Although the Milwaukee brace was first designed to be used in the postoperative treatment of idiopathic scoliosis, in 1958 Blount et al. suggested its use for the non-operative treatment of idiopathic scoliosis. Since that time, the Milwaukee brace and other braces have been used in an attempt to halt the progression of curves in immature patients who have idiopathic scoliosis1,6,8,9,11,13,15-18,24,27.
The literature is replete with studies of the non-operative treatment of scoliosis, but it is not known whether bracing is effective. There are several reasons for this dilemma. First and foremost, it is not surprising that there have been no prospective randomized studies with long-term follow-up comparing patients managed with a brace with a similar group of patients managed without one. Such a study could be considered unethical, as the present standard of care for a progressive curve in immature patients necessitates some form of treatment. As far as we know, all published studies on the effectiveness of Milwaukee bracing, as well as the present one, are retrospective and rely on comparisons with previously published studies of the natural history of scoliosis4,10,12,14,19,21,23,25. In addition, many of the published reports have involved non-consecutive series of patients, often with exclusion of patients who were managed operatively, and have not included a uniform group of patients who were at high risk for progression of the curve1,6,8,9,11,13,15-18,24,27. Other problems have included small sample groups, failure to measure compliance, failure to document skeletal immaturity, and short-term follow-up.
The purpose of the present report was to study the results in a series of patients with idiopathic scoliosis who were at a high risk for progression and were managed with the Milwaukee brace. The variables that were evaluated for their effect on treatment included the sex of the patient, skeletal immaturity (as determined by the Risser sign, the presence of menarche, and the age of the patient), the pattern and magnitude of the curve, the rotation of the apical vertebra, spinal balance, the response of the curve, and the compliance of the patient.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
†Department of Orthopaedics, Orlando Regional Medical Center, 1314 Kuhl Avenue, Orlando, Florida 32806.
‡Department of Orthopaedic Surgery, The University of Iowa Hospitals and Clinics, 200 Hawkins Drive, Iowa City, Iowa 52242-1009. Please address requests for reprints to Dr. Weinstein.
Selection of the Study Group
We identified 111 patients who were at least eight years old when management with the Milwaukee brace was begun. One patient could not be located, one still used the brace at night, and seven were interviewed but follow-up radiographs were not available. These seven patients had not had any additional treatment for the scoliosis. This left 102 mature patients (92 per cent) who had been followed until the performance of an arthrodesis or had had a follow-up radiograph made. To be considered at high risk for progression, the patients had to have met one of four criteria at the time of initial bracing: a curve of more than 25 degrees and a Risser sign of 0 or 1 (sixty-six patients); a curve of 20 to 25 degrees with 5 degrees of documented progression, a Risser sign of 0, 1, or 2, and no menses (eight patients); a curve of less than 20 degrees with 10 degrees of documented progression, a Risser sign of 0, 1, or 2, and no menses (three patients); or a curve of more than 35 degrees, a Risser sign of 0 to 3, and no menses (eleven patients). To improve the statistical validity, fourteen patients were excluded on the basis of these criteria. Four of the fourteen were excluded because they had been at low risk for progression of the curve, and ten were excluded because the magnitude of the curve was considered to have been too great (more than 45 degrees) to have been considered for bracing. This left a total of eighty-eight patients for statistical analysis in this study.
Ten patients were male. The remaining seventy-eight were female: forty-three were premenarchal and thirty-five were postmenarchal at the time of initial bracing.
Treatment
The patients wore a custom-fitted Milwaukee brace full-time (twenty-three hours a day) and were taught daily postural and paraspinal strengthening exercises. The braces were all manufactured by the same company (American Prosthetics, Iowa City, Iowa), and a certified orthotist was always present at the clinic. Typically, the patients returned every four to six months to have a posteroanterior radiograph made while standing and wearing the brace, until weaning from the brace was begun on the basis of skeletal maturity. Patients were intentionally weaned by decreasing wear by four hours a day after each visit to the clinic. Weaning was continued if there was no noticeable progression on serial follow-up radiographs made with the patient out of the brace. The brace was then worn at night only, until there was no noticeable progression or until the patient outgrew the brace. The patients were advised to have spinal correction and arthrodesis if the curves progressed despite bracing or progressed to more than 50 degrees. The patients who did not have an operation were instructed to return to have comparative radiographs made at one, two, three, and five years after use of the brace was stopped and every seven years thereafter.
Assessment of Compliance
To evaluate the effect of compliance with bracing on the outcome, the patients were classified as either compliant or non-compliant, as determined by review of the chart and a follow-up questionnaire. A patient was considered to have been compliant if he or she had returned for most of the scheduled appointments, had no mention of non-compliance on the chart, and had stated on the questionnaire that he or she had followed all of the treatment guidelines. If any one of these criteria had not been met, the patient was considered to have been non-compliant. On the basis of these criteria, sixty-four patients were compliant and twenty-four were non-compliant. Non-compliance varied from discontinuation of use of the brace prematurely, less than full-time use of the brace, or any deviation from the weaning protocol.
