Results

The details of the literature search are outlined in figure 1. One hundred and forty-four full text articles were identified for further analysis. Ten were excluded because the full text version could not be located, including three online reports that are no longer available. Eighty-six did not meet the entry criteria on closer scrutiny, most of which related only to minor head injuries or the challenges of implementing the guidance on CT scanning.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Selection of articles to review.

Fifty-eight papers were reviewed. This includes the seven papers cited by NICE in their review, which was limited to outcomes. Fifty-one further documents related to the provision of neurosurgical services, including nine on bed occupancy, eight on travel times to RNUs and 20 on transfer policies.

NICE guidelines

NICE undertook systematic literature reviews to identify papers relevant to two aspects of emergency neurosurgical care: direct transfer from the scene to a RNU, and secondary transfer of all patients with severe head injuries from DGHs to a RNU, irrespective of their need for surgery.

Direct transfer

According to NICE, a policy to transport all TBI patients directly to a specialist centre would require a shift of an additional 84 000–105 000 bed days to neurosurgery from the existing services that care for this group. They concluded that there is insufficient evidence to support this currently.

The two papers NICE cited to support this are summarised in table 1. Both are based on data from US trauma centres. Hannan et al compared mortality in patients in New York State taken directly to a trauma centre (with neurosurgical facilities), versus those taken to a non-trauma centre.17 Mortality rates were in favour of direct transfer. The second paper demonstrated that paediatric patients intubated in the field had worse outcomes than those intubated in hospital.18 However, the study was designed to examine the effect of place of intubation, not direct transport, as NICE acknowledged.

Secondary transfer

There was no uncertainty about the transfer of patients requiring surgery, but the GDG drew on four studies (see table 1) to recommend that all patients should ideally receive treatment in a neurosurgical unit, irrespective of their need for an operation. The first paper was a study of head injuries treated in hospitals contributing data to the UK Trauma Audit and Research Network (TARN).20 The case-mix adjusted odds of death for TBI patients in non-neurosurgical centres was 2.15 times that of those who received specialist care; mortality rates were 26% higher. However, it was not clear which aspects of specialist neurointensive care improved outcome. The remaining papers NICE cited described results from other countries to support the case that all patients should receive specialist care.17 21 22

The generalisability of the non-UK studies is questionable. For example, the paper by Hannan17 described care in New York State, where there is approximately four times the density of RNUs compared to the UK, so travel times are likely to be much shorter.23 The UK also remains the only First World country not to have a system of designated trauma centres currently.

The GDG called for more research on both transfer options, but the papers they cited highlight the difficulties of conducting rigorous research in this field. There is a paucity of randomised trials, and the emergency medicine literature is dominated by observational cohort studies and small case series, which may be subject to confounding and bias.

The aim of NICE guidance is to maximise the quality and consistency of patient care, but this document has been criticised because the evidence base for many of the other recommendations is limited and from non-UK settings.24 Resource availability was also outside their terms of reference,25 prompting calls for a ‘more realistic’ alternative taking this into account.26

Neurosurgical facilities

Only 53% of TBI patients who receive care at a DGH are transferred to a RNU.20 Some may not be suitable for further intervention, but would it be feasible for more patients to be transferred to receive specialist neurosurgical care, particularly when over half would not require surgery?

Only 9% of DGHs are able to transfer all their TBI patients for specialist care, due to bed shortages and local policies.27 Admission to neurosurgical units is currently governed by the availability of facilities, and patients with operable lesions are understandably given priority, but this has important implications for the implementation of the NICE guidance.

According to an audit carried out for the Department of Health in 2005, neurocritical care bed occupancy was consistently over 90%, with demand outstripping supply by 9–22% across units in England and Wales. Twenty-three per cent of head injury patients were managed in non-specialist critical care units at that time.28 In another survey,9 only 43 neurocritical care beds were available nationally - far from sufficient to cope with the projected 28.6 head injuries per day, when trauma forms only part of the neurosurgical workload. The same study demonstrated large regional variations in bed availability.

