How we treat recent thoracic and lumbar spine fractures

Hôpital de la Conception, Marseille


About 10,000 thoracic and lumbar spine fractures are treated every year in France. In our experience, a few very minor cases can be treated with bed rest and physiotherapy; 60 % of lesions can be managed with closed treatment; and only 30 % will require surgery. Our experience includes more than 500 cases treated in our Department over a 25-year period.

It should be emphasised that no treatment program can be implemented safely and effectively unless a sufficiently skilled surgical team, sufficiently competent paramedical personnel and adequate equipment are available.


I. A. Clinical assessment

This must be as detailed as possible as it will determine how urgently treatment is required and which medical facility is most appropriate for the patient’s particular needs.

A few specific points about the assessment should be borne in mind :

1 Fractures of the thoracic and lumbar spine are often the result of high-energy trauma (a fall from a height or a road accident); the mechanism which caused the lesion should be investigated as this will provide useful clues to the injury pattern that may be encountered.

2 If a patient has been involved in an accident and is complaining of back pain, the examination should include palpation of the contour of the spinous processes with the patient in the dorsal or lateral decubitus position. Any tenderness elicited will then pinpoint the location of the lesion and can be used to target further investigations more precisely. Excessive protrusion of a spinous process with an enlarged space between the spines, a subcutaneous haematoma, or even transverse stretch marks, are suggestive of a lesion caused by distraction of the posterior ligamentous structures.

3 A full and systematic neurological examination should be carried out, not forgetting the perineal region, which is often overlooked in emergency situations. The result of this examination should be clearly recorded in writing, as it will be referred to during the subsequent management of the patient. If the patient has to be transferred to a specialist unit, the clinical examination should be repeated in case there are any changes in the neurological signs and symptoms.

4 It should be borne in mind that a common finding in fractures of the thoracolumbar junction and the lumbar spine is reflex ileus related to a retroperitoneal haematoma around the fracture; this may be mimicking or masking an intra-abdominal or a retroperitoneal lesion.

5 Finally, we would emphasise that the rest of the spine (C-spine, sacrum and coccyx), the pelvis and ribcage should be examined for lesions, which may require their own forms of treatment, and may rule out certain treatment methods.

I. B. Further investigations

The examination of the vertebrae should include :

1 Good quality standard films showing the whole of the damaged vertebra (lateral views from the anterior prevertebral space right out to the spinous process, and anteroposterior views from one transverse process to the other) including as many adjacent vertebrae above and below as possible, with right and left oblique views if appropriate. There should also be a lateral view of the lumbosacral region. As we will see later, it is important to know the degree of lumbosacral lordosis in the patient’s spine, so that kyphotic sequelae in the thoracolumbar spine can be assessed.

2 A CT scan centred on the damaged vertebra and including adjacent vertebrae above and below the trauma site, with sagittal reconstructions.

3 Coronal and sagittal tomograms, which provide high quality information if CT is not available.

4 We do not perform magnetic resonance imaging (MRI), except in the rare cases of paraplegia without any radiographically demonstrable bony or ligamentous lesions, to screen for haematomyelia.

5 Myelography/radiculography are not usually done when the examinations described above are available.

I. C. Injury patterns

Vertebral lesions can be classified on the basis of their clinical and radiological characteristics. For this purpose we use the Magerl(12) scheme (Fig. 1) modelled on the AO classification of limb fractures, which provides an alphanumerical 3-3-3 grid: There are three types, A, B and C (Fig. A1), each of which is further classified into 3 groups (1, 2 and 3), which themselves each contain 3 sub-groups (1, 2 and 3), and further specification. Each type represents a principal injury mechanism, while the groups and sub-groups are based upon morphological characteristics, with ranking (1, 2 or 3) of the lesions according to progressive severity.

Without going into too much detail, the system can by explained in simple terms by saying that type A represents a compression mechanism (lesions located mainly in the vertebral body), with group A1 representing wedge impaction fractures; group A2, split fractures, and group A3, comminuted or burst fractures (Fig. 1B). The mechanism in type B is distraction of the posterior or anterior structures (distinguishing in each case whether there is a lesion of the vertebral body and/or of the disc), with group B1 representing predominantly ligamentous posterior flexion-distraction injuries; group B2, predominantly osseous posterior flexion-distraction injuries; and group B3, hyperextension-shear injuries with disruption through the disc (Fig. 1C). Type C represents lesions caused by a rotational mechanism in addition to the mechanism of one of the types described above (Fig. 1D).

Within this grid, the injuries are hierarchically ranked according to the progressive severity of the pathomorphological findings and the instability caused. This in turn makes it possible to suggest a prognosis for recovery, and to choose the most suitable method of treatment.

We have used this very specific classification to create a database which enables us to decide on treatment in relation to type of fracture, and to compare the efficacy of the various techniques available. If it were more generally used, it should produce better comparisons of the therapeutic results obtained by different teams specialising in spinal disorders. Of course, in an emergency situation the most pressing need is not to identify the precise type of fracture, which is the domain of scientific research, but rather to recognise the principal causative mechanism of a given lesion and to decide whether the function of the spine in terms of stability and neural protection has been compromised.


fig 1a

1A : Main characteristics of the three injury types : A Type A, injury caused by compression of the anterior column; B Type B, injury of the anterior column and the two posterior columns with distraction of the anterior or posterior elements; C Type C, rotational injury of all three columns.

fig 1c

1B : Type A, vertebral body compression injury : A1 Group A1, vertebral body wedge impaction fracture; A2 Group A2, split fracture; A3 Group A3, comminuted fracture or burst fracture.

fig 1b

1C : Type B, anterior and posterior element injuries with distraction : B1 Group B1, predominantly ligamentous posterior flexion-distraction injury; B2 Group B2, predominantly osseous posterior flexion-distraction injury; B3 Group B3, injury involving hyperextension and shearing through the disc.

fig 1d

1D : Type C, anterior and posterior element injuries with rotation : C1 Group C1, type A injury with rotation; C2 Group C2, type B injury with rotation; C3 Group C3, oblique fracture with rotational shear.

I. D. Spine functions and therapeutic indications

In an attempt to refine the therapeutic indications, we routinely determine the effects of the fracture on the 4 main functions of the spine : stability, posture, neural protection, and neurological function (14, 16).