Radiographic Analysis
All of the available radiographs for each patient were reviewed by two of us (K. J. N. and S. L. W.) to determine the pattern, apex, and end points of the curve. The curves were classified as thoracic, lumbar, thoracolumbar, double thoracic, or thoracic and lumbar double major. A curve was judged to be structural if there was substantial rotation at its apex and to be compensatory if there was very little apical rotation. A thoracic curve pattern indicated a structural curve in the thoracic spine with compensatory thoracic and lumbar curves cephalad and caudad. A lumbar curve pattern included a structural curve in the lumbar spine with an apex at the first or second lumbar vertebra and a compensatory thoracic curve cephalad. A thoracolumbar curve pattern consisted of a structural curve with an apex at the eleventh or twelfth thoracic vertebra and compensatory thoracic and lumbar curves. A double thoracic curve pattern included cephalad and caudad thoracic structural curves with a compensatory lumbar curve. The cephalad thoracic curve was determined to be structural if it had major apical rotation and the first rib was depressed on the concave side of the curve (Fig. 1). Finally, a thoracic and lumbar double major curve pattern consisted of structural curves in the thoracic and lumbar regions with a compensatory thoracic curve cephalad. The structural curves had major apical rotation and crossed the midline in all patients (Fig. 2).
Structural curves were additionally grouped into thoracic or lumbar types to assess the prevalence of failure and the effect of prognostic factors. All of the structural thoracic curves in the thoracic and lumbar double major curves, the thoracic curves, and the caudad thoracic curves in the double thoracic curves comprised the thoracic group. All of the lumbar curves in the thoracic and lumbar double major curves, the lumbar curves, and the thoracolumbar curves comprised the lumbar group.
All of the radiographic measurements were made by one of us (K. J. N.) with use of the same ruler. The 120 structural and 139 compensatory curves were measured in the coronal plane according to the method of Cobb, and the rotation of the apical vertebra was measured according to the method of Perdriolle and Vidal. Spinal balance (the distance from the apexes of the curve as well as from the seventh cervical vertebra to a line drawn perpendicular to the center of the sacrum) was measured and recorded in the coronal plane (Fig. 3).
The extent of ossification of the iliac apophysis (the Risser sign) was determined by two of us (K. J. N. and S. L. W.) from the radiographs made at the time of initial bracing and at weaning. The Risser sign at the time of initial bracing was determined from the radiographs of seventy-three (83 per cent) of the patients; for nine of the fifteen patients whose initial radiographs were missing, the Risser sign was ascertained from the chart. The Risser sign was 0 for fifty-one patients, 1 for sixteen, 2 for eight, and 3 for seven. Four of the six patients who did not have initial radiographs available or a Risser sign noted on their chart were premenarchal. Two of the sixty-four patients who were intentionally weaned from the brace had a Risser sign of 2 at the time of weaning; two, a Risser sign of 3; fifty-six, a Risser sign of 4; and one, a Risser sign of 5. Three of the patients whose previous radiographs had been discarded had no Risser sign recorded at the time of weaning.
Definition of Terms
All of the Cobb angles were recorded in relation to the ages of the patient during the course of the treatment (Table I). These included the ages at presentation, defined as the time when the scoliosis was diagnosed radiographically; at initial bracing, defined as the time when the brace was first applied; at the time of the best correction, defined as the maximum correction of the Cobb angle of the structural curve in the brace; at weaning, for the patients for whom weaning was begun intentionally; at the time when the patient stopped using the brace completely; and at follow-up, defined as the time when the latest radiograph was made for patients who were managed non-operatively, or at operation. The age at which use of the brace was stopped was also determined separately for the patients who had an arthrodesis and for those who did not. If a patient had stopped wearing the brace prematurely, the age at which the last radiograph was made with the patient in the brace was used. If a patient had continued wearing the brace until the date of the operation, the same age was recorded both for the time at which use of the brace was stopped and for the operation; however, most patients stopped wearing the brace an average of six months before the operation was performed.
Apical rotation was recorded for only the structural curves at initial bracing and at the time of follow-up or the operation. Spinal balance was also recorded, from the apexes of all structural curves as well as from the seventh cervical vertebra to the central sacral line, at initial bracing and at the time of follow-up or the operation.
No previous radiographs were available for thirteen of the patients who returned for follow-up, and five additional patients had one radiograph missing from their file. For these patients, the values for the Cobb angles were taken from the chart, and follow-up measurements were made from the same end points. Because spinal balance and apical rotation had not been uniformly recorded on the charts, this information was not available at initial bracing for the fifteen patients whose initial radiograph had been discarded.
