Factors That Influence Outcome in Bracing Large Curves in Patients With Adolescent Idiopathic Scoliosis

As previously published in Spine Magazine

Study Design. A retrospective review of 51 patients with adolescent idiopathic scoliosis (AIS) treated with a Boston brace for curves ranging from 36° to 45°.

Objectives. To determine what radiographic or clinical observations may be predictive of outcome.

Summary of Background Data. Patients with AIS who are braced for curves .35° are less likely to respond to conservative treatment than patients of similar maturity with smaller curves.

Methods. Skeletally immature patients with AIS with no history of prior treatment were treated with a Boston brace. Cobb angles, vertebral tilt angles, coronal decompensation, apical vertebral translation(s), apical vertebral rotation, lateral trunk shift, rib vertebral angle difference, pelvic tilt, and the lumbar pelvic relationship (LPR) were measured at brace prescription, initial in-brace, brace discontinuation, and follow-up.

Results. At the time of brace discontinuation, 31 patients (61%) were judged treatment successes. With follow-up observation, an additional eight patients progressed beyond 5°, and a total of 16 patients (31%) required surgical correction. Only patients with double curves were found to have radiographic values predictive of outcome. The LPR angle, the association between the thoracic curve vertebral tilt angles and the amount of in-brace correction of the Cobb angle, were significant predictors. A patient's reported wear schedule significantly influenced outcome.

Conclusions. Patients with a double curve pattern in which the thoracic curve is .35° and the LPR angle is .12° are significantly more likely to demonstrate curve progression. In-brace correction for double curves of at least 25% and a patient's ability to wear the orthosis .18 hours/day significantly increased the likelihood of success. [Key words: scoliosis, brace, lumbar pelvic rela-tionship, wear schedule] Spine 2001;26:2354–2361

Orthoses have been shown to be capable of altering the natural history of curve progression in adolescent idiopathic scoliosis (AIS).^{13,16,21,23,25} Orthoses are typically prescribed for skeletally immature patients (Risser sign 0, 1, or 2) with curves ranging from 25° to 45° in an effort to prevent further curve progression.^{6,8,11,13,16,20,25} There is a positive correlation between the magnitude of a scoliotic curve at time of detection, as measured by the Cobb angle, and the likelihood for curve progression.⁴ Similarly, the likelihood that an orthosis will prevent curve progression decreases as the size of the curve increases.^4,8,13,16 Despite recognizing the evidence that a brace is capable of preventing progression of curves .35°,^1,8,13,16,18 a successful outcome in these patients appears to be less predictable compared with those with smaller curves. We sought to determine what, if any, radiographic or clinical parameters may be significantly prognostic for outcome in bracing curves ranging from 36° to 45° in skeletally immature patients with AIS.

To be included in this retrospective study, patients had to have a diagnosis of idiopathic scoliosis, be at least 10 years of age, have a Risser sign of 0, 1, or 2, and have a curve between 36° and 45°. All patients were treated with a Boston brace from the same institution without any prior treatment. Between October 1979 and April 1991, 51 patients (47 female and 4 male) ful-filled the study's inclusion criteria.

The clinical data compiled included the patient's gender, age, menarcheal status, heights and weights over time, and the physician's perception of a brace wear schedule. The timing of menarche in girls was determined, as was the timing of each patient's primary adolescent growth spurt. The timing of this growth spurt, designated as the time in which a patient is demonstrating his or her peak height velocity (PHV), was calculated by comparing the height measured for each visit to the first height measurement from at least 6 months (183 days) before that visit. A height velocity of 9 cm/year has been shown to be 2 standard deviations (SDs) from mean values 6 months before and after the growth peak.³ Thus, the visit date in which a patient's height velocity was calculated to be $9 cm/year was designated the time of PHV.

