Epidemiological Data
We retrospectively reviewed the cases of seventy-two consecutive patients who had a lumbar discectomy, between January 1950 and May 1983, when they were sixteen years of age or younger. These patients represented approximately 3 per cent of the 2467 patients who were less than seventeen years of age when they were evaluated, during the same time-period, at our institution because of pain in the low back or lower limb. The patients in our series represented only 0.4 per cent of the 16,586 patients who had a lumbar discectomy during the same period. DeOrio and Bianco7 reported the earlier results for fifty of these patients in 1982.
The series included forty boys and thirty-two girls (a ratio of 1.25:1). The mean age at the time of the operation was fifteen years (range, eleven to sixteen years). The mean age of the boys was fifteen years and two months and that of the girls was fifteen years. The ages, heights, and weights of the patients were plotted on growth charts27. With the numbers available, we detected no significant difference between the mean weight of the boys or girls and the mean for normal subjects matched for age and gender. However, a standardized t test indicated that the patients were approximately 2 per cent taller than control subjects matched for gender and age.
Preexisting Trauma
An injury preceded the symptoms in forty-three patients (60 per cent). Ten patients were injured while lifting a heavy object; nine, during a fall; one, while shoveling; and twenty-three, during athletic activities. These activities included basketball (seven patients); gymnastics (six); wrestling (three); running (two); and soccer, football, tennis, roller-skating, and baseball (one each).
Presenting Symptoms and Clinical Findings
All patients had sciatica at the time of presentation. Sixty-five patients (90 per cent) had low-back pain and sciatica, and seven (10 per cent) had sciatica only. Six patients (8 per cent) had numbness, and three (4 per cent), subjective weakness. Although only nine (13 per cent) of the seventy-two patients reported weakness or numbness, thirty-five (49 per cent) had objective findings of radiculopathy, as manifested by weakness, reflex changes, and sensory loss, at the time of the examination. The main clinical signs included a positive straight-leg-raising test in sixty-four patients (89 per cent), weakness in thirty-four (47 per cent), changes in the deep tendon reflex in twenty-three (32 per cent), and sensory loss in four (6 per cent). Forty patients (56 per cent) had lumbar muscle spasm; twenty-nine (40 per cent), clinical evidence of mild scoliosis that was due to low-back pain and lumbar muscle spasm; twenty-two (31 per cent), tenderness in the low-back region or over the sciatic notch; and nineteen (26 per cent), limited motion of the spine.
Radiographic Findings
Preoperative radiographs of the lumbar spine revealed at least one developmental or congenital structural abnormality in twenty-five patients (35 per cent). These abnormalities included spina bifida occulta in eleven patients (15 per cent), sacralization of the fifth lumbar vertebra in nine (13 per cent), and a sixth lumbar vertebra in nine (13 per cent). Mild non-structural scoliosis (that is, a flexible curve of the spine due to nerve-root irritation) was seen in eighteen patients (25 per cent): fourteen had lumbar scoliosis and four had thoracolumbar scoliosis. Ten of the patients who had scoliosis had no structural abnormality of the spine, six had a structural abnormality, and two had narrowing of the disc space. Nine additional patients had narrowing of the disc space, which was associated with a structural abnormality in five and was seen in isolation in four.
All patients had myelography with contrast medium to confirm the protrusion or extrusion of a disc and to rule out other intraspinal abnormalities. Most of the patients were managed before the introduction of water-soluble contrast agents and computed tomography. The myelograms were diagnostic for herniation of a disc in sixty-five patients (90 per cent). Four patients had no abnormal findings on myelography. The findings in three patients were non-diagnostic and were interpreted as showing a posterior-lying cul-de-sac in two and a possible disc herniation in one. A posterior-lying cul-de-sac is an anatomical variation that occurs in conjunction with increased epidural fat anterior to the thecal sac and may obscure the myelographic identification of a protruded disc. This phenomenon is rarely a diagnostic problem in the era of computed tomography and magnetic resonance imaging.
Non-Operative Treatment
All patients were managed non-operatively before the discectomy. Non-operative treatment included hospitalization with bed rest, physical therapy, pain medication, muscle relaxants, and, for some patients, traction. The duration of non-operative therapy varied, depending on the response to treatment. Patients who had radicular pain that was unresponsive to non-operative therapy or who had a progressive neurological deficit, or both, had additional radiographic evaluation in preparation for the operation. For the seven patients who had non-diagnostic findings on myelography, the decision to operate was based on the presence of pain in the lower extremity and a progressive motor deficit. All seven patients had a positive straight-leg-raising test, and all were found to have a herniated disc at the time of the operation.
