We reviewed the medical records and radiographs of 109 consecutive patients with juvenile idiopathic scoliosis who had been followed between 1970 and 1992 at the Princess Margaret Rose Orthopaedic Hospital, Edinburgh. The study cohort was derived from 1362 patients who had been evaluated for idiopathic scoliosis during this time-period. Ninety-one of the more recently seen patients have been under the care of the senior one of us (M. J. McM.). During this period, the mean prevalence of new patients who had juvenile idiopathic scoliosis was 8 per cent (range, 4 to 22 per cent) per year.
The patients were asked to return for follow-up every four to six months, depending on the severity of the curve. Anteroposterior and lateral radiographs of the spine were made, with the patient standing, at the initial visit and at each follow-up visit. Side-bending radiographs of the spine were not made routinely. Initial measurements were made on all of the radiographs by the senior one of us, and then measurements were made by the other one of us (C. M. R.). The final values were based on a consensus between us; intraobserver and interobserver errors were not estimated.
The levels of the neutral, end, apical, and stable7 vertebrae were recorded, the scoliotic curve was measured2, and the rib-vertebra angle difference10 and the vertebral rotation12 were measured at the apical vertebra of the thoracic curve (Table I). Spinal decompensation, or list, was measured on the anteroposterior radiograph as the amount of lateral displacement of the seventh cervical vertebra with respect to the central sacral line. Lateral translation of the thoracic cage, distinct from list, was the degree of displacement of the thorax from the central sacral line14. Sagittal alignment was assessed on the lateral radiograph by measurement of the thoracic kyphosis from the third to the twelfth thoracic vertebra and the lumbar lordosis from the first to the fifth lumbar vertebra1.
Age, Gender, and Side of the Curve
The mean age at the time of the diagnosis of the curve in the 109 patients was six years and ten months (range, three years and six months to nine years and eleven months). There were sixty-seven girls and forty-two boys (Table I). The mean age of the boys when they were initially seen was five years and eight months (range, three years and six months to nine years and seven months), which was slightly younger than the mean age of the girls (seven years and two months; range, three years and eight months to nine years and eleven months) when they were first seen. The ratio of girls to boys was 1:1.6 in the age-group of less than six years old and 2.7:1 in the age-group of more than six years old. Female predominance (11:1) was most pronounced in the group that had a primary progressive lumbar curve.
The primary progressive thoracic and thoracolumbar curves and the resolving curves were right-sided in sixty-seven children and left-sided in thirty (Table I). Eleven of the twelve primary progressive lumbar curves were left-sided. In the group of children who were less than six years old, there were an equal number of right and left-sided curves, whereas the older children had more right-sided curves (ratio, 3.9:1). There was no meaningful difference in the mean age at the time of diagnosis among the four groups of children who had a progressive curve; however, the children who had a resolving curve were younger when they were first seen (mean age, four years and two months; range, three years and six months to five years and six months) than those who had a progressive curve.
Pattern of Progressive Curves
Attempts to classify the curves with use of the radiographs that were made at the time of the initial visit were unsuccessful, as the deformity often changed as a result of extension of the primary curve or the development of secondary curves that later became fixed. The apical vertebra remained constant, even when the curve extended to involve additional vertebrae, and it was used to predict the final pattern of the deformity. The development of secondary curves and the degree of spinal imbalance were similar for patients in whom the apical vertebra was at the same level; hence, these patients were grouped together (Figs. 1-A and 1-B).
Group 1
Fifty-seven of the children had a progressive thoracic curve with the apex usually at the eighth thoracic level (Group 1). As the curve progressed, the caudad neutral vertebra remained at the eleventh or twelfth thoracic level and the lumbar vertebrae rotated in a direction opposite to the rotation of the thoracic curve. There were two curve patterns: a single mid-thoracic curve (Group 1A) and a major mid-thoracic curve with a secondary minor lumbar curve (Group 1B).
There were twenty-eight patients in Group 1A. The apical vertebra was at the eighth thoracic level in twenty-four patients and at the ninth thoracic level in four. The curves extended from a neutral vertebra at the fourth or fifth thoracic level to a neutral vertebra at the eleventh or twelfth thoracic level, and the stable vertebra was usually at the twelfth thoracic or first lumbar level. As the thoracic curve progressed, a slight secondary lumbar curve developed, but no patient had a lumbar curve that crossed the mid-sacral line. The body of the fourth lumbar vertebra remained relatively square to the sacrum (Figs. 2-A and 2-B). Lateral translation of the thoracic cage increased as the curve progressed because of the absence of a compensatory lumbar curve, and there was a list toward the convexity of the curve. There was hypokyphosis of the thoracic spine and normal lumbar lordosis.