Statistical Analysis
To compare the over-all results of use of the Milwaukee brace and the relative effects of various prognostic factors, we defined three types of failure. Type-1 failure indicated that the curve had increased 5 degrees or more from initial bracing to the time that the patient stopped using the brace; type-2, the patient had an operation or had a structural curve of more than 50 degrees at the time of follow-up; and type-3, the patient did not have an operation and the major curve had progressed 10 degrees or more from initial bracing to the time of follow-up. The result in the patients who had a double structural curve was classified as a failure if one or both of the structural curves met the criteria just mentioned.
On the basis of these criteria for failure, we were able to assess the effect of several different parameters and variables on the result. Test variables included the sex of the patient, the Risser sign at initial bracing and at weaning, the presence of menarche at initial bracing, the age of the patient at initial bracing, the curve pattern, the compliance of the patient, the Cobb angle at initial bracing, the percentage correction of the structural curve during bracing, apical rotation, and spinal balance as measured from the seventh cervical vertebra as well as from the apexes of the structural curve. We excluded information regarding the cephalad thoracic curve in the double thoracic curves, as the Cobb angle and the apical rotation20 for this curve were always smaller than those for the caudad thoracic curve, which is the structural curve that dictates treatment.
The data were analyzed with three different statistical tests. The two-tailed Fisher exact test was used to assess the statistical prevalence of failure with regard to the sex of the patient, the Risser sign at initial bracing and at weaning, the curve pattern, the compliance of the patient, and the presence of menarche. This test was also used to assess the prevalence of failure in subgroups according to the age at presentation and according to the Cobb angle at initial bracing. The t test was used to compare the prevalence of failure and non-failure with regard to the apical rotation, the percentage of correction, the spinal balance as measured from the structural apexes, the average age at presentation, and the average magnitude of the Cobb angle at initial bracing. The Wilcoxon rank-sum test was used to assess the significance of failure when spinal balance measured from the seventh cervical vertebra was examined. Because our patients were similar with regard to risk for progression (according to the age of the patient, the Risser sign, and the magnitude of the curve at initial bracing), a p value of 0.01 or less (a type-I error) for each test was considered significant. A rigorous alpha level distinguishes differences that are both significant and possibly clinically applicable.
Two possible limitations must be kept in mind when these data are interpreted. First, because of the large number of comparisons made (117), the prevalence of a random result (type-I error) increases. In our study, the likelihood of finding a significant difference in at least one subgroup of comparisons when one does not exist is 69 per cent (1 - [1 - 0.01]117). In addition, because of the limited sizes of the study groups and subgroups, the statistical power lessens and the probability of a type-II error (a non-significant result when significance actually exists) increases. Studies such as this are limited by these statistical facts as attempts are made to discern importance from statistical results that are often difficult to apply to the clinical situation.
Sixty-one (69 per cent) of the eighty-eight patients who were included in the statistical analysis did not have an arthrodesis; of these, fifty-five (90 per cent) were intentionally weaned from the brace. Six non-compliant patients stopped wearing the brace prematurely, before skeletal maturity. The average duration of bracing for the sixty-one patients was one year and eight months, and they were followed for an average of six years and four months after they stopped using the brace. Twenty-seven patients (31 per cent) had an arthrodesis. Eighteen of these patients had progression of the curve while they wore the brace, and nine had progression of the curve after a trial of intentional weaning.
The average Cobb angle at initial bracing was 34 degrees. When use of the brace was stopped, the average Cobb angle was 45 degrees for the patients who eventually had an arthrodesis and 35 degrees for those who did not; however, the curves progressed an average of 5 degrees by the time of follow-up (Table II). These data were combined with the information regarding the age of the patient, the presence of menarche, and the Risser sign at initial bracing and at weaning and were arranged graphically for all structural curves, the structural lumbar curves, and the structural thoracic curves (Figs. 4, 5, and 6). The same patterns of initial curve response and progression until use of the brace was stopped were noted. Curves that were not treated with an arthrodesis continued to progress from the time that the patient stopped using the brace to the time of follow-up.
Statistical Analysis
For all patients, the prevalence of type-1, type-2, and type-3 failures was 48, 42, and 33 per cent, respectively. The prevalence of patients having at least one of these types of failure was 63 per cent.
Type-1 failure: The significant variable for all of the patients who had type-1 failure was an average correction of the Cobb angle of 8 per cent during bracing (p = 0.002). The patients who did not have type-1 failure had an average correction of 20 per cent.
Type-2 failure: The significant variables for all of the patients who had type-2 failure were an average age of eleven years and nine months at presentation (p = 0.002), an average Cobb angle of 37 degrees at initial bracing (p = 0.009), and an average correction of the Cobb angle of 8 per cent during bracing (p = 0.002). The patients who did not have type-2 failure had an average age of twelve years and nine months with an initial average curve of 34 degrees and an average correction of 20 per cent.