Several parameters were recorded from standing posteroanterior radiographs (Figures 1–3). Other measurements in this review include the rib–vertebral angle (RVA) difference, as described by Mehta,¹⁷ with separate angles recorded for both the convex side rib (RVA-cx) and the concave side rib (RVA-cv); the apical vertebral rotation (AVR) for both thoracic (AVR-T) and lumbar or thoracolumbar (AVR-L) curve apexes, as described by Nash and Moe ²² ; and the association between the vertebral tilt angles (VTAs), as described by Appelgren and Willner.2 Last, ratios between the thoracic and lumbar or thoracolumbar Cobb angles were calculated (Cobb-R = thoracic Cobb/lumbar Cobb), as was the ratio between the apical vertebral translations (AVT-ratio = AVT-thoracic/AVT-lumbar) (Figure 2). Each measurement was recorded at the time of brace prescription, initial in-brace, brace discontinuation, and, for patients that did not undergo immediate surgical correction because of treatment failure, at final follow-up.

Figure 1. The center sacral line (CSL) is a vertical line parallel to the outside edge of the radiograph, bisecting the spinous process of the second sacral vertebrae. Cobb 7 angles were measured. Vertebral tilt angles (VTA) for both thoracic and lumbar vertebrae used in measuring the Cobb angles were measured at the intersection of a line perpendicular to the CSL. For each thoracic (T) and lumbar (L) or thoracolumbar (TL) curve, both the superior (VTA-T-S; VTA-L-S) and inferior (VTA-T-I; VTA-L-I) vertebral tilt angles were measured. For lumbar or thoracolumbar curves the pelvic tilt angle indicated the association between the iliac crest line and the CSL. The lumbar pelvic relationship (LPR) is the angle formed between a line drawn on the inferior endplate of the last vertebra in the lumbar curve and the iliac crest line.

Figure 2. The plumbline is a line parallel to the CSL that bisects the body of C7. The distance between these lines is considered the coronal decompensation (CD). The apical vertebral translation (AVT) for thoracic curves (AVT-T) is the lateral distance from the thoracic apical vertebra center to the plumbline; for lumbar or thoracolumbar (AVT-L) curves this measurement was the lateral distance from the center of the apical lumbar vertebral body to the CSL. The relative apical distance (RAD) is the lateral distance between the vertebral body centers of the thoracic and lumbarapexes.24

To compare our findings with a previous report² on the association of VTAs in double curves, we averaged the values for VTA-T-I and VTA-L-S (Figure 1) to create a single value for the vertebra shared as an endpoint for both the thoracic and the lumbar or thoracolumbar curves, respectively. This angle was considered the VTA of the "transitional" vertebra and was termed end-vertebra angle "B" (EVA-B). The tilt of thoracic vertebrae considered the superior endpoint of the thoracic curve (VTA-T-S) was termed end-vertebra angle "A" (EVA-A). By doing so, we were able to categorize these patients into three distinct groups:
   Type 1: EVA-A.B: Where the tilt angle recorded for VTA-T-S exceeded that of the transitional vertebrae by .5°;
   Type 2: EVA-B.A: Where the tilt of the transitional vertebrae exceeded that recorded for VTA-T-S by .5°;
   Type 3: EVA-Symmetric: Where these two measurements were within 5° of each other, thus indicating a more symmetric association.

Patients were categorized as having either a double curve pattern (combination of thoracic and lumbar or thoracolumbar curve), a single thoracic curve, or a single lumbar or thoraco-lumbar curve. For purposes of this study, a double curve was classified as such when the concave side pedicle of the apical vertebra of the thoracic curve crossed the plumbline, and the concave side pedicle of the apical vertebra of the lumbar or thoracolumbar curve crossed the CSL (Figure 2). Patients with double curves had both a "primary" and a "secondary" curve designated. For those with double curves, the larger curve (by $4°) was assigned primary curve status. If the curves were within 3° of each other at the time of brace prescription,¹⁴ primary curve status was assigned by the primary investigator.

A successful outcome resulted if the primary curve increased brace prescription to that taken when the brace was discontin-ued. Additionally, for double curve patterns to be considered a success, the secondary curve could not exceed the primary curve by more than 5° at the time of brace discontinuation. A brace failure occurred if the primary curve progressed .5° or if the secondary curve (if present) exceeded the primary curve by .5° at brace discontinuation. Progression to surgical intervention was also examined as a subset of those that failed treatment.