Operative Treatment
The interval between the onset of symptoms and the initial operation was 6.4 months (range, two weeks to thirty-six months). The operation consisted of a unilateral partial hemilaminectomy with a partial medial facetectomy and removal of the extruded or protruded disc, with decompression of the affected nerve root. The end plates were not removed intentionally, and the disc usually was removed only on the symptomatic side. According to the operative reports, only two patients had extruded free fragments. Most patients had a protruded disc that compressed the overlying nerve root. According to the operative reports, the surgeon often had difficulty removing the discs because they were often adhered to the end plates. It appears that, because of this adherence, removal of a disc from the interspace is frequently more difficult in children than in adults. The consistency of the disc was frequently fibrotic, and the nucleus pulposus was moist and gelatinous.
Operative discectomy was performed at the interspace between the fourth and fifth lumbar vertebrae in thirty-four patients (47 per cent), at the interspace between the fifth lumbar and first sacral vertebrae in thirty-two (44 per cent), at the interspaces between the third and fourth lumbar vertebrae and the fourth and fifth lumbar vertebrae in three (4 per cent), and at the interspaces between the fourth and fifth lumbar vertebrae and the fifth lumbar and first sacral vertebrae in three (4 per cent). Twelve patients (17 per cent) had a two-level exploration; ten (14 per cent), a one-level bilateral exploration; and two (3 per cent), a three-level exploration. A multilevel unilateral exploration was performed because of non-diagnostic radiographic findings, evidence of multilevel myelographic defects, or failure to find notable nerve-root compression at the level of the initial operation. From 1977 to 1983, no patient had a multilevel unilateral exploration of nerve roots.
All bilateral procedures were performed at one level only, and such procedures were done because of an anterior midline defect (five patients), a bilateral myelographic defect (three patients), and indeterminate findings on myelography (two patients). One patient who had indeterminate myelographic findings had a bilateral exploration of the interspace between the fourth and fifth lumbar vertebrae that revealed a midline disc protrusion. Of the two patients who had myelographic findings of a posterior-lying cul-de-sac, one had spina bifida occulta of the first sacral vertebra and had a partial hemilaminectomy of the fifth lumbar vertebra as well as excision of a laterally protruded disc at the interspace between the fifth lumbar and first sacral vertebrae. The other patient had a partial unilateral hemilaminectomy of the fourth and fifth lumbar vertebrae and excision of a laterally protruded disc at the interspace between the fourth and fifth lumbar vertebrae. Of the four patients who had normal findings on myelography, one had a two-level exploration that revealed a midline disc protrusion at the interspaces between the fourth and fifth lumbar vertebrae and between the fifth lumbar and first sacral vertebrae. The other three patients had a single partial hemilaminectomy: one had a midline disc protrusion at the interspace between the fifth lumbar and first sacral vertebrae and the other two had a laterally protruded disc at the interspace between the fifth lumbar and first sacral vertebrae.
At the time of the initial operation, twelve patients (17 per cent) had a posterolateral spinal arthrodesis. The arthrodesis extended from the fourth lumbar to the first sacral vertebra in nine patients, from the fourth to the fifth lumbar vertebra in two, and from the fifth lumbar to the first sacral vertebra in one. The indication for arthrodesis included bilateral incompetence of a facet joint caused by developmental or postoperative changes. This finding was determined by visual observation at the time of the operation. On the basis of this indication, an arthrodesis was performed in eight of the twenty-five patients who had a coexistent structural anomaly (Table II), in two of the fourteen patients who had a multilevel hemilaminectomy, and in two of the ten patients who had bilateral exploration at the same disc space.
Collection and Analysis of the Follow-up Data
Short-term follow-up data (from the first postoperative year) were obtained from medical records, office reports, or written reports from the local physicians. The overall duration of follow-up ranged from twelve to forty-five years (mean, 27.8 years). A limitation of the present study was that we could not personally evaluate the current neurological signs and radiographs of some patients because they lived too far away. Long-term follow-up data were obtained primarily by a review of the medical records, administration of questionnaires, telephone conversations with patients, and evaluation of the operative records of patients who had a reoperation. All patients were contacted during 1995. The pain status and the level of activity were used as the most representative and substantial criteria in these follow-up evaluations7,35.
