Safe management of paediatric penetrating head injury without a CT scanner: A strategy for humanitarian surgeons based on experience in southern Afghanistan

Abstract

Introduction

In many parts of the world, access to a CT scanner remains almost non-existent, and patients with a head injury are managed expectantly, often with poor results. Recent military medical experience in southern Afghanistan using a well-equipped surgical facility with a CT scanner has provided new insights into safe surgical practice in resource-poor environments.

Methods

All cases of children aged under 16 years with penetrating head injury who were treated in a trauma unit in southern Afghanistan by a single neurosurgeon between 2008 and 2010 were reviewed. Based on a previously published retrospective review, a clinical strategy aimed specifically at generalist surgeons is proposed for selecting children who can benefit from surgical intervention in environments with no access to CT scanners.

Results

Fourteen patients were reviewed, of whom three had a tangential wound, 10 had a penetrating wound with retained fragments and one had a perforating injury. Two operations for generalist surgeons are described in detail: limited wound excision; and simple decompression of the intra-cranial compartment without brain resection or dural repair.

Conclusions

In resource-poor environments, clinically-based criteria may be used as a safe and appropriate strategy for selecting children who may benefit from relatively straightforward surgery after penetrating brain injury.

The aims of neurosurgical intervention in penetrating head injury are the prevention of infection (as with all war wounds) and the prevention/treatment of critically raised intracranial pressure (ICP) in patients who will benefit from the procedure.^1–5 The use of imaging is an integral part of decision-making in the management of these patients, and the information provided by computed tomography (CT) remains the ‘gold standard’.^6–9

In an area of conflict, it may be impossible to establish the exact timing and circumstances of a head injury, and the results of clinical assessments and nature of any medical interventions (such as the administration of drugs) ‘en route’ to the surgical facility may be unknown. A decision to operate is therefore often based on clinical examination on arrival at the medical facility (if the patient is not sedated/intubated and ventilated) and a subsequent CT scan.¹⁰

Without a CT scan, it is not possible to accurately localise and estimate the size of intracranial focal traumatic lesions; it is instead left to clinical examination alone, comprising principally of assessment of the level of consciousness and limb weakness, and a head examination for signs of external injury to determine if an operation is indicated. The decision also takes into account the haemodynamic status and stability of the patient; neck and torso injuries may mandate either treatment-limiting decisions, or immediate surgery that takes priority over neurosurgery to control life-threatening bleeding from these areas.^11–14

In cases of penetrating head injury, clinical examination can also determine the site of proposed operation, unlike blunt trauma, where, in the absence of CT, exploratory burrholes are required. However, the decision may be complicated by the presence of both entry and exit wounds.

It is generally accepted that, even without the availability of a CT scan, some form of operative intervention is appropriate in patients judged to be salvageable. Much of the evidence on which this is based dates from before the introduction of the CT scanner, often involving large numbers of patients treated by military personnel.^15–18 Today, protocols for operative intervention in cases of penetrating head injury in circumstances where a CT scan is not available are promulgated by, amongst others, the British and American Military, and the International Committee of the Red Cross.^19–22

We propose a novel, purely clinical, method of selecting paediatric patients for either wound excision surgery or simple decompressive surgery, based on a published series of paediatric penetrating head injuries treated in southern Afghanistan in a well-equipped trauma unit by PM.¹⁰ The surgery involved is described in detail, and may involve delayed cranioplasty, but can be carried out by non-specialists using basic equipment and facilities.

Methods

Data were reviewed for all children under the age of 16 who had a penetrating head injury and a Glasgow Coma Score (GCS) motor response of localising to painful stimulus or better,²³ and who were transferred to a trauma unit in Kandahar, southern Afghanistan, when PM was the sole neurosurgeon for the region. The study periods were 15 September–17 December 2008, 1 October–30 November 2009 and 25 January–18 April 2010, at a total of 237 days. The trauma unit was equipped with two CT Scanners, three operating theatres and an intensive care unit. All medical, nursing and ancillary staff were fully trained, and provided by members of the International Security Assistance Force.