Figure 3. For thoracic curves the lateral trunk shift (LTS)9 was measured by first drawing a horizontal line to the edges of the ribs of the apical vertebra. A perpendicular line bisects the horizontal line. The distance between this perpendicular line and the center sacral line represents the LTS.

To evaluate the association between all data recorded and outcome, patients were grouped as treatment successes and failures. The mean differences for each subgroup's data at brace prescription were compared. The differences in the in-brace correction of the various radiographic parameters between the two groups were also analyzed. Differences were expressed as a percent change, the actual change in value, or a direct comparison between the brace prescription and in-brace values, re-spectively. Comparisons of both the prescription values for the two subgroups and the in-brace correction analyses were made by curve pattern and by primary curve assignment.

To determine if a particular value could be used as a threshold for predicting success, the following ad hoc method was used: For a given curve type, a specific value for percent correction was selected. If the percent success for those above that value was significantly greater than the percent success for those below that value, then that value could be considered a threshold value. The 2% success rates were compared using Fisher's exact test. The smallest value that could be used as a threshold is reported. If no such value could be found for a curve type, then we concluded only that we were unable to detect a threshold value with this sample size.

Mean differences were compared with traditional analysis of variance methods and the Tukey multiple comparison procedure. In the case of two-group comparisons, t tests were used. To compare rates, traditional x 2 methods were used, unless small sample size mandated the use of the Fisher exact test. A P value of #0.05 was required for statistical significance.

Results

Of the 51 patients (47 girls and 4 boys) reviewed in this study, 31 (61%) were considered treatment successes at the time of brace discontinuation. Twenty patients were considered treatment failures because of curve progression .5°. Ten of these 20 patients underwent surgical correction before being fully skeletally mature because of curve progression during brace treatment.

We compared the mean values for all radiographic measurements at the time of brace prescription in the 31 patients who were considered treatment successes versus the mean values in the 20 patients that failed treatment. This analysis yielded some trends, but no measurements were found to be statistically predictive of outcome (Table 1).

Maturity as a Predictor

We analyzed the impact of patient maturity on outcome. Chronologic age, Risser sign, and the timing of menarche in girls at the time of brace prescription were not significant predictors for outcome (Tables 2–4). We were able to reliably deduce the timing of PHV in 44 of the 51 patients reviewed. Twenty-one patients demonstrated a growth velocity of at least 9 cm/yr, thus designating a calendar date and age for PHV. Thirty patients lacked a visit date where growth velocity exceeded 9 cm/yr, seven of whom required surgical correction before skeletal maturity; thus, no PHV could be assigned. The remaining 23 patients that were followed to skeletal maturity without any demonstrable peak in growth velocity were presumed to have achieved their peak growth age before the start of brace treatment. Therefore, we determined there to be 20 patients who were braced before achieving PHV, 24 patients who were braced after PHV, and 7 in whom the timing of PHV was indeterminate. The treatment success rate between these two groups of patients was not significant (Table 5).

We evaluated the combined impact of both Risser sign and menarcheal status on outcome in the 47 girls reviewed. Twelve of 23 (52%) that were both Risser 0 and premenarcheal at the time of brace prescription were treatment successes compared with 17 of 24 (71%) girls that were Risser 1 or 2, being either premenarcheal or postmenarcheal. Although suggestive of a tendency toward a higher risk for progression in the less mature girls with large curves, this difference was not statistically significant (P = 0.188).

Association of Curve Pattern and Outcome

Thirty-one patients (30 girls and 1 boy) had double curves, 12 had single thoracic curves (9 girls and 3 boys), and 8 had either single lumbar or single thoracolumbar curve patterns (8 girls and no boys). Of the 31 patients with double curves, 17 (55%) were treatment successes compared with 8 of 12 (67%) patients with single tho-racic curves and 6 of 8 (75%) with single lumbar or thoracolumbar curves (P = 0.519). Of the 20 patients that failed treatment, 6 of 14 with double curves required surgical correction before completing brace treatment, as did 3 of the 4 patients with single thoracic curves, and 1 of 2 patients with single lumbar or single thoracolumbar curves.

