The decision to choose between a CTLSO (Milwaukee brace) and a TLSO (such as the Boston system, Wilmington, Miami or Rosenberger orthosis) is affected by several factors including cosmesis, geographical preference and popularity of a given orthosis. Performance usually is a secondary consideration since objective comparisons have been difficult.
A finite element model was used to quantify and compare the effects of the CTLSO and TLSO in increasing spinal stability as measured in terms of the critical load of right primary thoracic with left lumbar compensatory and left primary lumbar with right compensatory idiopathic scoliotic curves.
While the CTLSO and TLSO were equal in stabilizing lumbar primary curves, the TLSO was 25 percent less effective in stabilizing primary thoracic curves. Pad and counterforce placement were critical factors influencing stability for both the CTLSO and TLSO, demonstrating the importance of proper initial fit and timely growth adjustments to ensure proper pad placement throughout the duration of wear.
Key Words: Idiopathic Scoliosis, Milwaukee Brace, TLSO, Orthotic Stabilization, Biomechanics.
Several different orthoses currently are available for nonoperative treatment of idiopathic scoliosis (1). The Milwaukee cervico-thoraco-lumbo-sacral orthosis (CTLSO) (see Figure 1 ) is the traditional standard and has a well-documented history of results (2-6). During the 1970s, low-profile or "underarm-type" orthoses were developed to treat lumbar and thoracolumbar curves (see Figure 2 ) (7-10). Known as thoracolumbo-sacral orthoses (TLSOs), they were later advocated to treat curves with thoracic primary curves (1).
The decision regarding whether to use a TLSO or a CTLSO is primarily based on cosmesis, preference of a given orthosis in certain geographic areas and popularity of a specific orthotic technique (such as the part-time criteria of the Charleston TLSO).
The TLSO is a more cosmetically appealing orthosis since it can be easily concealed by clothing, unlike the CTLSO, which has a neck ring that protrudes above clothing. Recent developments of the low-profile neck ring have attempted to address this cosmetic concern.
In certain regions of North America, specific types of orthoses are geographically preferred because of variation in orthotic training between regions where specific TLSOs were developed.
A few areas still prefer the CTLSO in spite of the unsightly superstructure and neck ring, whereas TLSOs, such as the Miami orthosis, are preferred in much of the Southeast (the geographic region of its originators). The Boston TLSO system is preferred in the Northeast for the same reason.
Despite the void of longitudinal studies, the Charleston TLSO has become popular nationwide because it is cosmetically appealing and is prescribed for night-time wear only, allowing patients to be out of the orthosis during waking hours. Mechanical performance has not been a primary factor in deciding which orthosis to use for idiopathic scoliosis.
While several types of orthoses exist, they all function by the same mechanisms of action: endpoint control, curve correction and continuous transverse support. Endpoint control prevents sway of the vertebral column and reduces gross trunk motion while lumbar and thoracic pads reduce scoliotic curves and maintain curve reduction for the duration of wear, providing continuous lateral support at the apex of the reduced curve. The synergistic effects of endpoint control, curve correction and continuous lateral support in orthotic stabilization of scoliotic curves with a biomechanical model have been demonstrated in previous studies (11,12).
While the mechanisms of action of the CTLSO have been investigated experimentally (13,14) and with finite element models (15), little is known about the TLSO's mechanisms of action. There are no objective data on the effects of trimline height and pad placement on curve correction and stability achieved by a TLSO. As a result, objective comparisons of the effectiveness of the CTLSO and TLSO in treating different curve patterns have not been possible.
This study was designed to test the primary hypothesis that the CTLSO is superior to the TLSO in stabilizing primary thoracic idiopathic scoliotic curves and the secondary hypothesis that proper pad placement is essential in providing optimum curve stability with the CTLSO and TLSO.
A finite element model of the spine-orthosis system was used to evaluate the stability of spinal curves for a variety of conditions. This mathematical computer model is capable of analyzing spinal geometries and response to load and was validated from patient radiographs. The model allowed simulation of different curve geometries and variation of orthotic design parameters such as pad placement and trimline location relative to the curve apices and endpoints. These simulation studies are useful in identifying, for each curve pattern, the optimum placement of pads and trimline for the most stable curve, i.e., one that can support the most axial load without progressing.
The stabilizing effect of an orthosis was studied for two combined thoracic and lumbar curve patterns: a 35-degree primary right thoracic curve with a 20-degree compensatory left lumbar component and a 35-degree primary left lumbar curve with a 20-degree compensatory right thoracic component (see Table A ). These two curve patterns will be referred to in this article as the primary thoracic curve and the primary lumbar curve, respectively. The compensatory component was assumed to be twice as flexible as the primary component. This is consistent with clinical observations of curve correction seen in the primary and compensatory components upon lateral bending (16).