Type-3 failure: The average percentage correction of the Cobb angle during bracing was also significant with regard to the prevalence of type-3 failure in all of the patients. The patients who had type-3 failure had an average correction of 10 per cent, while the patients who did not have type-3 failure had an average correction of 22 per cent (p = 0.007).
The sex of the patient, the Risser sign at initial bracing and at weaning, the presence of menarche at bracing, the compliance of the patient, the apical rotation, the curve pattern, and the spinal balance had no significant effect on the prevalence of failure type in any patient.
Failure of bracing of thoracic structural curves: When just thoracic structural curves were considered, the significant variable for the patients who had type-1 failure was an average correction of the Cobb angle of 7 per cent during bracing (p = 0.002). The patients who did not have type-1 failure had an average correction of 19 per cent. The significant variables for the patients who had type-2 failure were an average age of eleven years and eight months at presentation (p = 0.005) and an average correction of the Cobb angle of 5 per cent during bracing (p = 0.001). The patients who did not have type-2 failure were an average of twelve years and eight months old at presentation and had an average of 21 per cent correction of the curve. There were no significant parameters that predicted type-3 failure. The sex and compliance of the patient, the Risser sign at initial bracing or at weaning, the presence of menarche at initial bracing, the Cobb angle at initial bracing, the apical rotation, the curve pattern, and the spinal balance had no significant effect on the prevalence of failure type in the subgroup of thoracic curves.
Failure of bracing of lumbar structural curves: When just lumbar structural curves were considered, fewer variables were found to be significant for predicting failure. The patients who had type-1 or type-3 failure had an average correction of 5 per cent (p = 0.004) and 9 per cent (p = 0.002), respectively, while the patients who did not have type-1 or type-3 failure had an average correction of 27 per cent and 23 per cent, respectively. Also, the patients who had type-2 failure were an average of eleven years and five months old (p = 0.003), while the patients who did not have type-2 failure were an average of thirteen years and two months old. The Cobb angle at presentation, the presence of menarche, the Risser sign at initial bracing and at weaning, the compliance and sex of the patient, the pattern of the curve, the apical rotation, and the spinal balance had no significant effect on the prevalence of failure type in the subgroup of lumbar curves.
To allow comparisons of our results with those in the literature, we further analyzed the effect of the Cobb angle at initial bracing and the age at presentation. When the Cobb angles at bracing were divided into those that were less than 30 degrees, those that were 30 to 39 degrees, and those that were 40 degrees or more, no significant effect was found on the prevalence of all failure types for all patients, for thoracic curves, or for lumbar curves. Although the finding was not significant, in the total group of patients (p = 0.06) and in that of thoracic curves (p = 0.09), an angle of 40 degrees or more at initial bracing was associated with a higher prevalence of type-2 failure than was an angle of less than 40 degrees. When the patients were divided into those whose age at presentation was less than twelve years and those whose age was more than twelve years, no significant effect on the prevalence of type-1 failure was found. However, an age of less than twelve years was significantly associated with type-2 failure in the group of all patients (p = 0.009), that of thoracic curves (p = 0.008), and that of lumbar curves (p = 0.007). Although the finding was not significant, this younger age was also associated with a higher rate of type-3 failure in the group of lumbar curves (p = 0.05).
Natural History
To assess the effect of bracing on idiopathic scoliosis in immature patients, it is essential to compare the results with the natural history of untreated curves in immature patients. Two prospective studies3,23 on the prevalence of scoliosis as determined by school-screening were reported in the mid-1970's. Brooks et al. reported structural curves of more than 5 degrees in 14 per cent (474) of 3492 children. On following 134 of these children, they noted an average spontaneous improvement of almost 8 degrees in 22 per cent (thirty children) and an average progression of 7 degrees in 5 per cent (seven children). In a later study, Rogala et al. noted curves of more than 5 degrees in 4.5 per cent (1222) of 26,947 school children who were screened; 2 per cent (584) had curves of more than 10 degrees. In a subset of 603 of these children who were followed for two years, Rogala et al. found that 2 per cent of the 252 initial curves that were 6 to 10 degrees and 10.3 per cent of the 351 initial curves that were more than 10 degrees had progressed. A subset of fifty-two students in whom the curve measured 20 to 30 degrees at presentation were also followed. Of these, 79 per cent (forty-one) had progression of at least 5 degrees. Fustier followed eighty-one prepubertal children who had had untreated idiopathic scoliosis for at least three years. He noted progression of 5 degrees or more in 56 per cent of the seventy patients who had had an initial curve of 10 to 29 degrees, in 75 per cent of the twenty-four who had had an initial curve of 20 to 29 degrees, and in nine of the eleven who had had an initial curve of 30 to 45 degrees.