CT scanning and neurosurgery were performed following resuscitation and stabilisation according to ATLS and DSTS principles.^24–26 Operating theatre staff at the trauma unit recorded the type of operation(s) performed on a spreadsheet, using an anonymous patient identifying number.

Results

A total of 14 patients were reviewed (mean age 9 years (2–15); 12 male:2 female). Six were fully conscious on initial examination, and remained so after serial assessments. In this group, three patients had a tangential wound, and three had a ‘penetrating with retained fragment’ pattern of injury. All patients in this group were treated with a limited wound excision, irrespective of CT findings (Figure 1).

Figure 1.

Fully conscious patients with no neurological deficits. (A) Tangential wound. Entry wound right frontal region (to the left of the picture); larger exit wound right parietal region. Patient fully conscious with no focal neurological deficit. (B) 3D reconstruction CT of the patient in A shows the boney disruption along the path of the bullet. Axial views showed minimal damage to the adjacent brain. (C) Wound excision in patient A, with elevation of larger bone pieces. (D) Penetrating wound with retained fragments. CT shows a ‘benign’ track. Although crossing the midline, the narrow haemorrhagic track has no associated brain swelling. (E) Penetrating wound with retained fragments. ‘Peppering’ of the face and scalp by small fragments, showing how difficult it can be to identify a penetrating wound. (F) Axial CT of the patient in E shows a retained fragment just underneath the skull in the adjacent brain.

Initially, the whole head was shaved to identify scalp wounds that may represent an entry wound. The midline and other parts of the skull, under which lie major venous sinuses, and the frontal air sinuses were regarded as ‘danger areas’ and avoided (Figure 2). In these areas, exploratory surgery was therefore limited to incision (lengthening of the scalp wound) and gentle irrigation. No attempt was made to remove of any bone pieces. In other areas of the head (principally the lateral convexity of the skull), the wound was incised down through to the galea to expose the underlying skull.

Figure 2.

‘Danger’ areas. (A) Frontal region, containing frontal air sinuses and the underlying anterior superior sagittal sinus. (B) Midline, with the underlying superior sagittal sinus, widening from anterior to posterior. (C) Posterior view of the skull, with the underlying confluence of sagittal and transverse sinuses (covered by neck musculature).

For small entry wounds with only a ‘slot’ fracture, obvious foreign material was removed by copious irrigation; in larger wounds, pieces of skull were gently lifted and the larger pieces replaced once the underlying dura had been irrigated. In tangential wounds not in danger areas, the entry and exit wounds were joined by a linear scalp incision to provide direct access to the underlying skull fragments (Figure 1A–C). If removal of bone in any wound revealed an underlying extradural blood clot, this was gently washed away using a large syringe, resisting the temptation to: (a) scrape the residual clot off the dura (which can lead to active bleeding); or (b) incise the dura to exclude the possibility of an associated acute subdural haematoma.

Following irrigation, any persistent bleeding was controlled by placing a saline-soaked swab on the wound for a few minutes. No attempt was made to close any dural defects. However, the scalp was closed at the end of the procedure in all wounds; an essential final step in this limited operation.

Eight patients had a localising response to painful stimulus (in effect, corresponding to a GCS of greater than 10) with or without associated lateralising signs in the form of mild-to-moderate limb weakness. In this group, seven patients were in the ‘penetrating injury with retained fragments’ category, with a single, and usually obvious, entry wound (Figure 3). One patient sustained a ‘perforating’ pattern of injury (ie a single entry and exit wound). In all cases, CT scanning revealed a well-defined linear haemorrhagic track with moderate mass effect (Figure 4).

Figure 3.

Penetrating wound with retained fragments. Obvious right frontal entry wound, patient localising to painful stimulus with a mild right-sided weakness.

Figure 4.

Penetrating wound with retained fragments (patient localising to pain). CT scan shows a linear track.

