 31 per cent of hospital deaths after head injury in Glasgow in 1971-1975 were in patients who had been transferred to the NSU, compared with 55 per cent during the 10 years after a more liberal transfer policy was introduced in 1978. Although death is a common outcome after severe head injury, and is often the result of overwhelming primary injury, the concept of ‘avoidable deaths’, is useful in auditing
473
Table IV. Head injury deaths in Glasgow hospitals; 1971-1975 and 1979-1988 (Glasgow patients only) 1971-1975
1979-1988
388 129’ (31%) 268 (69%)
606 336’ (55%) 270 (45%)
170 180
121 148
Total deaths (ommitting patients with missing case records) N (%) transferred to NSU N (%) not transferred to NSU Of those not transferred: N (%) aged under 70 years N (%) survived more than 24 h
(66%) (67%)
(45%) (55%)
*About one-half of these had been transferred to the NSU from outside Glasgow.
Table V. Transferred and untransferred fatal cases during 1979-1988 (Glasgow patients only)
Age distribution (years) Under 70 70-79 80-89 90 and over No motor response to pain, dilated unreactive pupils, or spontaneous apnoea on first arrival in hospital Multiple injuries Severe multiple injuries (three or more body regions) Survival less than 24 h
Transferred (N = 336)
Not transferred (N=270)
293 (87%) 36 (11%) 7 (2%) 0 (0%)
121 (45%) 60 (22%) 76 (28%) 13 (5%)
128 (38%) 108 (32%)
104 (38%) 141 (52%)
27 (8%) 77 (23%)
62 (23%) 118 (44%)
the effectiveness of policies for dealing with injured patients (Reilly et al., 1975; Rose et al., 1977; JenneH and Carlin, 1978; Phair et al., 1991). Deciding whether or not a particular death might have been averted by better management is not an exact science, but relies upon a value judgement to distinguish between what might have been achieved under ideal circumstances and what could reasonably have been expected in practical terms. For example, one-third of the patients who died from intracranial haematoma in this series reached hospital only after a delay of several hours, during which some who had been talking deteriorated info coma. Some of these deaths might well have been prevented had they attended hospital sooner, but it would be unreasonable to regard them as preventable by better medical care. Age is an important determinant of outcome after severe head injury or multiple trauma (Teasdale et al., 1979). Operating for intracranial haematoma over age 70 years gives uniformly poor results, unless the patient has been talking after injury, is operated upon before deterioration into coma, and has only an extracerebral clot. Age therefore plays a legitimate part in triage, and it is no surprise that during 1979-1988 there were many more elderly patients among the untransferred cases than among those who went to the NSU. The second major determinant of outcome after head injury is the severity of primary brain damage. At any age, the finding on arrival at hospital of no motor response, dilated unreactive pupils, or spontaneous apnoea is highly predictive of a fatal outcome. The third major influence on outcome - and the only one which can be corrected - is the severity of secondary
474
Injury: the British Journal of Accident Surgery (1992) Vol. 23/No. 7
hypoxic-ischaemic insults to the brain and other organs. This caused 14 deaths in patients aged under 70 years which might have been averted by different management after admission to hospital. These cases include a boy who had a cardiac arrest under anaesthesia for manipulation of a forearm fracture, two fatal cases of pulmonary embolism in multiply-injured men who were not given prophylactic anticoagulation, and four who died from unheated concealed thoracic or abdominal blood loss (one was a teenage motorcyclist who refused life-saving abdominal surgery despite warnings of the likely consequences). This leaves only seven patients under 70 years of age who had not had overwhelming primary injury to the brain or other organs, and who died from unsuspected intracranial haematomas. In the period during which these seven potentially salvageable cases were not transferred, some 3000 head-injured patients were transferred from the same hospitals, and some 600 of these had operations for significant haematomas, with an overall mortality of around 30 per cent. The system for identifying patients who should be transferred therefore appears to be working reasonably effectively. Delayed referral of patients is known to be a major contributor to mortality, as well as to permanent disability in some survivors (Mendelow et al., 197% Seelig et al., 1981). In the west of Scotland much work has gone into developing and refining a policy fo encourage early referral of appropriate patients to the NSU, and during 1979-1988 many fewer potentially salvageable patients were not transferred than in 1971-1975. Maintaining this policy has required considerable and continuing effort in education and publicity as junior staff change in the various referring hospitals. Despite these efforts, a few patients who might benefit from neurosurgical services are still not transferred, and it is a maHer for concern that avoidable death from secondary brain damage caused by compressive intracranial haematomas still occurs in the same city as a regional NSU. There is an urgent need for more systematic audit of head injury management, both in general hospitals and in NSUs nationwide. This could establish whether policies or practices need to be reviewed to minimize avoidable death and disability. It is important that consulfants in accident departments and those in charge of primary surgical wards should liaise regularly with their colleagues in the regional NSU, who in turn must be prepared to agree locally appropriate criteria about which patients call for consultation with the NSU, for possible transfer. Transfer should only happen after consultation, and after adequate steps have been taken to minimize adverse incidents en route that might add to the brain damage (Gentleman and Jennett, 1990). As most head-injured patients are initially managed by doctors in training, who change at frequent intervals, it is important that they be taught and know what is expected of them.