In 1982, Nachemson et al. reported on the risk of progression of 5 degrees or more in patients who have untreated scoliosis. For girls who were ten to twelve years old, the prevalence of progression was 60 per cent when the curve had been 20 to 29 degrees at presentation and 90 per cent when it had been 30 to 59 degrees. For girls who were thirteen to fifteen years old, the prevalence of progression was 40 and 70 per cent, respectively. The factors that were found to have prognostic significance included the remaining growth potential, the curve pattern, the sex of the patient, and the magnitude of the curve.
Lonstein and Carlson14 reported progression of the curve in 23.2 per cent (169) of 727 children with untreated scoliosis who were followed to the end of skeletal growth or until progression of the curve. Factors found to be associated with an increased risk of progression included curves of greater magnitude, skeletal immaturity, and double curve patterns. Rotational prominence, a family history, the Harrington factor, and the sex of the patient had no significant effect on progression.
In 1986, Bunnell4 reported on 123 patients in whom an initial curve of less than 50 degrees had not been treated. Sixty-eight per cent of these patients had at least 5 degrees of progression of the curve, 34 per cent had at least 10 degrees, and 18 per cent had at least 20 degrees.
Studies of the natural history of scoliosis in male patients have been somewhat limited because of small series, short-term follow-up, or inclusion of patients who had had treatment. However, it is clear that progression in boys is associated with curves of greater magnitude and a lower Risser sign12,25. In contradistinction to girls, boys who did not have treatment were noted12,25 to have continued progression of the curve until the Risser sign was 5. Suh and MacEwen noted such progression in twenty-six (44 per cent) of fifty-nine structural curves. Karol et al. reported progression of 10 degrees in five (15 per cent) of thirty-four male patients who had a Risser sign of 4.
In a recent review of the natural history of idiopathic scoliosis, prognostic factors were summarized to include menarche, the Risser sign, the chronological age of the patient, the severity of the curve, and the type of curve5.
Curves of greater magnitude have also been reported to progress after skeletal maturity in untreated patients; curves of less than 30 degrees tended not to progress, while curves of more than 50 degrees progressed an average of 1 degree a year26.
Results of Treatment
In 1970, Moe and Kettleson presented what we believe was the first study on the treatment of idiopathic scoliosis with the Milwaukee brace. Their study was restricted to eighty ideal patients who had been followed for an average of eighteen months; the average correction of high thoracic curves, thoracic curves, and lumbar curves was 10, 23, and 18 per cent, respectively, at the time of follow-up. Additional follow-up data on these patients was presented ten years later6. In seventy-four patients, the average correction of thoracic curves and lumbar and thoracolumbar curves was 2 and 4 degrees, respectively; however, more than 80 per cent of the seventy-four patients who had been followed for at least five years had continued to have progression after they had stopped using the brace. More than one-third of the patients who had had a curve of more than 40 degrees needed an operation. Limited conclusions can be drawn from these studies as it is difficult to determine whether all patients were at high risk for progression, and the exclusion of patients whose compliance was poor or who had an operation may bias results.
In 1976, Keiser and Shufflebarger reported eleven-month follow-up data on 123 of 300 consecutive patients in whom scoliosis had been treated with a Milwaukee brace. The average correction was 7 degrees, and better results were noted for less severe curves, flexible curves, thoracic curves, and younger patients who were managed for a longer duration. This study also included patients who were at questionable risk for progression and excluded from analysis patients who had been managed operatively.
Edmonson and Morris reported the twenty-two-month follow-up results for fifty-two of 125 patients, in 1977. The average correction of the lumbar and thoracic curves was 6 degrees, and the authors concluded that, with adequate treatment, few curves progressed. The study included only patients who had been managed non-operatively, many of whom were at questionable risk for progression.
Mellencamp et al.17 reported the results for forty-seven of ninety-four patients who had been managed non-operatively and had been followed for at least five years. The average correction when the patient stopped using the brace was 10 degrees; however, curves were noted to have progressed an average additional 6 degrees at the time of the latest follow-up. Additional study of these patients after more than eight years of follow-up led to the same general conclusions; five of the patients had had an arthrodesis16. Both reports presented very little information on skeletal maturity and, therefore, on the risk of progression.
In 1985, Cochran and Nachemson reported the results after a minimum of five years of follow-up of eighty-five (89 per cent) of ninety-five patients who had been managed with a Milwaukee brace. They excluded forty-three other patients who either had been non-compliant or had had an operation. In the eighty-five patients, the average magnitude of the curve at bracing was 30 degrees, with 4 degrees of progression at an average of 7.3 years. The authors believed that their study, and all previous studies, suggested that the Milwaukee brace may correct or maintain the magnitude of the curve in some patients but additional progression is often noted. They concluded that the true effectiveness of the Milwaukee brace had not been scientifically established.