Kaplan-Meier survivorship analysis18 was performed to estimate survival without an additional operation (a lumbar discectomy or an arthrodesis, or both) for the overall population of patients, with use of arthrodesis at the initial operation as well as the age and gender of the patient as risk factors. The log-rank test30 was used to compare the survivorship curves of the different groups of patients. Survival was estimated as the duration from the initial operation to the first reoperation (for patients who had a reoperation) or as the duration from the initial operation to the latest follow-up evaluation (for patients who had no reoperation).
Reoperation was chosen as the end point of survival and was used in the calculation of the cumulative rates of survival because it is easily defined. Neurological deficit, pain, and function were not considered because, although they may be used as end points, their usefulness is limited by subjectivity and by the difficulty in determining when pain began or when a neurological deficit first developed. The use of a reoperation as a measure of survival has been addressed previously31.
Postoperative Complications
Postoperative complications included a superficial wound hematoma in one patient and a deep venous thrombosis that was diagnosed clinically ten days after the operation in another. There were no postoperative wound infections or neurological deficits. One patient had a pseudarthrosis at the site of the arthrodesis that necessitated two subsequent operations to achieve fusion; the final outcome was successful. One patient who had an abnormal finding on myelography continued to have pain in the lower extremity after the discectomy. Subsequently, operative release of a contracted iliotibial band resulted in complete relief of pain.
Reoperations
Twenty patients (28 per cent) who had been followed for a mean duration of 27.8 years (range, twelve to forty-five years) had one reoperation or more because of recurrent pain in the low back or lower extremity. The first reoperation was performed at a mean of 9.7 years (range, one to forty-five years) after the initial procedure; sixteen patients had a repeat exploration of the same intervertebral level that had been treated during the index operation, and four patients had exploration of a different level. Overall, thirty reoperations were performed in twenty patients: fourteen patients had one reoperation, four had two reoperations, one had three, and one had five. Twelve of the twenty patients had an arthrodesis of the lumbar spine (Table III).
Short-Term Clinical Outcome
In the first year after the lumbar discectomy, the results were encouraging. Overall, twenty-eight (39 per cent) of the seventy-two children had no low-back pain or sciatica and could participate in any activity, and thirty-nine (54 per cent) had only mild discomfort and limitation of some strenuous activities. Five patients (7 per cent) continued to have moderate pain in the first postoperative year. Of those five patients, one who had had a discectomy and an arthrodesis needed a reexploration and arthrodesis after pseudarthrosis developed at the site of the arthrodesis. Another patient needed a release of the iliotibial band for complete relief of pain in the lower extremity. Two patients, who had low-back pain, had a gradual improvement during the first postoperative year. Only one of the seventy-two patients had a poor result because of persistent moderate pain in the lower limb.
Long-Term Clinical Outcome
The encouraging early results did not persist. One year after the initial operation, a patient who initially had an excellent result had a recurrence of the radicular pain and needed a second operation at the same level. As already stated, twenty patients (28 per cent) who had been followed for a mean duration of 27.8 years needed one reoperation or more because of pain in the low back or lower limb.
Forty-eight (92 per cent) of the fifty-two patients who did not need a reoperation had no pain in the low back or lower limb or had only mild pain related to strenuous activity. Fifty-one patients (98 per cent) could participate in daily activities with no or only mild limitations (Table IV). Fifteen of the fifty-two patients participated in strenuous activities. Three patients used a corset because of low-back pain.
At the latest follow-up evaluation, sixty-six (92 per cent) of the original group of seventy-two patients had no or mild pain in the low back or lower extremity (Table IV). Forty-seven patients (65 per cent) could participate in activities without any limitations related to the back. Twenty-two patients (31 per cent) could participate in most daily activities but with occasional discomfort in the back after performing strenuous activities. Some of these patients needed a second or third operation. At the latest follow-up evaluation, twenty-one (29 per cent) of the seventy-two patients participated in strenuous activities. Only six patients (8 per cent) had marked discomfort, and three of them had frequent pain in the low back or lower extremity that severely limited the ability to perform daily activities.