Neurosurgical workload increased by more than 3% annually between 1993 and 1999, but this was not matched by investment in facilities.29 In 1993, the SBNS proposed a target of 30 neurosurgical beds per million population as the minimum safe standard for 2000, but only five out of 37 units met this target by 1999, with some working at less than 15 beds per million. The same document also proposed a target of four dedicated neurosurgical critical care beds per million population, but only four out of 37 units met this target by 1999. There was again wide regional variation in the availability of both types of bed and, where resources are poorest, emergency work was carried out at the expense of waiting list cases, threatening equity of access to both trauma and elective patients.3

Twenty per cent of neurosurgical bed days were consumed by patients who did not require acute management.10 In one case, a lack of beds led to 33% of elective cases being cancelled over 1 month.30 The main reasons were delays in repatriation to the referring hospital, accessing theatre space or undergoing radiological investigations. The problem was compounded by the poor availability of rehabilitation services.31

Thirty per cent of Emergency Department (ED) consultants were unhappy about difficulties they faced in arranging referrals.32 However, there have been some successful attempts to strengthen the links between RNUs and the hospitals that they serve, notably in the Eastern region of England,33 with policies in place in other areas to facilitate the swift transfer of all appropriate patients, regardless of RNU bed status.

In the National Confidential Enquiry into Patient Outcome and Death (NCEPOD) review of trauma care provision in 1999, 16.2% patients with moderate or severe head injuries had no evidence of neurosurgical referral, let alone transfer. However, the authors acknowledge that within this group it is possible that consultation was not undertaken due to the extent of injuries.34

Geography and transfer times

The report Safe Neurosurgery 2000, called for a maximum surface journey time to a neurosurgical unit of 2 h, if equity of access is to be maintained.3 However, the geographical distribution of services in the UK has evolved due to local circumstances, rather than central planning.

The pathophysiology of traumatic brain injury is described in terms of primary and secondary injury. Primary injury is the damage sustained by the brain at the moment of impact. Secondary injury is any detrimental event that occurs subsequently, including hypotension and hypoxaemia, so adequate resuscitation is vital. The time taken for transfer impacts on the time to definitive management of the primary injury and the risk of secondary insults. Patients with an operable injury have a better outcome if their lesion is surgically evacuated within 4 h.35

Although 33% of EDs were within 10 miles of the nearest neurosurgical unit, 12% were over 50 miles away and the national average was 23 miles, according to the Galasko report in 1999.14 Table 2 shows four further reports that described the range of regional units. Travel times meant that in one survey only six out of 43 (14.0%) patients transferred secondarily had an operation within 4 h of injury, compared with 22 out of 33 (66.7%) taken directly to the neurosurgical centre.34

There are also wide variations in the range of specialist centres nationally. In the catchment areas for urban centres, patients are a median distance of 10–15 km away, whereas in largely rural areas patients are a median of 50 km away. Some centres operate between these extremes, with a mixed population a median distance of 20 km away, but an IQR of 5–80 km.39 Such variations mean that achieving equity of access even with respect to the same centre is difficult, particularly in rural areas. There are no data about whether distance from the RNU affects the likelihood of patients being transferred, or about the impact of transfer times in patients with non-operable brain injuries.

Delays in transfer

Delays in patient transfer are not solely due to travel times, but often also occur at the DGH, either from failure to institute appropriate treatment for non-cranial injuries or in realising transfer was necessary.40 Delays in CT scanning and time waiting for an ambulance are also important.41

Twenty per cent of head injury patients were still not scanned in a timely fashion, even though 95% of hospitals now have 24-h access to CT imaging. Delays were primarily due to organisational factors, such as waiting for staff or access to the CT scanner.34 Moreover, a recent study demonstrated that no single step in the transfer process was responsible for all delays, nor were any single steps consistently performed within an acceptable time period.37

Systems to electronically transfer CT images between centres have also reduced the number of unnecessary transfers, but 13% of hospitals without on-site neurosurgery still do not have effective image transfer systems, so scans must be sent by courier, causing further delays.27

Hazards of secondary transfer

Although often necessary, transfer is not without risk, as NICE acknowledged. Avoidable factors contributing to death in patients transferred to a RNU have been reported for many years, highlighting the need for adequate resuscitation prior to departure.42–44 As the management of a head injury often takes precedence, the initial first-line management of other injuries such as long bone fractures can be unsatisfactory.45–49 The problem seems to have persisted, despite the implementation of the Advanced Trauma Life Support (ATLS) guidelines.41 However, most of the papers about the risk of transfer again involved a limited number of patients, and data from only one regional centre.

In another study, 63% of transfers required further intervention at the second hospital to complete resuscitation.50 Several other papers identify problems with the care of patients being transferred, particularly as transfers often take place out of hours51 and vital equipment may not be available.52 The NCEPOD trauma audit described arrangements for transfers as ‘haphazard,’ with deficiencies in local protocols, use of national guidelines, consultant oversight and documentation.34

If more patients with TBI are to benefit from neurosurgical care, closer attention must be paid to adequate resuscitation and assessment prior to transfer. Consideration also needs to be given to staff training and the availability of suitable equipment.53