Twenty-nine patients were in Group 1B. The apical vertebra was at the eighth thoracic level in twenty-six patients and at the seventh thoracic level in three. In the early stage, the thoracic curves appeared similar to those in Group 1A, with the caudad neutral and stable vertebrae at either the eleventh or the twelfth thoracic level. However, as the thoracic curve progressed, a lumbar curve developed, with the apex at the second or third lumbar level and the neutral vertebra at the fourth lumbar level, which was always tilted toward the convexity of the curve (Figs. 3-A and 3-B). The two curves ultimately progressed at the same rate, and spinal balance was restored; this resulted in a less severe cosmetic deformity than in Group 1A; however, the sagittal profile was the same in all of the curves in Group 1.
Group 2
Group 2 consisted of twenty-seven patients who had a single progressive thoracic curve with the apex at the ninth thoracic vertebra (twelve patients), the tenth thoracic vertebra (eleven patients), or the eighth thoracic vertebra (four patients) (Fig. 1-B). In this group, unlike in Group 1, the lumbar vertebrae rotated into the primary curve with progression, resulting in a shift of the caudad neutral vertebra to a more caudad level (the second lumbar level in ten patients, the third lumbar level in three, and the fourth lumbar level in fourteen). The stable vertebra was at the second lumbar level in eight patients, the third lumbar level in four, and the fourth lumbar level in fifteen. A compensatory lumbar curve did not develop in these patients (Figs. 4-A and 4-B), which resulted in a severe list, elevation of the ipsilateral shoulder, translation of the thoracic cage to the side of the curvature, and prominence of the contralateral hip. The more caudad the neutral vertebra in the primary curve, the more severe the deformity. The sagittal profile of the thoracic spine was hypokyphotic, and lumbar lordosis was normal.
Group 3
Group 3 included eight patients who had a long, tapering progressive thoracolumbar curve. The apex was at the twelfth thoracic level, and the neutral vertebrae were at the fifth or sixth thoracic level and at the fourth lumbar level (Fig. 1-B). The neutral vertebra at the fourth lumbar level was always tilted into the curve. The translation of the thoracic cage and the list were not as severe as in Group 2. There was hypokyphosis of the thoracic spine, and lumbar lordosis was normal.
Group 4
Group 4 consisted of twelve patients who had a primary progressive lumbar curve and a secondary minor thoracic curve. The apex was at the second lumbar level in ten patients and at the third lumbar level in two (Fig. 1-B). In the early stage, the lumbar curve extended from the eleventh thoracic to the fourth lumbar vertebra with slight, if any, deformity in the thoracic region. As the lumbar curve progressed, a compensatory thoracic curve developed. The eleventh or twelfth thoracic vertebra remained the neutral vertebra between these two curves. As the thoracic curve developed, spinal alignment was maintained and the cosmetic deformity was slight. These patients had normal thoracic kyphosis and increased lumbar lordosis.
Size and Progression of the Curve and Treatment
Patients who were seen before the age of ten years were usually observed without treatment for six to nine months. A brace was used if the curve was more than 30 degrees. A Milwaukee brace was used until 1976; after that time, a modification of the Boston spinal orthosis was employed. The best correction of the curve in the brace usually occurred within the first six months, and the greatest degree of correction was seen in children who were less than six years old. Initially, the brace was worn full-time; however, if the curve was less than 30 degrees after the brace had been worn for six months to one year, the patient was allowed to take the brace off for four hours each day for four months. If the curve progressed, the brace was again worn full-time. Operative intervention was recommended if the child was more than ten years old and had a curve of more than 40 degrees that could not be controlled in the brace.
Group 1A
In Group 1A (twenty-eight patients), the diagnosis was made at a mean age of seven years and three months (range, three years and eight months to nine years and eight months). Initially, the mean curve was 39 degrees (range, 10 to 49 degrees). Twenty-one patients were managed with a brace for a mean of four years and three months (range, nine months to six years and six months); the curve initially decreased but subsequently progressed at a mean rate of 2 degrees per year before the age of ten years. The curve progressed more rapidly (mean rate, 8 degrees per year; range, 6 to 15 degrees per year) after the age of ten years. All twenty-one patients had an arthrodesis of the spine at a mean age of eleven years and nine months (range, nine years and three months to fourteen years and eight months); at that time, the mean thoracic curve was 46 degrees (range, 28 to 82 degrees). Four patients were first seen by us at a mean age of seventeen years and six months (range, sixteen years and six months to twenty years), at which time the curves were 74, 78, 80, and 120 degrees; an arthrodesis of the spine was done in each patient. Three patients, eight years and three months to ten years old, were still being managed with a brace at the time of writing.