Only patients with double curves were found to have radiographic values predictive of outcome. The association between the VTAs and the comparative mean values for the lumbar pelvic relationship (LPR) were signifi-cantly different between those that were treatment suc-cesses versus those that failed treatment. Of the 31 pa-tients with double curves, 4 patients were classified as EVA-A.B (Type 1), 11 patients as EVA-B.A (Type 2), and 16 patients as EVA-Symmetric (Type 3). None (0%) of the 4 four patients with Type 1 were treatment suc-cesses compared with a treatment success rate of 8 of the 11 (73%) patients with Type 2 and 9 of the 16 (56%) patients with Type 3. The difference between these three groups was significant (P = 0.04).

Combined Effect of Thoracic Curve Cobb Angle and the LPR

We found the LPR angle to be a significant predictor of outcome in patients with a double curve when the thoracic curve measured at least 36°. In the 24 patients comprising this subgroup, the average LPR angle in the 13 patients that were a treatment success was 9.6° versus an average LPR angle of 14.5° in the 11 patients that failed treatment (P = 0.001). An LPR of 12° was found to be a threshold value predictive of outcome. Of the 16 patients with both a thoracic curve of at least 36° and an LPR of #12°, 12 (75%) were treatment successes, whereas only 1 of 8 patients (13%) with both a large thoracic curve (.35°) and a large LPR angle (.12°) was a treatment success (P = 0.006).

Conversely, in patients with a primary lumbar or thoracolumbar curve, whether it be a single curve pattern or when it is considered the more structural curve in a patient with a double curve pattern, the LPR angle was not predictive of outcome. In the 11 patients with a primary lumbar or thoracolumbar curve that were considered treatment successes, the average LPR angle was 17.8°. The average LPR angle in the seven patients having a primary lumbar or thoracolumbar curve that failed treat-ment was 18.7° (P = 0.47). Thus, even though the LPR angle in patients with primary lumbar or thoracolumbar curves was considerably larger than those in whom the thoracic curve was at least 36° (18.2° vs. 11.8°, respec-tively), it was not a significant predictor for outcome in these patients.

Because the size of the lumbar or thoracolumbar curve in a double pattern can influence the size of the LPR angle, we evaluated the association of the two Cobb angles and outcome in these patients. Of the 10 patients with Cobb angles of at least 36° in both thoracic and lumbar or thoracolumbar curves, five (50%) were treatment successes. This success rate did not differ from the remaining 21 patients (57% success rate) with double curve patterns where only one curve measured at least 36°.

Impact of In-Brace Correction

There was a significant positive correlation (P = 0.02) between the amount of in-brace correction of the primary curve's Cobb angle and the orthoses' ability to prevent curve progression in patients with double curves (Table 6). A threshold of 25% correction of the primary curve's Cobb angle was statistically predictive of outcome. Of the 31 patients with double curves, 27 had in-brace radiographs available for review. Of these 27 patients, 15 patients had at least 25% in-brace correction, 11 of whom (73%) were treatment successes. Of the 12 patients who had ,25% in-brace correction, only three (25%) were treatment successes. This difference was significant (P = 0.021). No threshold value for in-brace correction of the Cobb angle was predictive of outcome in patients with single curves.

Impact of Wear Schedule

We examined the medical records for documentation of wear schedule as reported to the physician by the patient and family for all subjects. Every patient was found to have at least one report of a brace wear schedule. Documentation of a physician's prediction of wear schedule was noted in 229 of 335 (68%) clinic visits recorded. Patients considered a treatment success reportedly wore their orthosis significantly more than those that failed treatment (17.1 hours/day vs. 11.5 hours/day, respec-tively; P = 0.012). Categorizing these patients into three distinct, clinically relevant groups yielded similar results (P = 0.022; Fisher Exact Test) (Figure 4).

Curve Behavior After Brace Treatment

Of the 51 patients reviewed, 41 patients were kept in the brace until skeletal maturity was achieved. Brace treatment was discontinued before skeletal maturity in the remaining 10 patients to undergo surgical correction because of curve progression. Seven patients (six that were considered a treatment success and one that failed treatment) were lost to postbrace observation. Thus, 34 pa-tients were subject to postbrace follow-up review.

