Injury (1991) 22, (5), 353-356
Printed in Great Britain
Preventable deaths after head injury: a clinical audit of performance I. C. Phair, D. J. Barton, M. J. Allen and M. R. Barnes Accident and Emergency
Leicester Royal Infirmary, Leicester, UK
A retrospective frauma audit of 267 comecufive patients presenting fo fk resuscitation room ofan accident and emergency department idenfified 107 (M= 0.73) patients with CNS predominant injuries in a period of 1 year. The mortality rafe was 33 per cent. Application of fk TRISS method indicated that 18 (51 per cent) of fk 35 deaths (Z = 5.74, P< 0.001) were mathematically unexpecfed; no unexpected suroivors were identified. Subsequenf peer review agreed fhaf death was preven fable in 11 cases. The peer panelsu=esfed that fk Lackof an on-site nmrosurgical service may have ate&d fk oufcmne of this group of patients adversely.
Introduction The differential survival of patients with head injuries has been recently emphasized by a large review (Gennarelli et al., 1989). Conclusions were that head injury significantly increased mortality after major injury and that head injury is now the most important cause of death in patients treated in American Trauma Centres. Deficiencies in the care of patients with major injuries in the United Kingdom have been identified (Anderson et al., 1988) indicating a need for improved trauma care and audit. The Glasgow Coma Scale (GCS) (Teasdale and Jennet, 1974) is widely used as a measure of level of consciousness after head injury. Admission GCS has been shown to correlate with the severity of the head injury and subsequent outcome (Luerssen et al., 1988). However, among subsets of patients with the same GCS, differential survival rates dependent on the nature of the anatomical brain injury exist. Thus, the type of brain injury is as important as the GCS score in outcome determination (Gennerelli et al., 1982). The TRISS methodology (Boyd et al., 1987) attempts to address this problem by inclusion of the Injury Severity Score (ISS), (Baker et al., 1974, and physiological parameters weighted according to the Multiple Trauma Outcome Study (MTOS) US, data (Boyd et al., 1987; Champion et al., 1989). Calculation of the probability of survival, identifies mathematically unexpected outcomes of survival and death, the latter being the subject of critical peer review. This allows an assessment of the level of trauma care being delivered at any one institution (Boyd et al., 1987; Copes et al., 1989). As a teaching hospital, the Leicester Royal Infirmary (LRI) receives all patients with major injuries after accidents within its district boundary. However, although the facilities for 0 1991 Butterworth-Heinema 0020-1383/91/050353-04
emergency computed tomography are available on site, the regional neurosurgical unit is situated some 26 miles away. It was therefore thought appropriate to audit our management of head injuries in the light of this local specialty deficiency.
Method Patients who were admitted to the hospital via the resuscitation room of the accident and emergency department during a 12-month period (1 July 1987 to 30 June 1988) were studied retrospectively. Information on the patients’ injuries was collated from the clinical, radiological, and surgical details recorded in the hospital charts. Where death had resulted, additional post-mortem data were used to corroborate the injuries described. Anatomical injuries were scored according to The Abbreviated Injury Scale (1985) of the American Association of Automotive Medicine. Injuries were defined as CNS predominant when the Abbreviated Injury Scale code of the head injury was the highest or equal highest component of the patient’s Injury Severity Score. The Revised Trauma Score (Boyd et al., 1987; Champion et al., 1989) was derived from records of Glasgow Coma Scale, systolic blood pressure and respiratory rates recorded in the accident and emergency notes. On occasions where physiological records were absent, normal values were assumed to prevent patient exclusion (Harviel et al., 1989). Probabilities of survival together with M and Z statistics were calculated as described by Boyd et al. (1987). Deaths were judged mathematically unexpected in those patients who died in whom the P, value as derived by TRISS was 0.50 or more. Subsequent peer review was then undertaken by a panel comprising six consultants. The specialties represented were: orthopaedic, general and vascular surgery, accident and emergency, intensive care and neurosurgery. With the exception of the neurosurgeon, all of the peers were based at the LRI. Each member of the panel received patient summaries which included the mechanism of injury, initial physiological recordings, resuscitation room procedures, and fluids administered. Details of operations and grade of staff involved were provided, together with relevant time sequences, in-hospital care and post-mortem findings. They were asked the question: ‘On consideration of optimum standards within a system of trauma care, do you believe this death was preventable or non-preventable?‘. If they
Injmyz the British Journal of Accident Surgery (1991) Vol. 22/No.
