Experimental investigation of the mechanisms and pathology of spinal cord injury began early this century when Allen developed a standardized low-force model of injury to the spinal cord (1). He recognized that secondary damage develops progressively during the first few hours after the injury. These changes have been attributed to vascular mechanisms that ultimately lead to spinal cord infarction (2). The earliest changes occur within minutes of an experimental injury; they consist of hyperemia and small hemorrhages in the central gray matter of the cord. Microscopic pericapillary hemorrhages coalesce, and by 1 hour, grossly visible hemorrhagic areas appear in the central gray matter. Tissue oxygen saturation is decreased in the injured segments during these early stages (3). Osterholm and Mathews (4) observed that there is an accumulation of vasoactive amines in the injured segments of the cord during this initial period, suggesting that vasoconstrictive substances in the injured area might be responsible for vasospasm and decreased blood flow, which potentiate damage. Since that time there has been considerable controversy regarding the nature and significance of these changes and their role has not been emphasized (5,6). Within 4 hours of injury, the gray matter is infarcted and spreading white matter edema is noted that progresses at 8 hours to infarction of the white matter (7,8). The delay in the appearance of white matter necrosis has raised the hope of preventing irreversible damage to long tracts by therapeutic intervention during the first few hours. These attempts, for the most part unsuccessful, have included laminectomies and myelotomies to decompress the central edema and hemorrhage; early spinal cord cooling to prevent secondary damage to the white matter (9,10); and hyperbaric oxygenation during the early stages of experimental cord injury. A variety of pharmacologic agents, including corticosteroids, diuretics, plasma expanders (including lowmolecular-weight dextran), dimethyl sulfoxide, α-methyltyrosine (a monamine synthesis blocker), and endogenous opiate antagonists,
have been found to have a beneficial effect in experimental cord injury (11,12). However, none of these surgical or pharmacologic interventions have been proven to alter the prognosis of human spinal cord injury.

The one exception, albeit now controversial, has been the often-cited randomized controlled trial that reported beneficial effects of high-dose methylprednisolone administered soon after the injury (13). This study was driven by experimental models in which corticosteroids were found to be beneficial in a rather restricted range of mechanical force applied to the spinal column (typically between 300 and 500 g/cm²). Injuries produced by lower forces did not result in paraplegia, and higher forces caused irreversible damage regardless of treatment. It is only an occasional clinical injury that conforms to forces within this narrow window. At present, of all these forms of therapeutic intervention, only corticosteroids are sufficiently safe to be recommended for wide clinical use; however, the validity of the conclusions of this study has been questioned from a number of perspectives. Soon after the results of the study were published, treatment with high-dose corticosteroids was considered obligatory but subsequent criticisms have emerged based largely on the study design and the analysis of results (14,15). There now seems to be a consensus that corticosteroids may offer only a slight benefit. Nevertheless, many centers continue to utilize a loading dose of 30 mg/kg methylprednisolone over 15 minutes followed by 5.4 mg/kg per hour for 23 hours. Others have recommended that the medication be continued for only 4 hours.

It cannot be overemphasized that the care of the patient with spinal cord injury begins at the scene of the accident. The significance of this early phase of treatment is underscored in a study by Rogers (16), who, a half decade ago, found that one of ten patients with spinal cord injury deteriorated between the time of the accident and the time of arrival at a medical facility. Because most of these patients do not lose consciousness, simple questioning reveals that the patient is paralyzed or has the cardinal signs of spinal injury, namely, numbness or severe neck or back pain. It has become well-established practice that in the presence of any of these symptoms or if the patient is unresponsive after a severe injury, a spinal injury is suspected and the patient is handled in such a way as to avoid any further dislocation of the spinal segments (17). Car-diopulmonary resuscitation when appropriate, insurance of a proper airway, control of obvious hemorrhage, and immobilization of evident long bone fractures are the other integral parts of initial care. If spinal cord injury is confirmed, the option of using intravenous methylprednisolone is available at the dose described in the preceding.