- 1. Stability and instability.

It has to be decided whether the trauma sustained has destabilised the spine to the point where the spine can no longer maintain intervertebral displacement within normal limits. The basis for this decision is R. Louis’ three-column theory (Fig. 2A).

The function of stability will remain uncompromised only if the trauma has left all three columns intact, i.e. the major anterior column formed by the discs and vertebral bodies, and the two posterior columns formed by the partes interarticulares and the facet joints linked by bridges formed by the pedicles and the laminae. Each column consists of alternating sections of bony and of fibrous (discoligamentous) structures.

The more columns that have been damaged, the more unstable the spine will be. Each column may be damaged by loss of continuity in a section of bone or of fibrous tissue, or by a defect (loss of substance).

A stable traumatic lesion is one which damages only the bony portion of one column or a horizontal bridge. In contrast, lesions of two or three columns will be unstable.

Damage to the bony part of two or three columns causes transient osseous instability. The fracture may be adequately and rapidly stabilised by consolidation if treated with reduction and immobilisation (Fig. 2B).

In contrast, lesions of the articular constituents of the columns, i.e. of the discs and ligaments (B2, B3 and some C types) have only a small chance (30 % to 50 %) of achieving solid and stable consolidation with conservative treatment. These lesions are said to exhibit chronic ligamentous instability, and are therefore clear indications for surgery (Fig. 2C). Similarly, surgery will be required if there is a defect involving loss of substance in one or more of the columns. Such a defect may take the form of a severe wedge impaction fracture (A1 injuries) with anterior compression by 50 % or more of the height of the vertebral body; a burst fracture (A3.3) leaving a defect in the vertebral body after reduction (Fig. 2D); a coronal fracture of a vertebral body (A2), leaving a defect underneath the suprajacent vertebral body; or a pincer fracture (Fig. 2E). Defects will require bone grafting.

fig 2

2A : Diagram of the R. Louis three-column theory. 1. Large anterior column consisting of the discs and vertebral bodies; 2 and 3, posterior columns consisting of the partes interarticulares and the facet joints.
2B : Osseous instability (Chance fracture).
2C : Ligamentous instability (dislocation).
2D : Anterior defect in a reduced wedge impaction fracture.
2E : Anterior defect caused by coronal fracture (split fracture and pincer fracture).

- 2. Posture and deformity

The postural function of the spine depends on a physiological pattern of spinal curvatures. Any serious deformity compromises posture, exposing the patient to the risk of pain and imbalance at the fracture site and in the compensatory curves. This may involve kyphosis or scoliosis, which becomes abnormal at between 15 and 20° at the fracture site (local kyphosis), depending on the original curvature of the patient’s spine. This type of vertebral deformity should therefore be reduced by conservative treatment; or surgically, if conservative treatment is not effective or if it is likely that local kyphosis will be more than 15° or 20°, irrespective of regional kyphosis. This is especially true if the subject has hyperlordosis or narrow spaces between the spinous processes or between the facets and the laminae (see Indications for Conservative Treatment). Of course, the extent of normal kyphosis or lordosis at the site of the lesion should be taken into account in any assessment of postural disorders.

- 3. Protection of the nervous system, stenosis of the spinal canal

The morphology of the vertebral canal and the foramina must be preserved in order to protect the spinal cord and the nerve roots. If there is a severe deformity resulting in spinal stenosis and causing varying degrees of compression, the walls of the canal would have to be reconstructed as far as possible in order to relieve the compression or to prevent subsequent compression. Closed reduction will be effective in many cases, but if this fails and residual compression persists, the canal will have to be decompressed by surgery. However, it should be remembered that only 50 % of the volume of the canal is occupied by neural structures. Spinal stenosis should therefore not be treated surgically in cases where the reduction in canal diameter is less than 30 % (or less than 50 % of the cross-sectional area of the canal, which can be calculated from a CT scan). However, we do operate as a matter of course in all cases where there is a neurological deficit, irrespective of the extent of canal obstruction.

- 4 Nerve lesions

Some neurological lesions are clearly irreversible, such as when nerve structures have been overstretched, crushed, or severed. However, many other lesions caused by partial compression may recover when the compression is removed. Although closed reduction may be effective in treating compression, neurological deficits are a contra-indication to continued conservative treatment, and must be treated surgically. The chances of recovery are best if a compressed neural structure can be be released quickly, within 8 hours if possible. When lesions involve mainly the spinal cord, drug therapy may be necessary during the first 24 hours to reduce necrosis and oedema at the level of the injury. We have abandoned high-dose corticosteroid regimens, and now administer only more standard doses (120 mg of Solumedrol® [methylprednisolone sodium succinate] for the first two days, tapering off the dosage thereafter).

Depending on the extent to which the four functions are affected, there may be a clear indication for closed treatment, or there may be one or more arguments in favour of surgery. When surgery is required because of compromised function, an approach must be chosen which will allow each of the functions to be recovered as effectively and as easily as possible.


II. A. General remarks

We find that most cases of recent vertebral deformity can be reduced without any major problems or complications using our method of treatment, which is derived from the method proposed by Böhler, using closed reduction followed by a plaster cast.

The method consists in reduction with the patient placed on a Cotrel scoliosis frame(8,13,15,19,22), using well-defined manoeuvres under image intensifier control.

1 2 3
1 : 41-year-old man, road accident. Burst fracture of L1 without neurological deficit. 25 % canal narrowing.
2 : Result after closed reduction.
3 : Excellent functional result at 2 years.

II. B. Technique

While Böhler used a method of reduction involving lumbar hyperlordosis, we prefer a three-stage reduction on a Cotrel frame (Fig. 3). Neither local nor general anaesthesia is required.

1 The first stage of reduction consists in applying axial traction with a head halter and two pelvic straps. A dynamometer is used to check the force applied, which is in the order of 10 to 15 kilograms. The fracture site is therefore maintained in its position of deformity, i.e. in kyphosis at the level of the affected vertebra. Traction is monitored on the image intensifier, to check that the volume of the vertebral body has been restored, measuring the distance between the adjacent discs above and below the fracture (Fig. 3A-3B).