Statistical Analysis
With the numbers available for study, we could not detect any association between the preoperative duration of symptoms and the long-term outcome. Similarly, no significant difference could be detected, with respect to the long-term outcome, between the patients who had had an arthrodesis at the initial operation and those who had not (p = 1.000, Fisher exact test). Fifty-seven (95 per cent) of the sixty patients who had had a discectomy without an arthrodesis had a satisfactory result during the first postoperative year. Although seventeen of the sixty patients needed a late reoperation, fifty-five (92 per cent) had no or mild pain at the latest follow-up examination and fifty-seven (95 per cent) had no or mild limitation of their activities (Table IV). Ten of the twelve patients who had had an arthrodesis had a satisfactory result during the first postoperative year. Only three of them had a late reoperation, and all of them had a satisfactory result at the latest follow-up examination (Table IV).
The addition of an arthrodesis to the discectomy did not appear, with the numbers available, to have a significant effect on the long-term outcome for the patients who had a structural abnormality of the lumbar spine (such as sacralization of the fifth lumbar vertebra, spina bifida occulta, or a sixth lumbar vertebra) (p = 1.000, Fisher exact test) (Table V). Eight (32 per cent) of the twenty-five patients who had a structural abnormality of the lumbar spine had a discectomy with an arthrodesis. Although one of these eight patients had a reoperation, seven had no or mild pain at the latest follow-up examination and all had no or mild limitation of their activities. The other seventeen patients who had a structural abnormality of the lumbar spine had a discectomy without an arthrodesis. Six of these patients needed a reoperation, but fifteen had no or mild pain in the back or lower limb at the latest follow-up examination and sixteen had no or mild limitation of activities (Table V). With the numbers available for study, we could not detect a significant difference, with respect to the long-term outcome, between the patients who had a preexisting structural abnormality of the spine and those who did not have such an abnormality (p = 0.412, Fisher exact test).
Survivorship analysis showed that the overall probability that a patient would not need a reoperation after a lumbar discectomy was 80 per cent at ten years and 74 per cent at twenty years after the initial operation (Fig. 1). With the numbers available, we could not detect a significant difference, with respect to the probability that a reoperation would not have been performed by twenty years, between the patients who had had an arthrodesis at the initial discectomy and those who had not (83 compared with 72 per cent, p = 0.65) (Fig. 2), between boys and girls (77 compared with 72 per cent, p = 0.44) (Fig. 3), or between the patients who were fifteen years of age or older and those who were less than fifteen years of age at the time of the initial operation (69 compared with 77 per cent, p = 0.39) (Fig. 4).
The true prevalence of herniated lumbar discs in children and adolescents is unknown. The pediatric patients in the present series represented only 0.4 per cent of all patients who had a lumbar discectomy at our institution. Only 3 per cent of the 2467 children and adolescents who were first seen at our institution because of low-back pain, during the same time-period, eventually had a lumbar discectomy. Rugtveit32 reported that seven (0.8 per cent) of 840 patients who were managed operatively for a herniated lumbar disc were children and adolescents. O'Connell28 reported that thirty-eight (3.2 per cent) of 1200 patients who had a protrusion of a lumbar disc were children. The reported prevalence of herniated lumbar discs is higher among Japanese children19,26. Kurihara and Kataoka21 reported that seventy (15 per cent) of 456 operations performed for herniated discs in Japan were for patients who were less than twenty years of age.
Although some authors2,3,5-7,21,24,34 have suggested that the prevalence of herniated discs is greater in boys than in girls, others have reported that the prevalences are about equal9-12,17. The present series demonstrated that the results of treatment of herniated discs were not affected by gender.
In two series, more than 90 per cent of twenty-three and twenty-five patients who were managed operatively for a herniated disc had involvement of a caudad disc6,10. In the present series, the discs at the interspaces between the fourth and fifth lumbar vertebrae and the fifth lumbar and first sacral vertebrae were most frequently involved (92 per cent); the two levels were affected about equally. Six of the patients in the present series had involvement at two levels; however, protrusions of discs at multiple levels have rarely been reported38. In the series of Fisher and Saunders11, only one of forty-three patients who ranged in age from eleven to twenty-one years had a protrusion of a disc at two levels.
The cause of herniated nucleus pulposus in children and adolescents is not known12. Several authors have stressed the important role of trauma1,2,7,11,12,24,28,33. A single traumatic episode, however, is not likely to lead to prolapse of a disc unless preexisting degenerative alterations predispose the disc to protrusion1,5,6,21. In the present series, an injury preceded the symptoms in forty-three (60 per cent) of the seventy-two patients and about one-half of the injuries occurred during participation in athletic activities. Many of the herniations of discs in children may be apophyseal avulsions or fractures of the cartilaginous end plates. These disorders are difficult to diagnose with use of myelography with contrast medium, but they are more apparent with magnetic resonance imaging, which is now more commonly used.