In 1991, Styblo presented a well documented thesis on 290 patients who had been managed with a Milwaukee or Boston brace. A subgroup of 128 patients who had contemporary indications for bracing was followed for 3.5 years, at which time the average initial 34-degree curve had progressed an average of 1 degree. The curves in 72 per cent of the patients were unchanged or improved at the time of follow-up, while those in 28 per cent had progressed 5 degrees or more. The patients who had progression or an operation tended to have been younger and skeletally immature, to have had a larger curve at the time of bracing, and to have had less correction of the Cobb angle. Partially compliant patients had twice the rate of progression of those who were completely compliant. Through comparison with the natural history of scoliotic curves, the author found that bracing had a positive effect.
The results for 1020 patients who had been managed with the Milwaukee brace were reported in 1994 by Lonstein and Winter15. Some of the patients had been included in previous studies6,18. The findings in a subgroup of 524 patients were compared with the natural history of scoliotic curves15. This subgroup had bracing for contemporary indications. The authors documented improvement of the Cobb angles from the time of initial bracing to that of the best angle in the brace as well as some loss of correction at the time that treatment with the brace was stopped. On the basis of comparison with published and unpublished studies on the natural history of scoliotic curves, Lonstein and Winter concluded that the brace was significantly effective for curves of 20 to 29 degrees, as seen at the end of treatment with the brace. Continued progression of the 20 to 29-degree curves was found at the time of the most recent follow-up. Thirty to 39-degree curves were not significantly different from unpublished natural-history values at the time when the patient stopped using the brace and the time of the most recent follow-up.
Results in the Present Study
There are limitations in the design of any retrospective study that affect any conclusions that are drawn. In the present study, fifteen initial radiographs were not available and the values for the Cobb angle had to be determined from the chart. Although we chose to include the patients whose radiographs were unavailable because none had had an arthrodesis, their exclusion would have increased the percentage of failures. The issue of lost records also affects conclusions regarding the impact of apical rotation and spinal balance. Secondly, as no good method for measurement of compliance was available during the period of this study, this was necessarily assessed on the basis of review of the charts and the questionnaires administered to the patients. We hoped to minimize this limitation by considering compliance an all-or-nothing variable without gradations. In addition, all eighty-eight of the patients in the study returned the questionnaire, which reduced the chance of observer bias.
Many studies1,6,8,9,11,13,15-18,27 have included patients who were at questionable risk for progression, who were fully compliant, or who currently would be considered candidates for an operation. We analyzed the results of bracing in a series of patients with scoliosis who were believed to be at high risk for progression. As curves of more than 45 degrees were excluded, the study was limited to a group of patients who would be considered candidates for bracing at the present time. The rate (102 of 111 patients; 92 per cent) and duration of follow-up (average, six years and four months) allowed us to study and compare longitudinally the effect of bracing on the natural history of scoliosis in these immature patients.
The average progression of the Cobb angle from initial bracing to the time when the patients stopped using the brace was 4 degrees. This is greater than the average in previously reported studies (range, -7 to +3 degrees)1,6,8,9,13,15-18,24,27. In the present series, the curves that were not treated with an arthrodesis were an average of 6 degrees greater at the time of follow-up than they were at initial bracing. The percentage correction of the curve at the time of the best correction in the brace appeared to have a significant effect on the prevalence of all types of failure in all patients. Patients in whom the best correction averaged 20 to 22 per cent did not have a failure, while those in whom it averaged 8 to 10 per cent did. Factors that were associated with an increased risk of arthrodesis included younger age (average, eleven years and nine months compared with twelve years and nine months for the patients who did not have an arthrodesis) and curves of larger magnitude. Factors that did not have a significant effect on the prevalence of failure included the sex of the patient, the Risser sign at initial bracing and at weaning, the presence of menarche at initial bracing, the curve pattern, the compliance of the patient, the apical rotation, and the spinal balance.
Although the curve pattern did not have a significant effect on progression, the thoracic curves that were not operated on had progressed an average of 7 degrees from initial bracing to the time of follow-up. It is important to note that thoracic curves that were not operated on had increased an average of only 1 degree from initial bracing to the time when bracing was stopped; however, they progressed an average of 5 degrees after bracing was stopped. This progression can be interpreted either as progression of 1 degree a year as is seen in mature patients with larger curves26 or as a tendency for curves to progress to their natural magnitude. The lumbar curves that were not treated with an arthrodesis progressed an average of 5 degrees from initial bracing to the time of follow-up. The lumbar curves did not seem to progress as much (average, 2 degrees) after bracing was stopped.