A finite element model of the spine was used to simulate the response of the curve in the frontal plane to external loads. Each motion segment of the thoracolumbar spine (T1-sacrum) was modeled as a beam element with end nodes located at the centroids of adjacent vertebral bodies. Similar models have been used in previous studies of scoliosis (17,18).
The flexural rigidities of the beam elements were chosen based on stiffness identification made with the same type of spine model (17) and are comparable to data reported in the literature (13,19). Nonlinear effects due to large displacements were incorporated in the model by updating the geometry and global stiffness matrix at each incremental load step.
The effect of the pelvic interface on the CTLSO was simulated by constraining all degrees of freedom of the sacrum. Since the neck ring keeps the head and neck centered over the pelvis, the first thoracic vertebra was constrained from moving in the lateral direction. However, the T1 vertebra was not constrained in the inferior-superior direction and was allowed to rotate in the frontal plane.
The model parameters of the TLSO used in this study are consistent with the clinical parameters of a well-fitted Rosenberger, Boston or Miami orthosis (see Figure 2 ). These TLSOs rely on the use of a trimline on the concavity of the primary curve to provide a counterforce that, in conjunction with the lateral pad loads, provides the triangulated force system to correct and stabilize the curve.
In the TLSO model, the sacrum was fixed to simulate the stabilizing effect of the pelvic section of the orthosis, exactly like the Milwaukee brace. The trimline of a TLSO provides lateral support to the ribs. As the ribs are pushed against the trimline due to pad loads, the stiffness of the trimline-rib complex provides resistance to lateral displacement of the curve. The equivalent stiffness of the trimline-rib complex was simulated by linear springs along the medial-lateral degree of freedom at appropriate vertebral levels.
The values of spring constants were derived using the data reported by Agostoni et al. (20) and Andriacchi et al. (21). Since a TLSO imposes no kinematic constraints on the first thoracic vertebra, T1 was not constrained in the medial-lateral and inferior-superior directions and was allowed to rotate in the frontal plane.
To simulate the effect of the CTLSO on the correction and stability of primary thoracic curves, axillary sling forces must be considered in addition to the thoracic and lumbar pad forces (see Figure 3a ). The axillary sling eliminates the neck-ring reaction when pad loads are applied.
The axillary sling force was calculated to minimize the reaction force at T1 for each simulated case of thoracic and/ or lumbar pad placement. The effect was represented by equal forces along the medial-lateral line of action at T5 and T6 levels. The updated coordinates of the centroids of each vertebra (TiL5) due to application of pad loads were obtained from the finite element analysis.
The interaction of an orthosis with the scoliotic curve was simulated using a two-stage approach that simulated the procedure used in fitting a patient with an orthosis. In the first stage, the deformed shape of the spine was obtained under the application of external loads to simulate curve correction in the orthosis. These loads consisted of the transverse forces exerted by the thoracic and lumbar pads in the case of a TLSO (see Figure 4a ).
The second stage of the simulation represented the continuous lateral sup. port (or transverse load) provided by the pads to the corrected curve. Having arrived at the corrected geometry of the spine in the orthosis, kinematic constraints were introduced at the levels of pad placements and axillary sling location (in the case of the CTLSO). A linear spring was used between the vertebral centroid and the constraint boundary to simulate the stiffness of the orthosis wall-rib interface at that level (see Figure 3b and Figure 4b ).
The model spine then was loaded in axial compression to simulate the weight of the body segments above sacrum. The axial load was distributed over vertebral centroids T1-L5 using the mass distribution function (19). The stability of a scoliotic curve was expressed in terms of its critical load, defined as the value of the maximum axial load the spine can support without undergoing a permanent increase in its curvature. This concept is described in previous studies (11,12,18).
The stability of the primary thoracic and primary lumbar curve patterns (see Table A ) was evaluated as a function of location of the trimline of a TLSO relative to the superior endpoint of the curve and placement of the thoracic and lumbar pads relative to the corresponding apices of the curve components. These results were compared to those obtained with the CTLSO model for the same curve patterns.
In using the CTLSO for the primary thoracic curve, maximum stability can be achieved by placing the thoracic pad alone at the apex of the primary thoracic curve (see Figure 5 ). Incorrect placement of the thoracic pad inferior to the apex reduces the stability of the thoracic curve. Placement of the thoracic pad two levels inferior to the apex reduces the stability achieved with the correctly placed thoracic pad by 16 percent.
A lumbar pad at the apex of the compensatory lumbar curve tends to decrease the stability of the primary thoracic curve that is achieved with a correctly placed thoracic pad alone. A lumbar pad one level too superior causes a loss of 13 percent of the stability gained by a correctly placed thoracic pad alone.