These patients were treated with a decompression procedure. To gain access to the brain surface, a scalp flap and bone flap were fashioned based on the entry wound and the major venous sinuses and frontal air sinuses were avoided, as follows. After identifying the entry wound, the head was positioned and cleaned, and the sites of four proposed burrholes roughly equidistant from the entry wound were marked on the scalp. A proposed incision to join these four points and preserve the blood supply to the scalp flap was marked and infiltrated with local anaesthetic (Figure 5A–C). The exception to this rule was an entry wound in the temporal region, which was exposed by a linear incision (Figure 5D) and a self-retaining retractor used to expose the skull. In this situation, a single burrhole placed next to the entry wound was used as a starting point for piecemeal removal of the thin temporal skull using ronguers in a manner analogous to decompression of a temporal extradural haematoma. When further preparing the head, one drape was placed along the midline to provide a constant per-operative reminder of the presence of the sagittal sinus.

Figure 5.

(A–C) Planning of scalp incision based on four burrholes placed equidistant from the entry wound. (D) Linear scalp incision for a temporal entry wound.

The full thickness of the scalp was incised, haemostasis was secured with artery clips and the resulting flap was turned back on its base and secured. The four burrholes were then created at their proposed sites using a Hudson Brace. The extradural space beneath the skull between the burrholes was developed with a dissector to allow the passage of a Gigli saw-guide, and a Gigli saw used to join up the four burrholes. The resulting free bone flap was lifted to reveal the underlying dura, with the entry wound in the approximate centre of the field. Clot (often mixed with small bone fragments) was seen extruding from the dural wound rather like toothpaste from a tube, either immediately after lifting the bone flap or after a minute or two of waiting (Figure 6). This process was facilitated by gently enlarging the dural entry wound in a cruciate-like manner.

Figure 6.

Spontaneous emergence of clot through a dural entry wound from an underlying track haematoma. This occurred approximately 30 seconds after lifting the boneflap.

The only subsequent action was to copiously wash the area, including any resulting cavities in the brain substance (ie the track of the missile) left by the release of the haematoma. Residual clot lying in the depths of the track was left. It was unusual to find active bleeding from the (usually contused) brain surrounding the track. If there was any bleeding, it was easily controlled by placing a saline-soaked swab on the area for a few minutes. No attempt was made to close the dura.

In two patients, the boneflap was simply placed in position (not secured), and the scalp flap closed with a haemostatic running vicryl suture to the deep part of the scalp (ie the galea). Time-consuming attempts to stop bleeding from the scalp flap edges, once the haemostats were removed, were therefore avoided. No drains were used, and the wound was dressed with a standard head bandage to provide gentle pressure for a day or two.

Delayed cranioplasty

Despite decompression of the track at the end of the procedure, in six patients mild brain swelling caused the replaced bone flap to ‘ride high’, preventing scalp closure. However, it was always possible to close the scalp by leaving the bone flap out of the closure.

The bone flap was subsequently stored in an anterior abdominal wall subcutaneous pocket (Figure 7A), adding only 5–10 minutes to the operating time, and was easily retrieved at a later date (within 5–10 weeks) for a delayed cranioplasty. This latter procedure involved re-opening the scalp wound and placing the bone flap in position on the wound defect, over which a dura-like covering had formed since the initial operation (Figure 7B). If possible, the flap was secured by passing vicryl through holes pre-drilled in the flap (ie away from the operative field) and then passed through the periosteum of the surrounding skull (thereby avoiding the risks associated with drilling holes). The scalp was closed in the usual manner.

Figure 7.

(A) Storage of boneflap in a (left) anterior abdominal wall subcutaneous pocket. (B) Delayed (2 months) cranioplasty. Despite leaving the dura widely open at the initial operation, a dura-like covering has formed at the site of the bone defect.

Following initial surgery, all patients were woken up immediately postoperatively and discharged fully conscious with resolving neurological deficits (if present preoperatively) within 3 days of surgery. Five patients returned for a delayed cranioplasty procedure, with one lost to follow-up. All patients were fully conscious at this stage, with no neurological deficits. After this second procedure, all patients made an uncomplicated recovery.