2 The second stage consists in re-establishing a certain amount of lordosis by passing a sling under the fracture site to pull the spine vertically upwards. The sling is tensioned until physiological lordosis is obtained. This procedure is beneficial in cases where a fragment of the posterior wall has been displaced into the vertebral canal : The vertebral body can be opened at the site originally occupied by the fragment, and the fragment itself can be pushed home by pressure from the posterior longitudinal ligament. In contrast, if lordosis is induced before the axial tension has been applied in kyphosis, the site of origin would be closed and it would be impossible to reduce the posterior fragment; there would then be a risk of the fragment being driven back into the canal and causing or aggravating neurological compromise. If the patient is conscious he or she will be able, at this stage, to report pain and neurological phenomena. Reduction manoeuvres should be stopped as soon as neurological manifestations occur or if they become aggravated (Fig. 3C).

3 The third stage consists in applying a classical Böhler body cast with three points of contact on the sternum, back and pubis, leaving a window over the upper abdomen (Fig. 3D).

The patient is allowed out of bed the same day, and may be discharged on the second or third day if there is no posttraumatic ileus. When there is neurological loss, we use this method before surgery to reduce the fracture, which simplifies our operative technique (see Surgical Treatment). However, if the neurological deficit regresses completely (single-root compression), conservative treatment is given on a trial basis.

Outpatients are asked to return three weeks after the first plaster cast is fitted to check that there has been no loss of reduction and aggravation of the initial deformity, which would mandate surgical stabilisation. It should be remembered that some loss of reduction while the patient is in a plaster cast is normal; however, with appropriate patient selection, this should not be so severe as to require surgery.

3 abc

3 d

Controlled closed reduction done under anaesthesia and under fluoroscopic control. A Cotrel frame is particularly suitable for this type of reduction, which is performed in three stages : The patient is positioned in such a way that the initial kyphotic deformity at the fracture site is maintained (A), axial traction is applied (B), followed by induction of lordosis while traction is maintained (C). A plaster body jacket is applied after the closed reduction (D).

Key to diagram :

II. C. Subsequent treatment

After a patient has spent three to four months in a plaster cast, there is a danger of muscle atrophy. Isometric proprioceptive exercises should therefore be started as soon as the pain has subsided, in order to prevent wasting of the abdominal and spinal muscles.

A fracture will usually take between three and four months to consolidate. As a precaution, a reinforced canvas brace with shoulder bands is often used to immobilise the spine for a further period of one to three months. Active and dynamic exercises are begun during the fifth month, with emphasis on the patient learning the "lumbar locking" technique. Stress films are then taken in order to check that there is no instability. The patient can go back to work at six months. This time may be substantially shorter for subjects in sedentary occupations or in professional activities that do not require major physical exertion; however, labourers should not lift heavy loads (more than 25 kg) for a further 6 months.

II. D. Complications

Localised pain underneath the plaster brace suggests that there is a pressure sore, which should be treated with standard local care followed by a change of cast.

We have never observed any neurological complications during the reduction procedure; however, if such a situation were to arise, immediate surgical release would be required.

II. E. Indications

We have been able to analyse and evaluate the results of closed reduction performed in 240 cases treated over a 10-year period; this analysis has enabled us to define the indications (single or multiple factors).

The following types of fracture should be managed conservatively :

- fractures of the thoracolumbar junction down as far as L3, and of the lower thoracic spine. For fractures at other sites, reduction is much less effective and a simple support without reduction is indicated;

- fractures in which local kyphosis is between 10° and 15° and the coronal deformity is less than 10°. When local kyphosis is between 15° and 20°, only patients with an adequate hyperlordotic reserve should be treated this way. When local kyphosis is less than 10° a simple support without reduction will be sufficient;

- stable lesions and unstable injuries involving predominantly bone (types A and B2), apart from cases where there is a defect or a rotational lesion;

- fractures without any signs of cord or multiple-root compromise. Patients with signs suggesting compromise of a single root may be treated conservatively, providing that the signs are completely abolished by reduction.

- fractures where canal encroachment is less than 50 % of the cross-sectional area of the canal, without any neurological compromise; as well as

- fractures which are not associated with major chest injuries, or, more generally, multiple trauma. Subjects should be able to get up and take part in early and sustained rehabilitation while in plaster, and should not be obese.

In contrast, conservative treatment is not effective in the following cases, which are likely to require surgical treatment :

- any fracture causing more than 20° of local kyphosis and more than 10° of coronal deformity;

- any cases where the instability is mainly ligamentous (types B1 and B3), cases of bony instability with a defect (some types A1 and A2, and A3.3) and rotational lesions (type C);

- all fractures which have been complicated in the initial or later stages by neurological signs.

For technical reasons (impossibility of performing the reduction manoeuvres required, or nursing problems), spinal fractures associated with fractures of more than one rib, unstable fractures of the sternum, unstable fractures of the pelvic ring, and fractures occurring as part of a multiple-trauma pattern should be managed either by bed rest and physiotherapy or surgically, depending on the extent to which the major functions of the spine are affected.

However, we use preoperative conservative reduction whenever possible, so that we can operate on a spine which has already been reduced.

III. POSTERIOR SURGERY (Images - Cases No. 2-3)

III. A. General remarks

We will describe the method used in our department, which is based on R. Roy-Camille's principle of fixation by means of plates and pedicle screws (1,2,5,6,7,15,16,17,20,21). We use Louis plates.

2-1a 2-1b 2 3
1 : 21-year-old man, motorcycle accident. Multi-level fracture with oblique fracture line, T8, T9 and T10, no neurological deficits.
2-3 : Posterior internal fixation with fusion of T6-T12 without decompression. Excellent functional result at 3 years.

III. B. Indications

This is the approach used in neurological emergencies.

The posterior approach is the preferred route when there is trauma to the thoracic and lumbar spine with neurological lesions, and in other cases of trauma, when there is no imperative requirement to repair a defect of the vertebral bodies or to relieve severe anterior compression. However, in that last-mentioned case, we use a posterior approach first to remove any retropulsed fragments from the canal, in order to avoid an anterior approach.

Similarly, the posterior route would be used in preference to the anterior route in an obese subject or in one likely to be at high risk during surgery.




1-2 : 32-year-old man, motorcycle accident. Burst fracture with rotation of L2. Bilateral weakness of thigh muscles.
3 : Posterior decompression and fusion of L1-L3 with a short construct. No neurological sequelae, no pain at 2 years.