Group 1B
In Group 1B (twenty-nine patients), the diagnosis was made at a mean age of six years and nine months (range, three years and ten months to nine years and eleven months). Initially, the mean thoracic curve was 39 degrees (range, 17 to 68 degrees) and the mean lumbar curve was 24 degrees (range, 10 to 45 degrees). Twenty-five patients were managed with a brace for a mean of five years and six months (range, two to ten years). Before the age of ten years, there was an initial decrease of both curves, but then progression occurred at a mean rate of 1.5 degrees per year for the thoracic curves and 1 degree per year for the lumbar curves. After the age of ten years, there was much more rapid progression of the curve: a mean rate of 7.5 degrees per year for the thoracic curves and 6.5 degrees per year for the lumbar curves.
Twenty-one patients had an arthrodesis of the spine at a mean age of twelve years and six months (range, nine years and eleven months to fourteen years and six months); at that time, the mean thoracic curve was 48 degrees (range, 24 to 86 degrees) and the mean lumbar curve was 34 degrees (range, 16 to 50 degrees). Four patients were first seen by us when they were fifteen years to sixteen years and eight months old; the thoracic curves were 78, 85, 86, and 95 degrees, and the lumbar curves were 46, 52, 62, and 73 degrees. Three of these patients had an arthrodesis of the spine, and the fourth declined additional treatment. Only one patient in this group reached skeletal maturity without rapid progression of the curve during the adolescent growth spurt. Three patients, nine years and three months to eleven years old, were still wearing a brace at the time of writing.
Group 2
In Group 2 (twenty-seven patients), the diagnosis was made at a mean age of seven years and three months (range, three years and nine months to nine years and six months). The mean thoracic curve was 40 degrees (range, 19 to 56 degrees). Twenty-six children were managed with a brace for a mean of four years and three months (range, nine months to nine years and three months). After initial partial correction in the brace, the curves progressed at a mean rate of 3 degrees per year before the age of ten years and 11 degrees per year after the age of ten years. Twenty-five of the twenty-six patients had an arthrodesis of the spine, at a mean age of eleven years and four months (range, nine years and three months to thirteen years and eleven months); at that time, the mean thoracic curve was 47 degrees (range, 29 to 90 degrees). The remaining patient, who was twelve years old and had a 35-degree curve when first seen, was still wearing a brace at the time of writing. The twenty-seventh patient was seven years old at the time of the initial evaluation and had a 34-degree curve that was not treated. At the time of the most recent follow-up examination, she was thirteen years and four months old and the curve was 53 degrees. The family declined all forms of treatment, despite a marked rib hump, spinal decompensation, and lateral translation of the thoracic cage.
Group 3
In Group 3 (eight patients), the patients were first seen at a mean age of five years and eleven months (range, three years and four months to nine years). The mean thoracolumbar curve measured 28 degrees (range, 18 to 45 degrees). One patient, who was seven years old at the time of the initial examination, had a 20-degree curve that was not treated. At skeletal maturity, the curve was 31 degrees and there was a mild cosmetic deformity. Two children, who initially had curves of 18 and 23 degrees, reached skeletal maturity without needing a spinal arthrodesis. One wore a brace for five years and six months, and the curve was 17 degrees at the age of sixteen years. The other wore a brace for nine years and one month and had a curve of 33 degrees at the age of seventeen years and three months. Three patients had worn a brace for a mean of five years and were still wearing a brace at the time of writing; the curve had decreased from 37 to 17 degrees by the age of twelve years in one child, from 45 to 19 degrees by the age of eight years in another, and from 40 to 14 degrees by the age of nine years in the third. Two children had a spinal arthrodesis after progression of the curve in the brace. Both children were nine years old when they were first seen; the curve progressed from 22 to 33 degrees by the age of fourteen years in one and from 28 to 36 degrees by the age of twelve years and seven months in the other. In retrospect, both operations may have been done prematurely.