Discussion

Whilst the clinical diagnosis of penetrating head injury may be obvious, hair and blood may combine to make the detection of small holes in the scalp extremely difficult. In the absence of a CT, which, of course, is very helpful in equivocal situations, complete shaving and cleaning of the head facilitates the search for small wounds, which may require further exploration (Figure 1E–F).²⁷

Penetrating head injuries may be usefully classified into three categories, based principally on the configuration of the scalp wounds, with additional information provided by a CT scan: ‘tangential’; ‘penetrating with retained fragments’; and ‘perforation’. This is a simplified version of that used in the US Emergency War Surgery Manual.20 Without a CT scan, it is possible to make an ‘educated guess’ at the likely pattern of injury. Examination of the head may reveal only a single wound. In the absence of wounds to the face and neck, it is reasonable to assume that this is an entry wound, caused by a fragment from exploding munitions or a bullet that has failed to exit the head (the so-called ‘penetrating with retained fragments’ pattern of injury). After entering the head, a retained fragment may come to rest just under the skull (Figure 1F) or travel a variable distance through the brain, producing a spectrum of injury that, on CT scan, ranges from a well-defined haemorrhagic track with little mass effect and no associated swelling (Figure 4) through to widespread destruction of brain tissue with associated severe bleeding and brain swelling. It is worth noting that all penetrating wounds of the face and neck (including the orbit and mouth) have the potential to enter the intracranial cavity;²⁰ however, they are beyond the remit of this paper.

If there are multiple wounds to the head, the situation is more complicated. The possibilities are as follows.

• >Multiple entry wounds with retained fragments or, less likely, bullets.
• >A tangential wound or wounds where the trajectory of the missile (usually a bullet) has struck the head a ‘glancing blow’ and has travelled under/through the scalp for some distance, leaving an entry and exit wound, or simply disrupting the scalp.
• >A ‘perforating’ wound, or wounds. In the case of a single perforating wound, there will be an entry and exit wound (although it may not be possible to determine which is which) configured in such a manner as to suggest the missile (usually a bullet) has travelled through the brain (and is therefore distinct from a tangential wound).

Perforating head injuries therefore have an associated brain injury. A CT scan usually demonstrates a linear haemorrhagic track connecting the entry and exit wounds. Larger, more destructive wounds (where portions of scalp and skull may be missing or disrupted, with brain extruding) are associated with large fragments that exit the head, producing devastating, unsalvageable, damage. In these circumstances, identifying exit/entry wounds may be impossible.²⁸ Finally, all penetrating wounds have the potential to produce an extradural haematoma at the entry or exit site (Figure 8). This may be associated with minimal initial direct damage to the brain.

Figure 8.

Penetrating wound with retained fragments. CT shows an extradural haematoma at the site of the left frontal entry wound. Initially GCS 15 immediately after the injury, the patient deteriorated to localising and confused (with no obvious neurological deficit) at the time of the scan.

Head injuries may be classified as ‘minor’, ‘mild’ ‘moderate’ and ‘severe’, based on GCS assessment as follows.²³

• >Minor injury (GCS 15). Where direct damage to the brain is minimal, there may be no obvious focal neurological deficit; in this situation, there is usually little or no associated bleeding and brain swelling and the patient is fully conscious (Figure 1). The presence of an evolving extradural haematoma at the entry wound will lead to an increase in intracranial pressure and a consequent reduction in the level of consciousness. A patient with this type of injury may have little associated initial brain damage, and be initially fully conscious, with no focal neurological deficit.
• >Severe injury (GCS 8 or less). In the most severe cases, focal bleeding and brain swelling, together with mechanisms that produce brain damage distant from the path of the missile(s), rapidly result in a critical rise in ICP and a corresponding profound reduction in consciousness (which, in practice, means they are flexing to painful stimulus).^29,30 Gross neurological deficits, such as pupillary signs and an asymmetric limb response to painful stimulus, may be present.
• >Mild (GCS 12–14) and moderate injury (GCS 9–12). Between the extremes of minor and severe head injury, patients with a localising response to pain (defined as hand movement above a line drawn level with the tip of the chin towards a supra-orbital stimulus) may be found in both these categories.