ISS group Figure 1. Injury Severity Scores for: 0, survivors; and Cl, deaths.
Table I. Outcomes for 107 patients studied
Table II. Outcome related to admission GCS
80 27 107
0.85’ 0.43’ 0.73’
61 11 72
19 16 35
4.12* 4.052 5.742
Except for the two cases mentioned above, of the 35 patients who died, 33 had post-mortem examinations; thus pathological diagnoses were available. All five patients with extradural haematomata survived; decompressive surgery was effected without transfer to the Regional Neurosurgical Unit in four patients; however, a neurosurgeon was present for three of those procedures. A post-mortem diagnosis of diffuse axonal injury was made for 10 patients, all of whom died. The majority of
Number of patbents
Figure 4. Main anatomical diagnosis of intracranial injuries, for: W, survivors; and 0, deaths. ICH - intracerebral haematoma. EDH - Extradural haematoma. SDH - Subdural haematoma. DA1
- Diffuse axonal were reviewed. The opinion that death was preventable was made by at least one peer in 11 of the 16 cases reviewed, a preventable death rate of 31.4 per cent. Associated extracranial injury of a maximum AIS score of 2 was noted in six of the II deaths judged preventable by the peer panel. Those injuries were to either the extremity or external regions. Therefore, it may be concluded that preventable deaths were attributable to the CNS injury in those 11 patients. The level of agreement that death was preventable as judged by TRISS and peer review decreased with increasing severity of injury as defined by admission GCS score (TableII). Errors made in the clinical management of the patients as identified by the peer group are shown (TableD). Mistakes were identified in all areas from initial primary assessment to surgical intervention. Mismanagement was deemed to have occurred in eight of the 11 preventable deaths. Recommendations by the peer panel to improve the standard of care included better access to, and communication with, senior neurosurgical opinion. In particular our inability to transfer CT scan images to the Regional Neurosurgical Centre was severely criticized.
patients with subdural or intracerebral haematomata died (Figure4). Computed tomography was delayed until deterioration following admission on four patients with skull fractures, all of whom had subdural haematomata. From analysis of admission GCS scores, three of those four patients, ‘talked and died’.
Discussion Data from the Multiple Trauma Outcome Study (US) state that CNS predominant injuries account for one-third of admissions to participating Trauma Centres and two-thirds of the deaths (Gennarelli et al., 1989). This is similar to the findings at this hospital. However, a direct comparison cannot be made due to our small numbers and low value of the A4 statistic. This is in spite of our catchment population of approximately I million inhabitants. Patient outcome, as specified by survival, has also been shown to relate to injury severity defined by admission GCS. An earlier investigation from this hospital on the management of severe head injuries reported a mortality rate of 52 per cent for patients with admission GCS scores between 3 and 8 (Miller et al., 1985). The 67 per cent mortality rate for that group of patients in this study may represent an unfavourable comparison. A suggested modification to the guidelines on the management of patients with recent head injury has been suggested by Teasdale et al. (1990). The conclusion that the presence of an intracranial haematoma is more likely when the skull is’ fractured has reaffirmed an earlier opinion from Glasgow, that the presence of a skull fracture is an adequate
Injury: the British Journal of Accident Surgery (1991) Vol. 22/No. 5
reason for computed tomography (Teasdale et al., 1982). The adoption of a ‘if sM1 fracture, CT’ policy may have led tp an earlier diagnosis of four acute subdural haematomata. mhough acute subdural haematomata in patients who present in coma is associated with a mortality rate in excess of 70 per cent (Gennerelli et al., 1982), three of our patients who talked and died may have been saved. Peer opinion indicated that a locally based neurosurgeon may have improved the outcome of those in whom death was defined as preventable. Despite the presence of a CT scanner in the accident and emergency radiographic department and relative ease of obtaining a scan at any hour of the day or night, preventable deaths were still identified. Although the report of the working party of the Royal College of Surgeons (1988) on the management of patients with major injuries levels criticism at the management of trauma in the UK, it appears from our peer comments that we have yet to practise in full the recommendations of a previous report on head injuries (RCS Working Party, 1986). Implementing the recommendations of the latter report may, in part, prove of greater benefit to the local management of head injuries than the establishment of trauma centres in selected locations. The preventable death rate measured by peer review did not equate with that indicated by application of the TRISS definitiorl in this analysis. A predictive value of the TRISS methodology in peer review preventable death designation of 54 per cent has been reported (Pories et al., 1989). As stated by Champion et al. (1990), ‘no numeric characterisation infallibly predicts P, = 1 for survivors and F’,= 0 for deaths, clinical acumen remains an essential component of trauma quality assurance’. In accepting imperfections of the TRISS methodology, it must be remembered TRISS identifies problem cases, and final arbitration is achieved by peer review (Boyd et al., 1987; Copes et al., 1989). As an audit tool, TRISS is to be recommended, for it is through critical audit that mistakes may be identified and lessons learned (Dearden and Rutherford, 1985).