As we are not aware of any randomized prospective studies of bracing with long-term follow-up, it is important to compare our results with the existing data on the natural history of idiopathic scoliosis in immature patients. However, a strict comparison is not possible, as studies of the natural history have involved different populations of patients and different durations of follow-up3-5,10,12,14,19,21,23,25. In other words, it is not possible to determine if patients had a worse result because they were followed longer and had progression after maturity or because the natural history was masked during bracing, with accelerated progression after bracing was stopped.
On comparison of our results with those of studies of natural history, it is not clear whether bracing has any value or not. The reported percentage of untreated patients who had more than 5 degrees of progression of the curve (type-1 failure) has ranged from 43 to 82 per cent4,10,11,14. In the present study, the prevalence of such failure for all patients was 48 per cent (forty-two patients).
More importantly, we found that 42 per cent (thirty-seven) of the patients who were managed with a brace either had a subsequent operation or would meet the current criteria for an operation. Thirty-three per cent (twenty-nine) of the patients who were not managed operatively had progression of the structural curves of at least 10 degrees from initial bracing to the time of follow-up. When we combined the prevalences of at least one type of failure, we found that 63 per cent of the patients who were managed with a brace either had an operation; would be a candidate for an operation at the present time; had progression of at least 5 degrees at the time that bracing was stopped; or, for patients who were not managed operatively, had progression of 10 degrees or more at the time of follow-up.
On comparison of our results with composite natural-history data19, we found additional evidence that bracing may not alter the natural history of scoliosis. In the present study, the prevalence of more than 5 degrees of progression for patients who were less than twelve years old was 58 to 75 per cent; for those who were more than twelve years old, it was 40 to 42 per cent. In the report by Nachemson et al., the prevalence of progression from the age of ten to twelve years and from the age of thirteen to fifteen years ranged from 60 to 90 per cent and from 40 to 70 per cent, respectively.
In a study of the natural history of idiopathic scoliosis before maturity in 123 girls, Bunnell4 reported a 52 per cent prevalence of progression of more than 5 degrees in patients who had had an initial curve of 20 to 30 degrees and a 67 per cent prevalence of such progression in patients who had had an initial curve of 30 to 40 degrees. This is similar to the 48 per cent rate of type-1 failure in our study. The prevalence of progression of more than 10 degrees, as reported by Bunnell, was 30 per cent in the patients in whom the curve had been 20 to 30 degrees and 48 per cent in those in whom it had been 30 to 40 degrees. Even with exclusion of the patients who were operated on, this is similar to the 33 per cent rate of type-3 failure in our study.
It is currently impossible to state that bracing effectively alters the natural history of scoliosis in immature patients who are at high risk for progression. Although bracing can maintain some curves at the prebracing level, progression after cessation of bracing has been documented in many patients. In other patients, bracing has had no effect on progression of the curve and an arthrodesis was indicated. A full assessment of the efficacy of bracing requires a well controlled prospective study in which patients at similar risk for progression are compared and the duration of follow-up is at least five to ten years after the cessation of bracing.
NOTE: The authors thank Jeff Ralston, M.D., for his assistance with the review of the charts.
Bassett G.S.; Bunnell, WP; and |and |MacEwen, G. D.: Treatment of idiopathic scoliosis with the Wilmington brace. Results in patients with a twenty to thirty-nine-degree curve. J. Bone and Joint Surg,68-A: 602-605, April 1986.68-A602
1986
Blount, W. P.; Schmidt, A. C.; Keever, E. D.; and |and |Leonard, E. T.: The Milwaukee brace in the operative treatment of scoliosis. J. Bone and Joint Surg,40-A: 511-525, June 1958.40-A511
1958
Brooks, H. L.; Azen, S. P.; Gerberg, E.; Brooks, R.; and |and |Chan, L.: Scoliosis: a prospective epidemiological study. J. Bone and Joint Surg,57-A: 968-972, Oct. 1975.57-A968
1975
Bunnell, W. P.: The natural history of idiopathic scoliosis before skeletal maturity. Spine,11: 773-776, 1986.11773
1986
[PubMed][CrossRef]
Bunnell, W. P.: The natural history of idiopathic scoliosis. Clin. Orthop,229: 20-25, 1988.22920
1988
[PubMed]
Carr, W. A; Moe, J. H.; Winter, R. B.; and |and |Lonstein, J. E.: Treatment of idiopathic scoliosis in the Milwaukee brace. Long-term results. J. Bone and Joint Surg,599-612: June 1980.599-612
1980
Cobb, J. R.: Outline for the study of scoliosis. In Instructional Course Lectures, The American Academy of Orthopaedic Surgeons. Vol. 5, pp. 261-275. Ann Arbor, J. W. Edwards, 1948.