Many authorities have emphasized the utility of standardized protocols that cue physicians and nurses to various aspects of treatment and, more importantly, the avoidance of further injury at each stage (18).

Once the patient arrives at the emergency room, continued immobilization of the suspected spinal injury must be ensured until definitive radiologic studies are obtained. The first priority is an adequate airway. Tracheal intubation is indicated if the patient is hypoventilating, unable to handle secretions, or comatose. Most patients can be intubated nasotracheally or by direct laryngoscopy without extending the head on the neck. The recent development of a fiberoptic laryngoscope has greatly facilitated intubation while eliminating the need to alter the position of the head and neck. With the flexible laryngoscope passed through a nasotracheal tube, the tube may be manipulated and its placement in the trachea determined under direct vision.

Most patients with high thoracic and cervical injuries have mild hypotension and bradycardia that is caused by sympathetic failure and peripheral vascular vasodilation that accompanies the state of “spinal shock.” These patients appear well perfused; their skin is warm; they do not appear in any way hypovolemic
but nonetheless they respond to rapid infusions of crystalloid solutions or plasma expanders. Only infrequently is it necessary to use vasopressors to support blood pressure. Pressor use should be dictated by signs of hypoperfusion of the brain, heart, or kidneys and not solely by the failure of fluid infusions to bring the systemic pressure to normal. Atropine also can be used to restore blood pressure briefly if the hypotension is associated with bradycardia.

Recognition of other life-threatening injuries is of vital importance and sometimes can be quite difficult in the paralyzed patient. With spinal gunshot wounds a second small bullet entry in the back may be missed while concentrating on the more obvious neck wound. Signs of hypovolemic shock may be absent in a patient with even profuse visceral bleeding because sympathetic tone is absent; therefore, the expected vasoconstriction and pallor may not be present; instead of the expected tachycardia, the patient may be bradycardic. It is then possible to misinterpret hypotension as caused by spinal shock while the patient continues to bleed. It should be emphasized that spinal shock is usually not responsible for profound hypotension, and it is not associated with a continuing drop in systemic blood pressure while the patient is receiving fluids. Furthermore, an acute abdominal injury, even when resulting in massive spillage of blood or gastrointestinal contents, may not be associated with any signs of peritoneal irritation in the patient with a complete spinal lesion. Therefore, when in doubt, it is appropriate to look for free air by performing a careful abdominal tap and abdominal plain films or computed tomography (CT) scan.

Head injuries frequently accompany and complicate spinal injuries, and their proper recognition is, of course, vital. As an extension of preceding comments, it is almost always incorrect to attribute a deteriorating state of consciousness to cerebral hypoperfusion related to spinal shock. Conversely, a spinal cord injury must be suspected in the comatose, head-injured patient who exhibits muscular flaccidity rather than reflex posturing. Only in the agonal stages of deterioration from a brain injury does flaccidity arise. Obvious long bone fractures should be properly immobilized to assist in the prevention of fat embolism and skin erosions.

During the acute stages of spinal cord injury, as also noted in the following, the bladder and gastrointestinal tract are atonic. An indwelling urinary catheter must be inserted except in institutions where a special team of trained personnel is available continuously for intermittent catheterization. If the latter is available, the patient should be committed to this form of treatment from the beginning. Otherwise, an indwelling catheter, placed with careful sterile technique, may be preferable (19). Gastric atony can result in significant gastric dilatation that compounds respiratory failure by causing upward pressure on the diaphragm, or it may promote aspiration. These patients are understandably anxious and as a result swallow a considerable amount of air. Because they do not feel pain, gastric dilatation can reach massive proportions rapidly and, if untreated, result in gastric rupture, a frequent cause of early death in past eras. These problems can be prevented by early insertion of a nasogastric tube.