The mean follow-up duration was 32 months (2.7 years), with a range of 3–138 months (11.5 years). Of the 34 patients with follow-up, 25 were considered treatment successes and nine were considered treatment fail-ures at the time of brace discontinuation. Of the 25 patients originally considered treatment successes, 17 remained stable at follow-up, as there was still no pro-ression of their primary curve of .5° comparing the Cobb angles at the time of brace prescription to that at final follow-up. Eight patients originally considered treatment successes at brace discontinuation demon-strated curve progression beyond the 5° threshold at final follow-up (average change in Cobb from brace prescription to final follow-up 9.5°; range 7–14°). Five of these eight patients ultimately required surgical correction. Of the nine patients with follow-up that were considered treatment failures because of progression at the time of brace discontinuation, one went to surgery.

Figure 4. The influence of reported brace wear on outcome.

To summarize, in the final analysis of the 51 patients reviewed in this study, there were 23 patients (45%) [note: six of whom were lost to postbrace discontinuation follow-up] that never demonstrated .5° of curve progression, 12 patients (24%) [note: one patient lacking follow-up] that ultimately progressed .5° but did not require surgery, and 16 patients (31%) who ultimately required surgical correction.

Discussion

The purpose of a brace in the treatment of AIS is to prevent progression of a curve that would otherwise be expected to worsen without treatment. Although there is evidence that an orthosis can alter the natural history of progressive scoliosis in adolescents,^{13,16,21,23,25} there tends to be agreement in the literature that larger curves are more likely to worsen than smaller curves.^8,13,16,25 Ideally, an orthosis would be prescribed for curves that are known to be progressive so as not to provide unnecessary treatment. Because a higher number of patients with curves .35° will ultimately progress to the point of requiring surgical correction versus patients with smaller curves, we set out to determine what, if any, clinical or radiographic findings may be predictive of outcome. The primary intent of this review is to provide clinically useful tools not yet reported in the literature that the physician can use when considering brace treatment in pa-tients with AIS presenting with a curve ranging from 36° to 45°. The secondary intent is to provide further insight into whether or not brace treatment should be continued for a given patient based on in-brace radiographs taken early in a treatment regimen as well as on a patient's reported wear schedule.

In the 51 patients reviewed in this series, 31 (61%) did not progress .5° from the time the Boston brace was prescribed to the point in which it was discontinued because of skeletal maturity. Ten patients (20%) had their orthoses discontinued before skeletal maturity to pursue immediate surgical correction. With postbrace treatment follow-up observation, the treatment success rate diminished to 23 patients (45%), with 16 (31%) ultimately requiring surgical correction. It is difficult to determine the best time throughout a treatment and observation regimen to assess the efficacy of brace treatment. Lon-stein and Winter argued that the time of brace discontinuation at skeletal maturity is the most appropriate time to make accurate comparisons with natural history data. ¹⁶ Prior bracing reports have defined outcome by comparing radiographs at brace prescription with those at brace discontinuation.^{8,11-13,18,25} We concur that an orthosis should only be judged for efficacy while prescribed to be worn, and to be consistent with prior reports, we adopted this criterion for assessing outcome and all other statistical analyses. Because of the increased likelihood that patients with large curves may demonstrate curve progression after reaching skeletal maturity,³¹ we elected to report a more in-depth post-treatment follow-up analysis.

We evaluated the usefulness of various maturity indicators as a possible tool in predicting treatment outcome. No maturity indicators were prognostic for outcome in these patients. These data suggest it is as appropriate to initiate a brace treatment program in a patient with AIS and a curve ranging from 36° to 45° who is clearly skeletally immature as it is for one who is suspected to have less growth remaining.

Outcome was based on changes in Cobb angles between the time of brace prescription versus brace discontinuation. No radiographic measurement recorded at the time of brace prescription was a significant predictor of outcome by analyzing all 51 patients reviewed as a group (Table 1). Prior bracing reports have shown variances in outcome as related to curve patterns.^{8,13,16,19,25} We were able to identify radiographic values predictive of outcome only in patients with double curves.