Cochran, T., and |and |Nachemson, A.: Long-term anatomic and functional changes in patients with adolescent idiopathic scoliosis treated with the Milwaukee brace. Spine,10: 127-133, 1985.10127
1985
[PubMed][CrossRef]
Edmonson, A. S., and |and |Morris, J. T.: Follow-up study of Milwaukee brace treatment in patients with idiopathic scoliosis. Clin. Orthop,126: 58-61, 1977.12658
1977
[PubMed]
Fustier, T.: Évolution radiologique spontanée des scolioses idiopathiques de moins de 45 degrees en période de croissance. Étude graphique rétrospective de cent dossiers du Centre de réadaptation fonctionnelle des Massues. Thesis, Université Claude-Bernard, Lyon, France, 1980.
Gardner, A. D.; Burwell, R. G.; Wozniak, A. P.; MacPherson, I. S.; Denn, P. G.; Pursell, L. M.; and |and |Pursell, A. G.: Some beneficial effects of bracing and a search for prognostic indicators in idiopathic scoliosis. Spine,11: 779, 1986.11779
1986
[CrossRef]
Karol, L. A.; Johnston, C. E., II; Browne, R. H.; and |and |Madison, M.: Progression of the curve in boys who have idiopathic scoliosis. J. Bone and Joint Surg,75-A: 1804-1810, Dec. 1993.75-A1804
1993
Keiser, R. P., and |and |Shufflebarger, H. L.: The Milwaukee brace in idiopathic scoliosis. Evaluation of 123 completed cases. Clin. Orthop,118: 19-24, 1976.11819
1976
[PubMed]
Lonstein, J. E., and |and |Carlson, J. M.: The prediction of curve progression in untreated idiopathic scoliosis during growth. J. Bone and Joint Surg,66-A: 1061-1071, Sept. 1984.66-A1061
1984
Lonstein, J. E., and |and |Winter, R. B.: The Milwaukee brace for the treatment of adolescent idiopathic scoliosis. A review of one thousand and twenty patients. J. Bone and Joint Surg,76-A: 1207-1221, Aug. 1994.76-A1207
1994
Mellencamp, D. D., and |and |Blount, W. P.: The natural history of idiopathic scoliosis. Late results revisited. Spine,11: 805-806, 1986.11805
1986
[PubMed][CrossRef]
Mellencamp, D. D.; Blount, W. P.; and |and |Anderson, A. J.: Milwaukee brace treatment of idiopathic scoliosis. Late results. Clin. Orthop,126: 47-57, 1977.12647
1977
[PubMed]
Moe, J. H., and |and |Kettleson, D. N.: Idiopathic scoliosis. Analysis of curve patterns and the preliminary results of Milwaukee-brace treatment in one hundred sixty-nine patients. J. Bone and Joint Surg,52-A: 1509-1533, Dec. 1970.52-A1509
1970
Nachemson, A.; Lonstein, J.; and Weinstein, S.: Report of the Prevalence and Natural History Committee of the Scoliosis Research Society. Read at the Annual Meeting of the Scoliosis Research Society, Denver, Colorado, Sept. 22, 1982.
Perdriolle, R., and |and |Vidal, J.: Étude de la courbure scoliotique. Importance de l'extension et de la rotation vertébrale. Rev. chir. orthop,67: 25-34, 1981.6725
1981
[PubMed]
Picault, C.; deMauroy, J. C.; Mouilleseaux, B.; and |and |Diana, G.: Natural history of idiopathic scoliosis in girls and boys. Spine,11: 777-778, 1986.11777
1986
[PubMed][CrossRef]
Risser, J. C.: The iliac apophysis: an invaluable sign in the management of scoliosis. Clin. Orthop,11: 111-119, 1958.11111
1958
[PubMed]
Rogala, E. J.; Drummond, D. S.; and |and |Gurr, J.: Scoliosis: incidence and natural history. A prospective epidemiological study. J. Bone and Joint Surg,60-A: 173-176, March 1978.60-A173
1978
Styblo, K.: Conservative treatment of juvenile and adolescent idiopathic scoliosis. A clinical, roentgenological and comparative retrospective study on the effects of conservative treatment by brace in patients. Thesis, Aun De Rijksuniversiteit Te Leiden, Leiden, The Netherlands, 1991.
Suh, P. B., and |and |MacEwen, G. D.: Idiopathic scoliosis in males. A natural history study. Spine,13: 1091-1095, 1988.131091
1988
[PubMed][CrossRef]
Weinstein, S. L., and |and |Ponseti, I. V.: Curve progression in idiopathic scoliosis. J. Bone and Joint Surg,65-A: 447-455, April 1983.65-A447
1983
Winter, R. B.; Lonstein, J. E.; Drogt, J.; and |and |Noren, C. A.: The effectiveness of bracing in the nonoperative treatment of idiopathic scoliosis. Spine,11: 790-791, 1986.11790
1986
[PubMed][CrossRef]