Group 4
In Group 4 (twelve patients), the diagnosis was made at a mean age of seven years and ten months (range, five years to nine years and three months). The mean thoracic curve was 22 degrees (range, 10 to 57 degrees), and the mean lumbar curve was 30 degrees (range, 18 to 58 degrees). Only one patient had a spinal arthrodesis. At the time of the initial examination, this child was nine years old, the lumbar curve was 58 degrees, and the thoracic curve was 57 degrees. The arthrodesis was done at the age of eleven years and three months, at which time the lumbar curve was 86 degrees and the thoracic curve was 80 degrees.
The remaining eleven patients were managed with a brace for a mean of seven years and eight months (range, one year and six months to ten years and two months). After initial partial correction in the brace, the mean rate of progression was 1.5 degrees per year for the lumbar curves and 2 degrees per year for the thoracic curves before the age of ten years; after this time, the curves progressed more rapidly (5 and 4.5 degrees per year, respectively). Weaning from the brace commenced at a mean age of thirteen years and five months (range, eleven years and one month to fourteen years and six months), at which time the mean lumbar curve was 36 degrees (range, 18 to 80 degrees) and the mean thoracic curve was 39 degrees (range, 26 to 86 degrees). Seven of these patients were seen at skeletal maturity, at which time the mean lumbar curve was 35 degrees (range, 28 to 42 degrees) and the mean thoracic curve was 31 degrees (range, 25 to 40 degrees). The spine was well balanced in all of these patients, and operative treatment was not necessary.
Resolving Curves
In the five patients who had a resolving curve, the diagnosis was made at a mean age of four years and two months (range, three years and six months to five years and six months). The mean curve was 23 degrees (range, 15 to 40 degrees), with slight vertebral rotation. Three of the five patients had a thoracic curve, and two had a thoracolumbar curve. The patients were managed with a brace, and the curves decreased to less than 10 degrees at a mean of four years and six months (range, two years to six years and eight months). The curves had remained stable at the time of the latest follow-up evaluation, at skeletal maturity.
Rib-Vertebra Angle Difference and Rotation of the Apical Vertebra
The mean rib-vertebra angle difference in the patients who had a progressive thoracic or thoracolumbar curve was 31 degrees (range, 12 to 45 degrees). In the patients who had a resolving curve, the mean difference was 9 degrees (range, 5 to 20 degrees) and the difference decreased with time. In Groups 1A, 1B, and 2, both measurements changed only slightly before the age of ten years; however, they appeared to increase in proportion to the progression of the curve during the adolescent growth spurt. In Group 3, the changes were smaller and more gradual, regardless of age.
At our institution, the prevalence of patients with juvenile idiopathic scoliosis has remained relatively constant (mean, 8 per cent per year) as a proportion of patients with idiopathic scoliosis (infantile, juvenile, and adolescent). In his London clinic, James found a prevalence of juvenile scoliosis of 12 per cent (sixteen of 134). In the United States, where infantile scoliosis is relatively uncommon15, the prevalence of juvenile scoliosis has been reported to be 13 to 16 per cent (forty-four [13 per cent] of 33513, twenty-six [15 per cent] of 16911, and twenty [16 per cent] of 1236). In our study, the patients were first seen at a mean age of six years and ten months, and the over-all ratio of girls to boys was 1.6:1. Others have reported a similar age at onset, with a higher proportion of girls: the ratio of girls to boys has ranged from 3.8:1 in one study9 to 4.4:1 in another study17. We identified five types of curves that could be classified only as they progressed.
Juvenile scoliosis develops before a child is ten years old and progresses slowly during the period of steady spinal growth. However, after the age of ten years, progression is rapid during the period of accelerated spinal growth, despite the use of a brace. In our study, the curve progressed in 104 (95 per cent) of the 109 patients. Seventy-seven of the eighty-nine patients who were followed until skeletal maturity had an arthrodesis of the spine to prevent additional progression.
In the early stage, the level of the most rotated vertebra at the apex of the primary curve was most closely associated with the prognosis. The level of the caudad neutral vertebra in the thoracic curves could help to predict the final deformity. It has been suggested that factors that are associated with a poor prognosis (progression of the curve) in juvenile idiopathic scoliosis include a serial increase in the rib-vertebra angle difference at the apex of the primary curve, a thoracic kyphosis of less than 20 degrees, and a left-sided curve in a boy4,9,17. We did not find any of these factors to be of value. In our patients, all of the primary thoracic curves were hypokyphotic and the rib-vertebra angle difference increased as the curve progressed, reflecting increasing vertebral rotation at the apex of the curve.