Following these initial steps, a careful neurological assessment must be undertaken to determine the level and severity of the spinal cord injury. In cervical injuries the motor examination usually establishes the level. Patients with cervical and high thoracic lesions will have lost intercostal movement and their respirations are fully diaphragmatic. With injuries in the upper cord above C5, the phrenic motor neuronal pools are damaged and respiration is impaired. With complete injuries at the T1 level, there is paraplegia and some impairment of intrinsic hand function. At the C7-T1 level, there is also impairment of finger and wrist flexors (C8 nerve root). At the C6-C7 level, the triceps, wrist extensors, and forearm
pronators (C7 root) also are lost. At C5-C6, in addition to the preceding, the biceps and forearm supinators are weak (C6 root), and at C4-C5 the deltoid and supraspinatus and infraspinatus muscles are impaired (C5 root). Thoracic injuries result in paraplegia, and the level can be determined by correlating the radiographic findings and the sensory examination. Lower thoracic and thoracolumbar injuries often result in complex neurological syndromes with mixtures of upper and lower motor neuron (conus), and peripheral nerve (cauda equina) deficits. Lumbar injuries, of course, result in cauda equina deficits, which are usually incomplete, and as noted in the following, reversible for a considerable period.

The sensory examination also is of utility in determining the level, type, and severity of the neurological injury. Careful testing of pain sensation must be carried out to identify areas of spared sensation, especially in the sacral and perineal regions, which are of considerable prognostic significance. In determining the level of injury by pinprick examination, recalling that cutaneous branches of the cervical plexus, corresponding mostly to C3 and C4, innervate the skin over the lower collar and upper chest areas, sometimes as far down as the nipples, avoids the error of calling a cervical lesion incomplete because sensation is preserved in the upper chest. Careful sensory examination of the upper extremities (C5 through T1) and the axilla (T1 and T2) usually demonstrates a correspondence between the sensory and motor functions. It is also of value to identify cases of complete loss of motor function and pain sensation but with preserved proprioception (posterior column) function.

Deep tendon reflexes initially usually are absent below the level of a complete spinal cord lesion; however, they may return within a few hours of the injury, and their presence should not be taken as an indication of an incomplete lesion. In thoracolumbar injuries, the presence of an anal or bulbocavernosus reflex in the absence of motor power or sensation indicates that spinal shock has subsided and there is an upper motor neuron lesion as opposed to a cauda equina lesion. The latter, as discussed, has a much better prognosis. During the initial evaluation, appropriate films can be obtained, as discussed, and it is advisable to also obtain routine blood tests and arterial blood gases.

Partial spinal cord injuries generally fall into one of the following neurological syndromes. The Brown-Sequard syndrome of hemiparesis with ipsilateral loss of proprioception and contralateral impairment in pain perception can occur in relatively pure form from penetrating injuries such as stab wounds. The prognosis is good when caused by closed injuries, and nearly 90% of these patients make a functional recovery.

The central cord syndrome of Schneider is characterized by loss of strength of the upper extremities out of proportion to the weakness of the lower extremities. The sensory deficits and impairment of bladder and bowel function are variable and unpredictable (20,21). The prognosis is good, and close to 60% of patients regain the ability to ambulate, although they may be left with some impairment of hand function. Usually, function of the legs returns first, followed by bladder function and ultimately function of the upper extremities.

The anterior cord syndrome consists of loss of motor function and pain sensibility below the level of the lesion, with preservation of posterior column function. The prognosis here is poor, but occasionally a patient improves after decompression of the cord by removal of an anterior disc or bone fragment. The rare instances of relatively pure impairment of posterior column function carry a good prognosis. As discussed, lesions of the conus medullaris are associated with a poor prognosis for return of bladder and rectal function, but the prognosis is good when the deficit is due to root dysfunction, which is sometimes hard to distinguish from conus lesions in cases of thoracolumbar injuries.

When there is spinal malalignment with obvious or possible compression of the spinal cord, we and the neurosurgeons who advise us have felt that rapid reduction of the dislocation is preferable in all patients and is essential
in those patients with only a partial neurological injury. There is less urgency in reducing lumbar fractures that compress only the cauda equina.