The association between the EVAs was prognostic for outcome in patients with double curves. This method allows the observer to differentiate the association between the endpoints of the two curves,² providing a curve pattern analysis not possible by using the Cobb method alone. We found that if the tilt of the superior vertebral endpoint of the thoracic curve (VTA-T-S) exceeds that of the tilt of the vertebra shared as an endpoint for both curves (EVA-A.B), a patient was significantly less likely to have a successful outcome. This EVA Type 1 pattern was the rarest of the three possible patterns, claiming only four of the 31 patients with double curves. Each of these four patients, however, failed treatment (P = 0.04). Our findings differ somewhat from those reported by Appelgren and Willner ² ; however, a direct comparison between the two reviews is difficult because of differences in study methodologies. In their larger series of 104 patients, each of whom were nonsurgical patients with double curves of varying Cobb magnitudes treated with a Boston brace, they reported a greater deterioration in the thoracic Cobb angle at the 2-year follow- up if the tilt of the superior EVA was smaller than that of the inferior end vertebra by .5° (EVA-B.A).²In patients with major curves where the thoracic curve was at least 36°, an LPR of 12° was found to be a threshold value that significantly influenced outcome. The significance of the LPR in combination with a large thoracic curve was not simply a matter of the lumbar or thoracolumbar curve also being large. Patients presenting with two large curves did not have a significantly lower treatment success rate compared with those in whom only one curve exceeded 35°. Further, the size of the LPR angle in patients with a primary lumbar or thoracolumbar curve far exceeded those found in patients with double curves where the thoracic curve was considered the primary curve, yet their treatment success rates were better. Therefore, an LPR angle of .12° should be considered for clinical decision making only in those patients with a double curve pattern where the thoracic curve is .35°. Seven of eight patients that met this description in this review failed treatment.

Why is the size of the LPR angle relevant only in patients with a double curve pattern, having a thoracic curve .35°? The combination of the poor base of support that exists with a large obliquity between the bottom half of a lumbar curve and the pelvis in a spine that is simultaneously attempting to balance itself in the presence of a large thoracic curve may be a contributing factor to these patients being less responsive to brace treatment. This finding is similar to a previous report 26 demonstrating a positive correlation with the obliquity found between the lower lumbar spine (L4) and the pelvis and postoperative imbalance in Type II ¹⁴ curves, the same curve pattern where the LPR was found to be significant in this review. Is this a matter of the inability of an orthosis to reduce the LPR to a level in which equilibrium can be achieved and thus prevent further curve progression, or is the combined effect of a large thoracic curve and a large LPR a contraindication for brace treatment? We found the LPR correction in double curves to be considerably less than in those with single lumbar or thoracolumbar curves (Table 6). Because this is a retrospective review, we were unable to accurately assess the impact of brace design on reducing the LPR angle. To prospectively study the usefulness of an orthosis designed to maximally reduce the size of the LPR and Cobb angles would be helpful in determining whether or not these patients should still be considered candidates for brace treatment.

Beyond seeking factors that may be considered useful in determining a criterion for treating large curves, we also analyzed the virtue of continuing treatment once prescribed. We evaluated the amount of in-brace correction for all radiographic measurements. Previous reports have stressed the importance of in-brace correction of the Cobb angle and how it may have a positive correlation with treatment success.^{5,6,8,10,13,15,19,23,28} –30 We found the percent in-brace correction of the Cobb angle to have a significant positive correlation with treatment success only in patients with double curves. Twenty-five percent in-brace correction of the primary curve was found to be a threshold value prognostic for outcome in these 27 patients having in-brace radiographs available for review. Our inability to identify a threshold value for success in those with single curves is thought to be an issue of sample size rather than clinical relevancy. With study populations of only 12 patients with single thoracic curves and eight with single lumbar or thoracolumbar curves, and treatment success rates of 67% and 75%, respectively, there simply were not enough cases that failed treatment to compare in-brace values with those that were considered treatment successes.