III. C. Advantages and disadvantages

Internal fixation by plating will produce a semi-rigid form of construct. A reinforced fabric brace will need to be worn for three months after surgery.

The advantages of our method are :

- The hardware generally does not have be removed, as our plates have a low profile and produce little bulk. In addition, we often use short constructs combined with fusion of the facet joints.

- As the approach does not require the transverse processes to be exposed, the neurovascular bundles of the paraspinal muscles passing just outside the pars interarticularis are preserved (Fig. 4).

4A : Exposure technique which preserves the neurovascular structures of the spine. Our plate-and-screw system and fusion mass add little bulk, so no pressure is applied to the structures during closure of the incision.
4B : Diagram illustrating the hazards to the neurovascular structures of the paraspinal muscles from vertebral exposure beyond the facet joints and the insertion of prominent hardware and bulky iliac grafts between the transverse processes.

- The need for iliac bone grafting can usually be obviated if corticocancellous bone chips taken from the spinous processes are used for facet joint fusion.

- Our procedure using lordosing spinal traction provides easy reduction during surgery, which generally means that over-aggressive manipulation during surgery can be avoided.

The disadvantages are :

- Our plate and pedicle screw system does not reduce fracturs in same way as a system using rods. For this reason the fracture is often reduced preoperatively by closed reduction, which is maintained by lordosing traction applied throughout the surgical procedure (see Position of the Patient).

- The use of a semi-rigid system involves a very small amount of postoperative loss of reduction.

III. D. Operative technique.

- 1. Position of the patient

The posterior approach means that the patient has to be turned on the operating table after general anaesthesia has been induced, while wearing a bivalved body jacket (after preliminary closed reduction has been performed). Lordosing spinal traction is then applied using a head halter, a pulley and a dynamometer fixed to a hook hanging from the operating table at the head end, and two boots fixed to the operating table. The operating table is gradually bent into a lordosing angle, while 10 to 15 kg of vertebral traction is maintained, to correct or maintain correction of the fracture site under image intensifier control. Surgery will then be performed on reduced lesions (Fig. 5).


Patient positioning with lordosing spinal traction for a posterior incision.

- 2 Incision and placement of the screw tracts

The incision is in the posterior midline. The fascia is detached from the spinous processes using cutting diathermy, and the muscles are stripped first from the spinous processes and then from the laminae using a very wide elevator which is too big to pass between two transverse processes. If possible, the muscles should be detached at both ends of the incision, away from the actual fracture site. The site can then be approached, holding the elevator or scissors very carefully so that they do not penetrate the dural sac through the fracture site. Laterally, exposure is confined to the outer margin of the facet joints, preserving the neurovascular structures lying between the transverse processes (Figs. 4 and 6).


Posterior incision, nerve decompression, and repair of the dural sac.

Retractors are inserted and haemostasis is obtained with bipolar diathermy. The vertebral canal can then be either decompressed or stabilised by fusion, depending on the specific requirements of the situation.

We recommend that if canal decompression is envisaged, the screw tracts should be prepared immediately, without inserting the screws. Decompression of the canal can cause bleeding which makes it difficult to identify the pedicles, and rapid closure may be necessary if bleeding is heavy and uncontrollable. Once the screw tracts are in place, the fracture can be stabilised rapidly.

The entry points for the screws will depend on the section of spine involved (Fig. 7) :

Anatomical landmarks for thoracolumbar screw tracts, and image intensifier control of the placement of the marker K-wires.

- Between T1 and T3, the insertion point is 3 mm below the facet joint and 3 mm medial to its lateral margin. In these vertebrae only, the screw tract is routed slightly obliquely and medially.

- Between T4 and T10, the insertion point is 5 mm medial to the lateral margin of the facet joint and 3 mm from its lower margin on a crest which medially continues the upper border of the transverse process.

- Between T11 and L5, the insertion point is located at the intersection of a vertical line that runs tangent to the inside of the notch formed by the outside limit of the pars, and a horizontal line drawn 4 mm above the upper margin of that notch, i.e. inside the lower part of the facet joint space.

Whichever insertion point is chosen, small K-wires may be introduced at the intended point; a.p. and lateral views on the image intensifier are used to verify that the position is correct and that the wire is in the centre of the pedicle. (If this marking is done before nerve roots are freed, the holes should be temporarily filled with bone wax).

- 3. Nerve decompression

A preoperative CT scan is used to decide on where decompression is required, and to show whether bits of bone have been displaced at the fracture site. It will also be used to guide the lifting and extraction of any fragments displaced towards the vertebral canal.

Dislocation of the facet joints can be reduced by resecting a third of the end of the upper facet. A tyre-lever manoeuvre can be used to replace the dislocated facet in a posterior position while applying posterior distraction by means of two bone-holding forceps placed on the spinous processes above and below the site of injury.

If there are neurological symptoms and the radiographs suggest that there is stenosis, laminectomy should be done, preserving the facet joints and the partes interarticulares.

If there is a complete or a greenstick vertical translaminar fracture (which is common in burst fractures), rootlets of spinal nerve roots are likely to have been entrapped, and the laminectomy should be started some way away from the fracture of the lamina.

The margins of the laminectomy may be covered with bone wax to obtain haemostasis.

The lateral recesses are explored with a dissector to find a focus of compression, and the posterior surface of the vertebral bodies and discs is likewise probed.

The main source of compression is one or more posterosuperior (or, more rarely, posteroinferior) fragments of a vertebral body that have been driven back into the canal. The fragment or fragments should be removed or pushed well forward. To do this, we use a special L-shaped bone impactor. To ensure that the procedure is effective, the offending fragment has to be reimpacted deeply into the posterior third of the vertebral body. This stage is likely to cause bleeding, and small neuro sponges should be packed liberally into the bleeding sites in each of the recesses, taking care to avoid compressing the dural sac. Strips of Surgicel may also be used, instead of these sponges, at the end of surgery.

The bone fragments removed from the spinous processes and the laminae are given to the scrub nurse, who will clean off any fibrous tissue or cartilage to make them into corticocancellous strips which can subsequently be used as bone grafts (Fig. 6).