In our series, fully conscious patients were effectively managed by what we have called a ‘limited wound excision’, as described above. In a patient with a penetrating head injury (and therefore a contaminated wound, by definition), antibiotic therapy and effective surgery are required to prevent infection.^31,32 There are therefore theoretical objections to limiting the wound excision, particularly in the case of a small ‘slot’ fracture, in which merely irrigating the skull surface and fracture site may miss in-driven fragments and foreign material. In fully conscious patients, however, the potential benefit of deeper exploration may be outweighed by the risks of aggravating brain bleeding and swelling, particularly in the hands of an inexperienced operator.^22,33 This risk is greatly increased in wounds lying over venous sinuses, and a here very cautious approach, comprising merely gentle irrigation and attempted primary scalp closure, seems justified, with or without a CT scan.

Our experience in Afghanistan suggests that patients with a single, obvious entry wound outside of the ‘danger areas’ and presenting with a localising response to pain (in practice, a GCS of 11–13) have the potential to do well with a simple decompression operation. The use of CT, combined with clinical assessment, makes it possible to determine whether decompressive surgery is appropriate and, of course, guides the exact site of the surgery. It is appropriate, therefore, to be cautious before embarking on surgery without CT, as the probability of a haemorrhagic track suitable for simple decompression in a patient with a single entry wound who is localising to pain cannot be determined. These considerations notwithstanding, all patients in whom conscious level assessment demonstrates a localising response to pain will require, at the very least, a limited wound excision, and the raising of a flap, as described, is a relatively straightforward further step. Should clot be seen to be spontaneously emerging from the underlying revealed dural entry wound, the subsequent surgical procedure involves no brain resection and no time consuming dural closure.^34,35 Storage of the bone flap, if required, is not a complicated procedure.

Following a relatively short operation, these patients can be immediately woken up. They make a rapid postoperative recovery; within 24hrs of surgery, they are fully conscious with, if present preoperatively, an improved neurological deficit. The lack of dural closure, although contentious, was not associated with any adverse short-term outcomes.^36–39 Consumption of local hospital time and resources is therefore kept to a minimum. Similar considerations apply to a subsequent delayed cranioplasty procedure; retrieval of and (re)placing the bone flap is relatively straightforward, and patients can be discharged from the ward within 3 days.^40,41

Non-operative management

For patients presenting with a best motor response of flexing to painful stimulus, CT scanning is an integral part of the decision to operate, the operative strategy and postoperative management. Notwithstanding the implications for the continuing postoperative care of a potentially severely disabled patient in a resource-poor country, the lack of a CT scan mandates conservative management in these patients. This means that patients in this category with a potentially salvageable extradural haematoma will not be treated. Irrespective of conscious level, similar considerations apply to patients presenting with a profound neurological deficit, such as hemiplegia.

Conclusions

If a CT scanner is unavailable, the following management strategy for children under the age of 16yrs with penetrating head wounds is suggested. Surgical intervention should be limited to gentle irrigation of the wound(s) and an attempt at primary closure of the scalp, irrespective of level of consciousness, in wounds potentially involving major venous sinuses or the frontal air sinuses (the so-called ‘danger areas’ described above), or in patients with a dense neurological deficit. In fully conscious patients and in those obeying commands, a limited wound excision, with removal of bone pieces and foreign material from the dura/surface of the brain, is appropriate. In patients who are localising to pain with a single (presumed entry) wound, it is reasonable to consider a simple decompression of a presumed track haematoma, without undertaking either brain resection or dural closure. If there is difficulty replacing the bone flap it may be temporarily stored in the anterior abdominal wall in a subcutaneous ‘pocket’, in anticipation of a further procedure 5–8 weeks later to retrieve and replace the bone flap as a cranioplasty. Patients with a best motor response of flexion to a painful stimulus should be managed conservatively. An algorithm (Figure 9) is suggested as the basis of safe practice and, as such, provides a guideline for the generalist surgeon.

Figure 9.

Algorithm: suggested strategy for management of paediatric penetrating head injury when CT scan is not available.

In conclusion, clinically-based criteria may be a safe and appropriate strategy In resource-poor environments for selecting children who may benefit from relatively straightforward surgery after penetrating brain injury.