Other in-brace radiographic measurements that correlated with a successful outcome included the amount of lateral translation seen in the apical vertebra for lumbar or thoracolumbar curves in patients with double curves, and for patients with single thoracic curves, the percent change seen in the rib–vertebrae angle (Table 6). The reasons for these findings are purely speculative, but, as with the ability to decrease the Cobb angle in-brace, it could be argued the more flexible the spine and its association with the thorax, the more responsive it may be to treatment.

Compliance with brace wear is an important and elusive factor in brace treatment. Some studies have suggested part-time brace wear to be as effective as a 23 hours/day treatment regimen.^1,8,11,25 A metaanalysis of 20 studies culled from the Prevalence and Natural History Committee of the Scoliosis Research Society suggested braces that were worn 23 hours/day to be significantly more successful in preventing curve progression versus those worn 8 or 16 hours/day.²⁷ One report on the use of the Boston brace in the treatment of high-risk patients with larger curves (range 35–52°) found those who reported wearing the brace 23 hours/day experienced a significantly higher success rate in stopping further progression versus patients who reportedly wore their orthoses #18 hours/day.¹⁸ We found the number of reported hours of wear per day to be prognostic for outcome (Figure 4). Orthoses designed for at least 16 hours/day wear have been shown to have significantly higher treatment success rates than those designed only to be worn during sleep.^12,13 Although there appears to be a dose-related criterion in controlling some curves, especially in curves .35°, each report on wear schedule to date is based on subjective reports by the patient or caregiver. Additional research is needed to better understand brace wear schedule thresholds required for success.

Acknowledgment

The authors thank Richard H. Brown, PhD, of the Texas Scottish Rite Hospital for Children for his expertise in providing the necessary statistical analysis for this study.

Key Points

• In patients with large, double curves from adolescent idiopathic scoliosis treated with a Boston brace, the size of the lumbar pelvic relationship angle, the thoracic vertebral tilt angles, and the amount of in-brace correction of the Cobb angle significantly influence treatment outcome.

• There is a significant positive correlation with the amount a brace is reportedly worn and treatment success.

References

1. Allington NJ, Bowen JR. Adolescent idiopathic scoliosis: treatment with the Wilmington brace. J Bone Joint Surg Am 1996;78:1056–62.

2. Appelgren G, Willner S. End vertebra angle: a roentgenographic method to describe a scoliosis. A follow-up study of idiopathic scoliosis treated with the Boston brace. Spine 1990;15:71–4.

3. Buckler JMH. A Longitudinal Study of Adolescent Growth. London: Springer-Verlag, 1990.

4. Bunch WH, Patwardhan AG. Understanding progression. In: Scoliosis: Making Clinical Decisions. St. Louis: CV Mosby, 1989:50–67.

5. Bunnell WP, MacEwen GD, Jayakumar S. The use of plastic jackets in the non-operative treatment of idiopathic scoliosis. J Bone Joint Surg Am 1980; 62:31–8.

6. Carr WA, Moe JH, Winter RB, et al. Treatment of idiopathic scoliosis in the Milwaukee brace. J Bone Joint Surg Am 1980;62:599–612.

7. Cobb JR. Outline for the study of scoliosis. In: Edwards JW, ed: Instructional Course Lectures, vol. 5. Ann Arbor, MI: American Academy of Orthopaedic Surgeons, 1948:261–75.

8. Emans JB, Kaelin A, Bancel P, et al. The Boston bracing system for idiopathic scoliosis: follow-up results in 295 patients. Spine 1986;11:792–801. 9. Floman Y, Denny N, Riseborough EH, et al. Osteotomy of fusion mass in scoliosis. J Bone Joint Surg Am 1982;64:1307–16.

9. Gardner ADH, Burwell RG, Wozniak AP, et al. Some beneficial effects of bracing and a search for prognostic indicators in idiopathic scoliosis. Spine 1986;11:779.

10. Green NE. Part-time bracing of adolescent idiopathic scoliosis. J Bone Joint Surg Am 1986;68:738–42.

11. Howard A, Wright JG, Hedden D. A comparative study of TLSO, Charles-ton, and Milwaukee braces for idiopathic scoliosis. Spine 1998;22:2404–11.