- 4 Repair of the dural sac

Any breaches of the dural sac must be exposed and then repaired using non-absorbable 5/0 vascular sutures on small curved atraumatic needles. A single figure-of-eight stitch will be sufficient to repair an isolated puncture wound, but wider tears will have to be oversewn.

If the breach is very lateral or anterolateral and is therefore difficult to suture, or if there is major defect, it is best to put Surgicel around the tear or defect, or to take down a piece of muscle from the margin and reinforce this patch with tissue glue.

When a tear in the dural sac is being repaired, a patty should be put over the end of the suction tip to prevent damage to nerve roots.

When there is a complete neurological lesion, it may be necessary to open the dural sac to verify the lesions, which can then be reported to the patient and his or her family. Postoperative MRI may be interfered with by the hardware inserted at surgery. It is also recommended that the sac should be opened if the cord is very swollen because of a central haematoma. The haematoma should be aspirated through the posterior commissure of the spinal cord with a very fine needle.

When there is incomplete neurological loss and all epidural compression has been relieved, the dura mater should not be opened, as this would add iatrogenic micro-trauma.

If there has been significant loss of CSF (floppy dural sac), the contents of the dural sac can be reconstituted by introducing physiological saline using a fine needle and a syringe, until the sac is once more smooth and cylindrical. This procedure also reduces epidural bleeding.

After the dural sac has been repaired, either a double layer of Surgicel or a layer of tissue glue should be applied (Fig. 6).

- 5 Stabilisation and fusion

For internal fixation we use paired symmetrical R. Louis stainless steel plates with closely-spaced holes, and pedicle screws. All the plates can be bent and cut as required.

For the pedicles of the upper thoracic spine we usually use 4.5 mm diameter screws (drill size 2.8 mm); for the others, we use 5.5 mm screws (drill size 3.2 mm).

The screws are inserted in the sagittal plane perpendicular to the plane of the neighbouring lamina. The screws should be inserted in such a way as to ensure that the tip does not penetrate further than the middle of the vertebral body (on the lateral views). For an average adult, screw length will range from 30 mm at T1 to 45 mm at L5. We have two plates with oblique screw paths which go towards the upper surface of the lateral bone mass of the sacrum for L3 or L4 fusion to the sacrum (which is done very occasionally in traumatology). The holes in the plate are arranged in such a way as to ensure that the sacral screws are routed with angulation in two planes.

Before putting the plates into position, the facet joints of the vertebrae between those receiving the end screws are freshened by shortening the inferior facet by a third at the thoracic level and excising visible cartilage, and by excising the sagittal portion of the lumbar facet joints with a narrow rongeur. Similarly, at the lumbar level, the part distal to the joint space is freshened between the tip of the inferior articular process and the subjacent lamina (Fig. 8).

The plate is applied by inserting the top screw first. It is then aligned on the facet joints and the bottom screw is inserted. The intermediate screws are then inserted guided by the contralateral landmarks.

If used, chips of corticocancellous bone are placed on the freshened facet joints and the plate is tightened onto the graft mass (Fig. 8).


Internal fixation procedure with posterior fusion.

In cases of dislocation or lesion of a motion segment, the construct should extend to one pedicle above and one below the lesion.

For fractures, the construct should run from the pedicle above to the pedicle below the fractured vertebra; in the case of complete neurological lesions, a longer construct is preferred, from two vertebrae above to two vertebrae below the level of the lesion.

For fractures of the thoracolumbar junction, an “extended short” construct may be used, covering 2 pedicles on either side of the fractured vertebra, with fusion of the motion segments above and below the fracture. In the latter case, we expect to remove the hardware after six months to avoid the screws at the ends of the construct breaking.

It is sometimes difficult to reduce fractures of the upper thoracic spine preoperatively. In this case, we would ease the spine onto prebent plates, gradually inserting the screws into the pedicles above the fracture. We always include the 2 vertebrae above the fracture, and the plates are contoured according to the curvature to be obtained in that region of the spine.

- 6 Closure

In most cases we close the muscles, fascia, subcutaneous tissue and skin with interrupted sutures, without inserting drains. However, if there was excessive bleeding during surgery, a suction drain should be used to prevent compression of the dural sac.

- 7 Postoperative care

To prevent cardiovascular shock and to avoid the risks involved in any heterologous transfusion, a Cell-Saver unit should be used to salvage the patient's red cells for reinfusion.

When the dura mater is damaged, we routinely prescribe postoperative antibiotic therapy (Bactrim®).

During the first few days after surgery, the patient is at risk from ileus. Nasogastric decompression should be performed for the first 24 hours, and the passage of flatus should be checked.

Anticoagulants should not be given for the first 12 hours, to prevent epidural haematoma formation.

We routinely perform postoperative CT scans to check that nerve decompression has been successful, and to verify that the pedicle screws have been placed correctly.

Patients with severe paralysis should be nursed in bed, with turning every 3 hours to prevent bedsores(9). After 2 to 3 weeks they may sit up, protected by a reinforced fabric brace. These patients will require rehabilitation at a specialist centre.

Patients who are not paralysed may be got up from the day after the operation, protected by a reinforced fabric brace (for long constructs) or by a plaster or heat-moulded resin cast (for short constructs), which will be worn for three to four months. The patient will be followed-up every month for 6 months, then every year. Dynamic exercises are started at the end of the third postoperative month. Patients in sedentary occupations may go back to work after four months; labourers may resume work after six months. In order to preserve healthy motion segments, we ask our patients to practise the technique of "locking" the lumbar spine in neutral position for the rest of their lives.

- 8 - Complications

The first complication to prevent is major blood loss during surgery, because recent trauma to the spine tends to cause substantial bleeding. Hot moist swabs should therefore be used to cover those parts of the operative field which are remote from the site that is being addressed. On entering the spinal canal, hot moist neuro sponges and Surgicel should also be packed into the lateral recesses, taking care not to compress the dural sac.

A screw may be inserted into the pedicle of the fractured vertebra to increase the stability of the construct. In the case of burst fractures, the interior of the canal should be checked after the screw has been tightened, as it is possible that cortical bone on the inside of the split and fractured pedicle may break off while the screw is being inserted, causing compression of a nerve by the screw and/or by bone fragments avulsed from the pedicle. The interior of the canal should also be checked in the case of a fracture of a pars or of an articular process. When the screw is tightened, the plate presses against the facet joints, which may displace one or more of the fractured joints into the canal.