The presenting signs and symptoms of the herniated discs in the children and adolescents in our series were similar to those observed in adults. The primary symptom in all of our patients was pain in the lower extremity. Unlike adults, children rarely have an overt neurological deficit5,12,33. In series ranging in size from twenty-five to 101 children and adolescents, the straight-leg-raising test was positive for more than 90 per cent of the patients10,14,33,34. The present study, in which the test was positive for 89 per cent (sixty-four) of the patients, confirmed the importance of this finding in the diagnosis of a herniated disc in young patients, particularly in the absence of any neurological deficit.
The radiographic findings did not reveal any common predisposing factors. Other authors have noted that certain structural anomalies of the spine are more prevalent in children and adolescents who have a herniated disc than in the general pediatric population5,7,10,17. In the present series, no difference was found, with respect to the long-term outcome, between the patients who had had a preexisting abnormality of the lumbar spine (such as spina bifida occulta or transitional segmentation) and those who had not had such an anomaly. A radiographic investigation of the lumbar spine is important to exclude other diseases, such as congenital malformations, tumors of bone or the spinal cord, and infectious disease.
Postmyelographic computed tomography or magnetic resonance imaging is important in the radiographic evaluation. In the present series, most of the two-level explorations and bilateral procedures were conducted in the years before the introduction of water-soluble contrast agents and computed tomography. We currently perform postmyelographic computed tomography or magnetic resonance imaging routinely before attempting any operative procedure for the treatment of lumbar disc disease in children. Magnetic resonance imaging provides excellent detail of the intrathecal anatomy, which is not seen with computed tomography12,16. Dysraphic states, benign or malignant neoplasms, and inflammation (discitis and osteomyelitis) are prominently revealed. In an earlier review, Love demonstrated that 5 per cent of 115 patients who were seen because of a lumbar disc syndrome had an unexpected lesion of the intraspinal canal22.
The treatment of herniated nucleus pulposus in children and adolescents remains controversial. Although most investigators have recommended non-operative measures initially, many have believed that operative treatment was more effective1,3,20,34; however, no controlled studies, to our knowledge, have been performed to support this hypothesis. Unless there is a notable neurological deficit, the initial treatment of a symptomatic herniated lumbar disc should be non-operative. However, in children and adolescents who have persistent pain and disability or who have findings of radiculopathy, operative intervention is appropriate. A unilateral partial hemilaminectomy, foraminal decompression (if it is needed), and partial discectomy without arthrodesis is the procedure of choice in such patients.
With the numbers available, we could not detect a significant difference, with respect to the probability that a reoperation would not be needed within twenty years after the initial operation, between the patients who had had an arthrodesis at the time of a discectomy and those who had not (p = 0.65, log-rank test). These data support the suggestions of other authors that, in this age-group, arthrodesis has little place in the treatment of a ruptured lumbar disc associated with normal vertebrae2,9,21. Arthrodesis rarely is necessary in pediatric patients who have a lumbar discectomy, and it should be performed only to prevent potential instability of the spine. The main indication for arthrodesis is incompetency of the facet joints that is due to either iatrogenic29 causes or developmental variations. In the present series, arthrodesis did not improve the clinical outcome or decrease the rate of recurrence either in the patients who had a structural abnormality of the lumbar spine or in those who did not.
The patients in the present study had no postoperative wound infections or neurological deficits, and they had a high rate of early satisfactory clinical results. The early results of a lumbar laminectomy and discectomy in children and adolescents have been predominantly satisfactory in most reported series (Table I)6-8,10,14,21. However, despite these early successful results, some patients might need a second or third operation to treat radicular symptoms due to herniation of the same or another lumbar disc or to treat symptoms suggestive of spinal instability. In the present series, in which the duration of follow-up was as long as forty-five years, twenty (28 per cent) of the seventy-two patients needed a reoperation. Because most patients (72 per cent) did not need another procedure after the initial operation, postoperative restrictions of normal activities and sports were not justified.
In conclusion, a discectomy does not always provide the final solution to lumbar disc disease in children. However, careful selection of patients, conscientious follow-up, and sometimes additional treatment can produce satisfactory long-term results in most patients. Spinal arthrodesis does not offer any benefit for children who have a herniated lumbar disc in association with normal vertebrae and no signs of spinal instability.