12. Katz DE, Richards, BS, Browne RH, et al. A comparison between the Boston brace and the Charleston bending brace in adolescent idiopathic scoliosis. Spine 1997;22:1302–12.

13. King HA, Moe JH, Bradford DS, et al. The selection of fusion levels in thoracic idiopathic scoliosis. J Bone Joint Surg Am 1983;65:1302–13.

14. Laurnen EL, Tupper JW, Mullen MP. The Boston brace in thoracic scoliosis: a preliminary report. Spine 1983;8:388–95.

15. Lonstein JE, Winter RB. The Milwaukee brace for the treatment of adolescent idiopathic scoliosis. J Bone Joint Surg Am 1994;76:1207–21.

16. Mehta MH. The rib-vertebra angle in the early diagnosis between resolving and progressive infantile scoliosis. J Bone Joint Surg Am 1972;54:230–43.

17. Mitchell TM, Smith BG, Thomson JD. Effectiveness of the Boston brace in the treatment of large curves in adolescent idiopathic scoliosis. Proceedings of the 31st Annual Meeting of the Scoliosis Research Society, Ottawa, On-tario, Canada, 1996.

18. Moe JH, Kettleson DN. Idiopathic scoliosis: analysis of curve patterns and the preliminary results of Milwaukee brace treatment in one hundred sixty-nine patients. J Bone Joint Surg Am 1970;52:1509–33.

19. Montgomery F, Willner S, Applegren G. Long-term follow-up of patients with adolescent idiopathic scoliosis treated conservatively: an analysis of the clinical value of progression. J Pediatr Orthop 1990;10:48–52. 21.

20. Nachemson AL, Peterson L-E, Brace Study Group of the Scoliosis Research Society. Effectiveness of treatment with a brace in girls who have adolescent idiopathic scoliosis. J Bone Joint Surg Am 1995;77:815–22.

21. Nash CL, Moe JH. A study of vertebral rotation. J Bone Joint Surg Am 1969;51:223–9.

22. Olafsson Y, Saraste H, So¨ derlund V, et al. Boston brace in the treatment of idiopathic scoliosis. J Pediatr Orthop 1995;15:524–7.

23. Patwardhan AG, Rimkus A, Gavin TM, et al. Geometric analysis of coronal decompensation in idiopathic scoliosis. Spine 1996;21:1192–200.

24. Price CT, Scott DS, Reed FR, et al. Nighttime bracing for adolescent idiopathic scoliosis with the Charleston bending brace: long-term follow-up. Pediatr Orthop 1997;17:703–7.

25. Richards BS. Lumbar curve response in type II idiopathic scoliosis after posterior instrumentation of the thoracic curve. Spine 1992;17(suppl): 282–6.

26. Rowe DE, Bernstein SM, Riddick MF, et al. A meta-analysis of the efficacy of non-operative treatments for idiopathic scoliosis. J Bone Joint Surg Am 1997;79:664–74.

27. Uden A, Willner S, Pettersson H. Initial correction with the Boston thoracic brace. Acta Orthop Scand 1982;53:907–11.

28. Upadhyay SS, Nelson IW, Ho EKW, et al. New prognostic factors to predict the final outcome of brace treatment in adolescent idiopathic scoliosis. Spine 1995;20:537–45.

29. Watts HG, Hall JE, Stanish W. The Boston brace system for the treatment of low thoracic and lumbar scoliosis by the use of a girdle without superstruc-ture. Clin Orthop 1977;126:87–92.

30. Weinstein SL, Zavala DC, Ponseti IV. Idiopathic scoliosis: long-term follow- up and prognosis in untreated patients. J Bone Joint Surg Am 1981;63: 702–12.

Address reprint requests to
Donald E. Katz, BS, CO
Orthotics Department
Texas Scottish Rite Hospital for Children
2222 Welborn
Dallas, TX 75219

From the Orthotics Department, Texas Scottish Rite Hospital for Chil-dren, Dallas, Texas.
Acknowledgment date: October 4, 1999.
First revision date: March 3, 2000.
Second revision date: July 3, 2000.
Acceptance date: April 24, 2001.
Device status category: 11.
Conflict of interest category: 12,14.