Occasionally, one or more screws may have been slightly misrouted in the pedicle (this will show up on a postoperative CT scan). Only screws which are causing symptoms of nerve root compromise should be repositioned. The screw tract is often excessively medially routed and too low down on the pedicle.

A suction drain should never be left in contact with a breach of the dural sac as it is likely to cause tonsillar herniation.

Postoperative infection means that asepsis was inadequate. The operative site should be irrigated frequently with an antibiotic solution.

Further surgery by the anterior route will be required if a postoperative CT scan shows persistent residual anterior compression of more than 50 % of the anteroposterior diameter of the vertebral canal, particularly if the patient was paralysed before surgery.

IV. ANTERIOR SURGERY (Images - Case No. 4)

4-1 4-2 4-3
1 : 21-year-old soldier, parachuting accident. Burst fracture of L2; no neurological deficits. Anterior defect and < 20 % canal encroachment.
2-3 : Anterior reconstruction with arthrodesis of L1-L3 via a left-sided thoracoabdominal incision without decompression. Return to work at 6 months. No pain 2 years later.

IV. A. General remarks

Most vertebral lesions occur at the front of the spine, and one would therefore logically expect to repair them directly using an anterior approach. However, the spine runs through anatomical regions which involve various specialties, so a surgeon working on the spine must be able to access the spine by the thoracic, thoracolumbar, lumbar or abdominal routes. This technical requirement is obviously an obstacle to the general use of the anterior approach to the thoracolumbar spine, unless the orthopaedic surgeon can call upon the services of a chest or general surgeon to establish the access. Without special training or help from an appropriate specialist, anterior surgery is difficult and dangerous. In addition, the internal fixation methods which are possible using this approach do not provide such good stability as those used in posterior surgery. For these reasons, the anterior approach should be performed only in stringently selected patients (3,4,10,11,16,20).

IV. B. Indications

There are two types of lesions which require an anterior approach : anterior column defects, and spinal canal encroachment by a vertebral body.

Loss of bone substance appears in the form of substantial defects which are visible on a CT scan of the vertebral body in severe wedge impaction fractures reduced by conservative treatment, or of compression of the vertebral body by more than 50 % of its normal height. Anterior exposure and fixation will also be required in the case of a coronal fracture of a vertical body creating a space below the anterior inferior margin of the suprajacent vertebral body.

Residual spinal canal encroachment by the vertebral body causing > 50 % canal lumen reduction will require decompression by the anterior approach, especially if there is incomplete neurological loss.

Thus, if there are no neurological complications and if there is an anterior column defect with no sign of posterior lesions, the repair may be done using an anterior approach alone.

In the other cases, posterior surgery will be done before anterior surgery.

Major obesity may be an obstacle to anterior surgery. Associated thoracic or lumbar disorders may be a contra-indication to this form of surgery.

IV. C. Operative technique

- 1 Position of the patient and incisions

We perform anterior surgery with the subject supine, so as to be able to work very close to the midline sagittal plane, and so have direct access to the vertebral canal. Similarly, internal fixation is easier when the spine remains straight, which is not always the case when the patient is lying on one side. We routinely put our patients in spinal traction, using the same equipment and the same forces that we use for posterior surgery. The table is broken in such a way as to place the hinge immediately beneath the fracture site, to induce lordosis (Fig. 9).


Patient positioning with lordosing traction for an anterior incision.

We use an image intensifier to ensure good reduction, so that the fracture can be reduced even before an incision is made.

We always harvest a piece of fibular shaft to act as an anterior strut graft. We remove it by making an incision below the middle of the leg with the lower end at least 10 cm from the lateral malleolus. A segment of fibula 3 to 4 cm long should be harvested for use in the upper thoracic spine, two fibular segments for the thoracolumbar spine, and three fibular segments for the lumbar spine.

To access the thoracic spine from T3 to T10, we use a submammary right anterolateral thoracotomy, resecting 3 to 4 costal cartilages, which will be reattached at the end of surgery.

For access to the thoracolumbar junction, from T10 to L2, we recommend a thoracoabdominal approach (thoracophrenolumbotomy) that is transpleural but retroperitoneal. All transthoracic approaches will require two chest drains.

For the middle lumbar vertebrae, from L2 to L4, it is best to use a left-sided retroperitoneal flank approach with an oblique incision parallel to the neurovascular structures of the abdomen.

To repair lesions situated in the lumbosacral region, from L3 to S1, we prefer a midline transperitoneal approach.

New possibilities for treatment have been opened up with the introduction of endoscopic or video-assisted surgery, making it possible to reconstruct a vertebral body or release a nerve root using internal fixation. These minimal-access procedures can be performed between T4 and T12 using a thoracoscopic approach; between T12 and L4 using a laparoscopic retroperitoneal approach; and between L4 and S1 using transperitoneal laparoscopic surgery. To reach the thoracolumbar junction, thoracoscopic surgery can be combined with a minimal-access technique using the retroperitoneal approach. A long learning period is needed for these endoscopic techniques, together with an excellent knowledge of the classical approaches in case conversion to an open procedure becomes necessary. The advantage of endoscopic techniques is that they reduce postoperative complications and leave smaller scars than do the classical anterior approaches (18). We can only mention these methods in passing, as we are only just beginning to use them.

- 2 Anterior decompression

If an anterior incision is used to repair a defect without narrowing of the spinal canal, resection of the bony part of the vertebral bodies and of the discs can be limited to the free or mobile anterior portion of these structures, allowing the posterior wall to be preserved.

In contrast, if there is any canal encroachment with neurological symptoms, a true rachiotomy must be done, exposing the anterior surface of the dural sac for the entire length of the lesions shown on the radiographs (including CT scans).

Each of the vascular structures is taken on a suture passer, ligated or clipped, and divided midway between the aorta and the neural foramina.

The surfaces of the vertebral bodies are then gently stripped using a medium-wide curved elevator, from the side of approach towards the anterior surface of the vertebral bodies, passing in front of the anterior longitudinal ligament towards the start of the opposite surface.

When exposure is complete, a malleable blade retractor bent into an S-shape may be inserted against the opposite surface of the vertebral body, protecting the large prevertebral vessels in the upper excavation of the S (Fig. 10).