Initially, the curves in Group 1 appeared to be very similar, but two different types of deformity developed with progression of the curve. Nonetheless, the apical vertebra usually remained at the eighth thoracic level and the neutral vertebra remained at the eleventh or twelfth thoracic level. We do not know why only a single structural curve developed in some of these patients (Group 1A) while a double structural curve developed in others (Group 1B). This was not related to management with a brace (because the double-curve pattern was present before the brace was used) or to the radiographic characteristics of the curve (the severity of the vertebral rotation in the primary thoracic curve, the rib-vertebra angle difference, or the sagittal deformity). Both types of thoracic curves had a poor prognosis, and most could not be controlled adequately in a brace. All twenty-five Group-1A patients who had reached skeletal maturity and twenty-four of the twenty-six Group-1B patients who had reached skeletal maturity had operative management.
The curves in Group 2 differed from the curves in Group 1 because, with progression, additional lumbar vertebrae rotated into the primary thoracic curve. This resulted in a shift of the caudad neutral and stable vertebrae to a more caudad portion of the lumbar spine, leaving insufficient room for the development of a compensatory lumbar scoliosis. The patients in Group 2 had the most pronounced cosmetic deformity because of greater lateral translation of the thoracic cage and the tendency for a list to develop toward the convexity of the thoracic curve. The more caudal the caudad neutral vertebra, the more severe the deformity. The curves in which the fourth lumbar vertebra was the last vertebra tilted into the primary curve were similar to a type-IV adolescent curve, according to the classification of King et al.; however, curves in which either the second or the third lumbar vertebra was the last vertebra to be tilted into the primary curve were intermediate between types III and IV curves according to King et al. Operative management was necessary for all of the twenty-five patients in Group 2 who had reached skeletal maturity.
Group-3 curves progressed slowly, and there was only a mild list and slight lateral translation of the thoracic cage. The curves were very mobile and were easily controlled in a brace; three patients had a curve of less than 35 degrees when they reached skeletal maturity. A spinal arthrodesis was performed to prevent progression of the curve in only two patients, who had curves of 33 and 36 degrees; in retrospect, these operations may not have been necessary.
In Group 4, the apical vertebra was at the second or third lumbar level and the curve initially presented as primary lumbar scoliosis. However, as the lumbar curve progressed, a second structural curve developed on the opposite side in the thoracic region, producing a double-curve pattern and thereby maintaining spinal balance. This pattern of deformity, which was relatively uncommon, occurred in twelve patients and was similar to a type-I adolescent curve according to the classification of King et al. This type of curve had a relatively benign prognosis, and only one patient, who had not been managed with a brace, needed a spinal arthrodesis. The remaining patients had a well balanced spine with a relatively mild cosmetic deformity when they were seen at the most recent follow-up evaluation.
The single lumbar curves seen in adolescent idiopathic scoliosis were not found in our patients who had juvenile idiopathic scoliosis. However, the prevalence of single lumbar curves in other studies of juvenile scoliosis has been reported to be 12 per cent (seven of fifty-nine patients17 and five of forty-three patients9). It is possible that many of these were actually Group-4 curves and the relevance of the thoracic curve was not appreciated.
Five patients (5 per cent) had a curve that decreased or resolved; this compares with rates ranging from 52 per cent (sixty-eight of 132)3 to 92 per cent (ninety-two of 100)8 that have been associated with infantile idiopathic scoliosis. In our series, the resolving juvenile curves, which were seen before the children were six years old, were similar to resolving infantile curves in that they averaged 23 degrees, there was little vertebral rotation, and the rib-vertebra angle difference was less than 20 degrees10. The five curves were treated with a brace, and they decreased to less than 10 degrees at a mean of four years and six months and remained stable throughout the adolescent growth spurt.
In most of our patients, the final curve was similar to adolescent idiopathic scoliosis. A few of our patients may have had an infantile idiopathic curve that was not recognized before the child was three years old. The etiologies of juvenile and adolescent idiopathic scoliosis may be similar, but juvenile scoliosis develops before the age of ten years. There is a rapid increase in the deformity during the adolescent growth spurt, and most patients need operative management early in the teenage years. In contrast, adolescent idiopathic scoliosis appears to develop from a straight spine during the adolescent growth spurt and is therefore subjected to a shorter period of accelerated growth of the spine before it stabilizes at skeletal maturity; only a few of these patients need operative management. We believe that the difference in prognosis warrants consideration of juvenile idiopathic scoliosis as a separate and distinct entity from adolescent idiopathic scoliosis.
NOTE: The authors thank Marianne McMaster for her help in the preparation of this paper.