Acknowledgements The authors gratefully acknowledge the help they received from many individuals. Dr John Bryden and MS Barbara Boyd of the Management Information Services Division of Greater Glasgow Health Board provided information to identify patients for the study. The Medical Records Officers of Glasgow Royal Infirmary, the Western Infirmary/Gartnavel General Hospital, the Victoria Infirmary, Stobhill Hospital, the Southern General Hospital, and Royal Hospital for Sick Children helped in the location of
hospital case records, and the consultants in these hospitals gave permission for the records of their deceased patients to be examined. The Crown Office, the Procurator Fiscals of Glasgow and Strathkelvin, Dumbarton, and Greenock, and the forensic pathologisb who carried out autopsies on these patients allowed access to their records. Dr Ian Strang collected the data on 1971-1975 head injury fatalities which were used as a comparison with this study. Dr Lilian Murray PhD provided data on head-injured patients transferred fo the NSU during 1979-1988. Miss Anne Semple prepared the final drafts ot’ the manuscript. This study was part of an investigation into avoidable mortality and morbidity after head injury funded by the Scottish Home and Health Department.
References A Group of Neurosurgeons (1984) Guidelines for initial management after head injury in adults. Br. Med. j. 288,983. Anderson I. D., Woodford M., De Dombal F. T. et al. (1988) Retrospective study of 100 deaths from injury in England and Wales. Br. Med. 1. 296, 1305. Baker S. P., O’Neill B., Haddon W. et al. (1974) The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care. 1. Trauma 14, 187. Bryden J. S. and Jennett B. (1983) Neurosurgical resources and transfer policies for head injuries. Br. Med. 1. 286, 1791. Gentleman D. and Jennett B. (1990) Audit of unconscious head injured patients transfeerred to a neurosurgical unit. Lancet 1, 330. Jeffreys R. V. and Jones J. J. (1981) Avoidable factors contributing to the death of head injury patients in general hospitals in Mersey Region. Luncef 2, 459. Jennett B. and Carlin J. (1978) Preventable mortality and morbidity after head injury. I+ry 10,31. Jennett B., Murray A., Carlin J. et al. (1979) Head injury in three Scottish neurosurgical units. Br. Med. J 2, 955. Mendelow A. D., Karmi M. Z., Paul K. S. et al. (1979) Extradural haematoma effect of delayed treatment. Br. Med. 1. fi), 1240. Phair I. C., Barton D. J. Barnes M. R. et al. (1991) Deaths following trauma: an audit of performance. Ann. R. Co/l. Swg. Engl. 73,53. Reilly P. L., Adams J. H., Graham D. I. et al. (1975) Patients with head injury who talk and die. Lance/ 2,375. Rose J., Valtonen S. and Jennett B. (1977) Avoidable factors contributing to death after head injury. Br. Med. J 2,615. Seelig J. M., Becker D. P., Miller J. D. et al. (1981) Traumatic acute subdural haematoma; major mortality reduction in comatose patients treated within 4 hours. N. Engl. J. Med. 304, 15 11. Teasdale G., Galbraith S., Murray L. et al. (1982) Management of traumatic intracranial haematoma. Br. Med. J 285, 1695.
Teasdale G. and Jennett B. (1974) Assessment impaired consciousness. Lancet 2, 81.
of coma and
Teasdale G.. Murray G., Anderson E. et al. (1990) Risks of acute traumatic intracranial haematoma in children and adults: implications for managing head injuries. Br. Med. J. 300, 363. Teasdale G., Skene A., Parker L. et al. (1979) Age, severity and outcome of head injury. A&a Neurocbir. Suppl. 28, 140.
Paper accepted 24 February
1992.
Requests for reprints shouM be addressed to: Mr D. Gentleman,
Consultant Neurosurgeon, Department of Neurosurgery, Dundee Royal Infirmary, Dundee DDl 9ND, UK.