References American Association for Automotive Medicine. The Abbreviated Injury Scale. (1985) revision. Des Plaines, Illinois. Anderson I. D., Woodford M., De DombaI F. T. et al. (1988) Retrospective study of 1000 deaths from injury in England and Wales. Br. Med. J 296, 1305.
Baker S. P., O’Neill B., Haddon W. et al. (1974) The injury severity score: development and potential usefulness. Proceedings Arnerican Associafion forAutomotive Medicine l&58. Boyd C. R., Tolson M. A. and Copes W. S. (1987) Evaluating trauma care: The TRISS method. J. Trauma 27, 3 70. Champion H. R., Sacco W. J., Copes W. S. et aI. (1989) A revision of the trauma score. J. Trauma 29, 623. Champion H. R., Copes W. S., Sacco W. J. et al. (1990) A new characterization of injury severity. J Trauma 30,539. Copes W. S., Sacco W. J. and Champion H. R. (1989) Guest Editorial. Arch. berg. Med. 6, 165. Dearden C. H. and Rutherford W. H. (1985) The resuscitation of the severely injured in the Accident and Emergency Department - a medical audit. Injtny 16,249. Gennarelli T. A., Spiehnan G. M., Langfitt et al. (1982) Influence of the type of intracranial lesion on outcome from severe head injury. J Neurosurg. 56, 26. Gennarelli T. A., Champion H. R., Sacco W. J. et al. (1989) mortality of patients with head injury and extracranial injury treated in trauma centers. 1. Trauma 29, 1193. Harviel J. D., Landsman I., Greenberg A. et al. (1989) The effect of autopsy on injury severity and survival probability calculations. ]. Trauma 29, 766. Luerssen T. G., Klauber M. R. and Marshall L. F. (1988) Outcome from head injury related to patient’s age. 1. Neurosurg. 68,409. Miller E. S., Neoptolemos J. P., Aitkenhead A. R. et al. (1985) Management of severe head injuries in a non-neurosurgical trauma centre. J. R. Coil. Surg. Edinb. 30, 82. Pories S. E., Gamelli R. L., Pilcher D. B. et al. (1989) Practical evaluation of trauma deaths. 1. Trauma 29, 1607. Royal College of Surgeons of England. (1988) Commission on the provision of surgical services. Report of the Working Party on the Management of Patients with Major Injuries. London. Royal College of Surgeons of England. (1986) Commission on the provision of surgical services. Report of the Working Party on Head Injuries. London. Teasdale G. and Jennet B. (1974) Assessment of coma and; impaired consciousness. A practical scale. hcet 2, 81. Teasdale G, Galbraith S., Murray L. et al. (1982) Management of traumatic intracranial haematoma. Br. Med. J. 295,1695. Teasdale G. M., 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.
Requests for reprints shouti be aailressed to: Mr I. C. Phair FRCS, Consultant in Accident and Emergency Medicine, North Staffordshire Royal Infirmary, Hartshi& Stoke on Trent ST4 7JW, UK.