Thoracic fractures usually present with either no neurological deficit or with a complete neurological syndrome. These fractures are usually stable and reduction is rarely necessary unless marked kyphosis is present. Thoracolumbar fractures frequently require open reduction to decompress the cornus medullaris and to ensure stability. In these cases, reduction should be accomplished as early as feasible once the patient is stable from the hemodynamic point of view and other urgent injuries have been attended to.

With rare exceptions, displaced cervical fractures should be treated initially by skeletal traction. The halo ring is preferred in our institutions, but other methods of skeletal traction are also satisfactory. There is no clear rational or scientific justification for the traditional method of reducing these dislocations by slow increases in weights over periods of many hours or days (22). In cases of high cervical dislocations, reduction usually can be easily accomplished with weights of no more than 10 to 15 pounds. For lower cervical dislocations, traction can be initiated with 20 pounds under careful radiographic control, preferably with an image-intensifying fluoroscopic unit, to detect early distraction. The weights can be increased by increments of 5 to 10 pounds every few minutes, with radiographic observation between each increase, until reduction is accomplished. This may take as much as 70 to 80 pounds, which is safe for a short period of time, provided that excessive distraction at any level is not detected radiographically. Only an exceptional dislocation (usually associated with a unilateral jumped facet) fails to be reduced within 1 or 2 hours with 80 pounds of traction and careful administration of parenteral muscle relaxants. Once reduction is achieved, the weight can be reduced to approximately 20 pounds, which is usually sufficient to maintain proper alignment until a definitive form of immobilization is chosen.

It should be noted that a multicenter study has documented delayed deterioration in approximately 5% of patients with spinal cord injuries (23). Others indicate the rate is closer to 1% if considering only irreversible syndromes of “ascending cord necrosis.” Usually, these are younger patients with cervical cord injuries and there is substantial swelling of the cord on magnetic resonance imaging scan. In most of these patients, the decline in function was associated with a specific event (e.g., surgery, moving the patient, or application of traction, etc.). In the case of cervical cord injury and delayed ascension of the deficit, there has been an association with mortality. Furthermore, deterioration within 24 hours was associated with traction, between 24 and 72 hours with hypotension, and later deterioration, with vertebral artery injury (24).

In all patients who are unconscious from a head injury, or who complain of neck or back pain, or in whom a spinal injury is suspected for any reason, initial anteroposterior and lateral plain films of the area in question should be obtained with the patient properly immobilized. In cases of suspected cervical spine injuries, an open-mouth view of the odontoid is also essential. Careful, systematic review of plain spine films should be performed to exclude injury at each spinal level (25). It is advisable to obtain a chest film during this initial period.

The availability of emergency CT scanning greatly facilitates the early evaluation of cervical injuries. This has become a major mode of diagnostic evaluation after plain films. In patients with a short neck or large shoulders, CT scanning may be the only way to properly evaluate the lower cervical area. The cervical spine must be studied at least down to the C7-T1 junction in cases of suspected cervical injury. Computed tomography studies may also be necessary to detect a unilateral jumped facet or bone fragments that have been displaced into the spinal canal or root foramen. They are valuable in confirming that complete
reduction of a dislocation has been achieved when an anterior fusion is contemplated. Between 15% and 20% of patients who have cervical spinal injuries have no overt radiographic abnormality on plain films, but two thirds of these have abnormalities on CT scan. More extensive fractures than appreciated with plain radiographs have been visualized in more than 50% of patients who had CT scans (26). Flexion and extension films of the neck in the acute stage of a spinal injury are not recommended except in patients with a normal neurological examination who complain of neck pain and have no obvious evidence of dislocation or fracture on plain radiography. We have had no experience with discography which was used in the past (27). Magnetic resonance imaging (MRI) has proved to be of benefit in evaluating the chronically injured spinal cord (28). This modality of imaging can be used after bony alignment has been achieved.