Anterior approach via a left-sided thoracoabdominal approach. Ligature and division of the segmental vessels, and exposure of the spine.

The trauma site may then be resected, starting from one centimetre in front of the neural foramina and continuing to the anterior surface of the vertebral bodies, leaving a small contralateral wall. Straight and angled rongeurs and curettes are used to resect progressively the vertebral body and one disc or the two adjacent discs. The resection often causes bleeding, so bone wax should be applied to the raw cancellous surface at frequent intervals. The resection gradually gets closer and closer to the posterior wall of the vertebral body. When extracting a large fragment impacted in the vertebral canal, no manoeuvres should be performed from front to back; instead, the disc immediately above the displaced fragment should be detached so that the dura mater is visible in the intervertebral space.

An angled curette can then be inserted underneath the trapped fragment and the fragment can be drawn out. A strip of Surgicel covered with hot moist swabs should be left on for a few minutes to obtain haemostasis. The anterior surface of the dural sac will then be completely free, with only a wall of bone by the foramina (Fig. 11).


Anterior decompression; the vertebral body is opened and fibular grafts are inserted.

- 3 Stabilisation and fusion

We prefer not to use inert replacement materials or internal fixation devices that may cause major damage if the implant fails. It should always be borne in mind that a major vessel may be damaged by an unstable construct, and that revision may be difficult if there are subsequent complications. Our method is based on the three factors which are necessary and sufficient for a good repair of the anterior column : The first factor is sagittal bone grafting using shaft of fibula; the second is the preservation of healthy endplates with sound cortical bone to act as supports for the flat ends of the fibular pegs; and the third factor is an anterolateral plate with screws inserted into the healthy vertebral bodies above and below the decompression site (Fig. 12).


Insertion of an anterolateral plate through a right-sided thoracotomy.

Although the fibula has traditionally been regarded as poor graft material, we can confirm that we have never observed non-union in any of the cases when fibular grafts have been inserted under the conditions described.

The fibular graft must be supported by cortical bone, since a cancellous bed would not be strong enough. An anterolateral plate will not interfere with the large vessels, since the screw heads and plate will be tucked away in a neutral zone along the thoracic and the lumbar spine. When the fibular graft or grafts have been cut to the exact dimensions of the defect between two healthy endplates, they are wedged in with some force, and spinal traction is released, which compresses the grafts.

The plates we use are the same as those for posterior fixation. The plate is chosen according to the length of the vertebral bodies to be fixed; any holes not required are cut off. The plate is positioned with a temporary pin at each end so that it is snugged down onto the anterolateral aspect of the vertebral bodies. The screws (diameter 5.5 mm) are also the same as those used in posterior surgery. A pilot hole 1 cm deep is drilled into the vertebral body, using a 3.2 mm drill bit. The screw is inserted in the direction of the opposite surface of the vertebral body, angling the tract slightly backwards, as if aiming at the opposite pedicle. It is not essential to try to engage the posterior cortex, as the construct will be sufficiently solid. In adults, the screw length used is about 25 mm for thoracic vertebrae, and between 35 mm and 45 mm for lumbar vertebrae. Two screws may be inserted into each thoracic vertebral body, and three into each lumbar vertebral body.

If L3-L5 are being fused, anterior internal fixation is not necessary if posterior fixation has been performed. An anterolateral plate is likely to interfere with the left iliac vessels.

Suction drainage should be by means of two tubes in the thoracic region and one drain in the lumbar region; it is not required if the midline transperitoneal route is used.

- 4 Postoperative care

After a thoracic approach, the patient remains supine, with an air and a fluid drain, which are generally kept in for 4 days. A Y-shaped connector with a one-way valve is inserted into the ends of the two tubes. When closing the chest wall, the surgeon and the anaesthetist should take care to evacuate the pneumothorax, clamping the drain after each expiration, to bring the lung back to the chest wall. Re-expansion of the lungs should be checked every day by auscultation and by bedside chest radiographs.

After surgery confined to the lumbar region, a suction drain should be left in for 48 hours.

When the transperitoneal or the left thoracoabdominal routes have been used, the patient should receive antibiotic cover, with nil by mouth and parenteral nutrition until flatus is passed.

Anticoagulation is not essential, but should be used after 12 hours in at-risk patients or if prolonged recumbency is expected. If anticoagulation is given too early it may cause dural sac compression by a haematoma.

It is generally sufficient to put the patient into a reinforced fabric brace as soon as the drains have been removed. The support may be loosened after meals and at night; it must be worn for about 4 months. If anterior surgery alone was used, it would be preferable to use a slightly more rigid external support.

No specific rehabilitation is required, apart from a good deal of brisk walking.

Postoperative follow-up is the same as after posterior surgery.

- 5 Complications

There are many possible types of complication, which may be vascular, neural, respiratory or mechanical.

Postoperative haematomas may be caused by failure to obtain adequate haemostasis.

Neurological complications may be caused inadvertently during decompression : There tends to be a great deal of bleeding at this stage, making it impossible for the surgeon to see what is happening before it is too late. It would be best to use curettes working outwards from the danger area (the deeper layers) towards the surface, and protecting the large vessels by malleable blade retractors.

Thoracic complications may include pneumothorax, haemothorax or atelectasis. Haemothorax may be caused by vascular leaks; if it is severe, revision may be required. Pneumothorax may be caused by inadequate re-expansion of the lung during closure, or by postoperative air leaks. The chest tubes should therefore be checked to ensure that they are patent and properly sealed. They may be connected to a suction apparatus (50 cm H2O). Atelectasis is caused by bronchial obstruction, which may need to be cleared endoscopically.

On the abdominal side, it is possible for gastrointestinal function to be inhibited for 4 days, without this being an indication that something is wrong.

It is not a good idea to pack screws into the vertebral body at all costs, as this might disrupt the vertebra. Two screws in each vertebral body are sufficient to ensure adequate stability of the construct.

If the vertebral bodies are very osteoporotic, the fibular strut grafts are likely to protrude through the endplates. In this case, we use another, more rigid type of internal fixation device (anterior thoracolumbar plates from Synthes).

V. COMBINED APPROACH (Images - Case No. 5)

5-1 5-2
1-2 : 25-year-old man, road accident. Burst fracture with rotation of L3 and incomplete cauda equina syndrome.
3-4 : Canal completely blocked, with major anterior defect.
5 : First, posterior decompression and fusion of L2-L4 with a short construct. On D8, anterior reconstruction via a left flank approach with fibular strut grafts without internal fixation after decompression. Sequelae of erectile and sphincter dysfunction. Occasional pain.

V. A. Indications

Both anterior and posterior surgery will be required when the objectives of surgery cannot be achieved with a single approach. This may apply in the case of very comminuted fractures (complete burst fractures) with neurological lesions where there is also an anterior defect and/or gross residual anterior compression after the first stage of surgery.

The development of minimal access techniques should extend the range of indications for a combined approach.

V. B. Sequence of approaches

Logically, the first stage should be the one which will give the most rapid results, especially in terms of nerve decompression. The posterior approach satisfies this requirement. In practically all other cases, the anterior approach is used as the first stage.

V. C. Timing

Obviously, the anterior and posterior stages can be done at one operation, but surgery is likely to take excessively long, particularly as there may be substantial blood loss which cannot be adequately compensated by the use of a Cell-Saver alone. For this reason, our usual practice is to leave a week between the two stages.


Our technique produces excellent results.

Purely conservative treatment can be used to correct all types of deformity, albeit with subsequent loss of reduction; this loss does not, however, exceed the original deformity, i.e. this form of treatment prevents the kyphosis that occurs in the wake of unreduced fractures.

In contrast, patients treated surgically experience only a very minor loss of reduction, particularly with anterior surgery. With posterior surgery, the loss is in the order of 7°, as compared with 2° following anterior surgery.

In our patient material, union was obtained in 97 % of cases.

In cases where there are neurological lesions, anterior decompression has been clearly shown to be effective; 87 % of patients who had neurological deficits prior to surgery improved. Our mean neurological recovery rate is 68 %, comparing the patients' neurostatus grading before treatment and at one year after surgery.

The potential benefit of endoscopic and video-assisted techniques is not yet clear; however, these modalities appear to hold promise for the future. They do make it possible to use very short constructs by allowing less invasive anterior reconstructive and/or neurological surgery.


1. ARGENSON C, DALZOTTO H, de PERETTI F, LOVET J. Les fractures du rachis thoracique (T1-T10). A propos de 105 cas. Rev Chir Orthop 1989 ; 75 : 370-386
2. BENAZET JP, MELCHIOR B, ROY-CAMILLE R, SAILLANT G. Les “montages courts” par plaques de Roy-Camille dans les fractures lombaires. Rev ChirOrthop 1989 ; 75 (suppl I) : 118-119
3. BREHIER P. Chirurgie antérieure des fractures du rachis thoracolombaire (Thèse). Marseille. 1992
4. DEHOUX E, BARRE J, MIRANCOURT T. Intérêt de l’abord antérieur dans le traitement des fractures du rachis thoraco-lombaire. Rev Chir Orthop 1988 ; 74 (suppl. II): 296-298
5. DICK W. The “fixateur interne” as a versatile implant for spine injuries. Spine 1987; 12 : 817-831
6. FLOMAN Y, FARCY JP, ARGENSON C. Thoracolumbar spine fractures. Raven Press. New York. 199 3
7. GAUTHIER V. Chirurgie postérieure des fractures du rachis dorso-lombaire: à propos de 105 observations (Thèse). Marseille. 1995
8. GOUTALLIER D, PIAT C. Le traitement orthopédique des fractures du rachis dorso-lombaire et lombaire avec recul du mur corporéal postérieur (Burst fracture) sans ou avec troubles neurologiques mineurs. Rev Chir Orthop 1988 ; 74 (suppl. II) : 77-80
9. GUTTMAN L. Initial treatment of traumatic paraplegia. Proc R Soc Med 1954 ; 47 : 1103-1121
10. KANEDA K, HASHIMOTO T, ALBUMI K. Anterior decompression and reconstruction with anterior instrumentation for thoracolumbar burst fractures with neurologic deficits. Orthop Trans 19 9 0 ; 14 : 777-783
11. KOSTUIK JP. Anterior fixation for burst fractures of the thoracic and lumbar spine with or without neurological involvement. Spine 1988 ; 13: 286-293
12. MAGERL F, AEBI M, GERTZBEIN SD, HARMS J, NAZARIAN S. A comprehensive classification of thoracic and lumbar injuries. Eur Spine J 1994; 3: 184-201
13. LOUIS C. Le traitement orthopédique des fractures du rachis thoraco-lombaire: à propos de 120 observations (Thèse). Marseille.1994
14. LOUIS R. Chirurgie du rachis (anatomie chirurgicale et voies d’abord).Springler-Verlag. Heidelberg.1993
15. LOUIS R, BONSIGNOUR RP, OUIMINGA R. Réduction orthopédique contrôlée des fractures du rachis. Rev ChirOrthop 1975; 61 : 323 -344
16. LOUIS R, GOUTALLIER D. Fractures instables du rachis (symposium). Rev Chir Orthop 1977 ; 63 : 415-481
17. 0LERUD S, KARLSTROM G, SJOSTRUM L. Transpedicular fixation of thoracovertebralfractures. ChirOrthop 1988 ; 277 : 44-51
18. REGAN JJ, Mc AFEE, MACK MJ. Atlas of endoscopic spine surgery. Quality Medical Publishing, Inc. St Louis, Missouri 1995
19. ROY-CAMILLE R, MASSIN P. Traitement des fractures du rachis dorso-lombaire par la méthode de Böhler. Rev Chir Orthop 1989 ; 75, 479-489
20. SENEGAS J, BAULNY D, GRENIER F. L’ostéosynthèse du rachis dorsolombaire par plaque vissée antéro-latérale. Rev Chir Orthop, 1987 ; 73 (suppl II) : 157-160
21. STEFFEE AD, BISCUP RS, SITKOWSKI DJ. Segmental spine plates with pedicle screw fixation. A new internal fixation device for disorders of the lumbar and thoracolumbar spine. Clin Orthop 1986 ; 203 : 45-53
22. WEINSTEIN JN, COLLALTO P, LEHMANN TR. Thoracolumbar “burst” fracture treated conservatively: a long term follow up. Spine 1988 ; 13 : 33-38

| Orthopaedic |