Maximum stability of the primary lumbar curve can be achieved by placing a lumbar pad at the apex of the primary curve with an apical thoracic counterforce of a magnitude great enough to minimize the neck-ring reaction (see Figure 6 ). A lumbar pad alone at the apex of the primary lumbar curve without a thoracic counterforce results in a 25 percent smaller critical load value than that achieved when a thoracic counterforce is used. The lumbar pad alone is not as effective in stabilizing the primary curve because as the spine is loaded axially (the normal result of being upright), the curve tends to shift away from the lumbar pad, thus reducing its stabilizing effect. A thoracic counterforce acts to maintain contact between the lumbar pad and the primary curve, thereby improving its stability. Incorrect placement of the lumbar pad one level too superior reduces the stability by 12 percent.
In primary thoracic curves, maximum stability and curve correction using a TLSO are achieved with the axillary trimline at T5-6 and a pad at the apex of the thoracic curve (see Figure 7 ). As the trimline of the TLSO moves inferior relative to the superior endpoint of the curve, curve correction decreases, and a loss of stability of nearly 20 percent results with each level.
Moving the thoracic pad one level superior to the apex decreases the curve correction and causes a loss in stability by 13 percent (see Figure 8 ). Placement of the thoracic pad up to two levels inferior to the apex has no appreciable effect on curve correction and stability. Adding a lumbar pad at the apex of the compensatory curve decreases the effectiveness of a correctly placed thoracic pad since the critical load value decreases by nearly 15 percent. A lumbar pad one level too superior further decreases the correction of the primary curve and causes a 20 percent decrease in stability as compared to that achieved by a correctly placed thoracic pad alone. This is consistent with the behavior of the lumbar pad noted for the CTLSO treatment of primary thoracic curves.
In primary lumbar curves, optimum stability with a TLSO is achieved with the trimline located one level inferior to the apex of the compensatory thoracic component and a pad load at the apex of the primary curve (see Figure 9 ). Moving the trimline inferior toward the superior endpoint of the primary lumbar curve results in a significant loss in stability. For example, as the trimline location is moved from T9-10 to T11-12, the critical load decreases by approximately 45 percent.
However, little additional gain in stability of the primary lumbar curve is achieved by placing the trimline at the superior endpoint of the compensatory thoracic component. For example, the critical load of the primary lumbar curve supported by the trimline at T5-6 with thoracic and lumbar pads at corresponding apices is only 8 percent greater than that achieved by placing the trimline at T9-10 and pad load at the apex of the lumbar curve.
Incorrect placement of the lumbar pad has a detrimental effect on the stability of the primary lumbar curve. A lumbar pad load applied one level superior to the apex decreases the stability achieved by a correctly placed pad by nearly 10 percent.
In primary thoracic curves, maximum stability with the CTLSO is achieved with the maximal pad load and continued lateral support at the apex of the thoracic curve in conjunction with an axillary sling force to minimize the neck-ring reaction. Maximum stability using a TLSO in these curves is obtained by placing the trimline at the superior endpoint of the curve with a pad at the apex of the thoracic curve. A comparison of optimum results of the two orthoses shows that in primary thoracic curves the optimum stability achieved with a TLSO is 25 percent less than that obtained with the CTLSO (see Figure 10 ).
In primary lumbar curves, the best result with a TLSO is achieved with the axillary trimline one level inferior to the apex of the compensatory thoracic component and maximal pad load and continued lateral support at the apex of the lumbar curve. In using the CTLSO to stabilize primary lumbar curves, optimum stability is obtained with maximal pad support at the apex of the lumbar curve with an apical thoracic counterforce (pad load) great enough to minimize the neck ring reaction. A comparison of these two orthotic treatment modalities shows nearly equal critical load values for the primary lumbar curve (see Figure 10 ).
The critical load is the smallest value of axial load at which the maximum bending moment in the spinal curve reaches the elastic limit in bending and causes inelastic failure (12,18). Thus, the critical load is a measure of the capacity of the scoliotic spine to carry an axial load without causing a permanent increase in its curvature.
In a previous study (12), the authors used the critical load of scoliotic curve as a measure of its stability to understand the clinically observed relationship between curve magnitude and progression; at any age of maturity a larger curve is much more likely to progress than a smaller curve. Subsequently, the critical load measure was used to investigate the problem of progression in untreated curves of adolescents with idiopathic scoliosis as a function of curve pattern (18). The variations in the incidence of progression of different curve patterns as noted in the clinical studies were accurately mirrored in the findings of the biomechanical model of curve progression. Thus, thinking about differences in the stability of a scoliotic curve as measured by its critical load is a way of understanding the clinically observed behavior of scoliotic curves under a variety of clinical situations.
In this work, the authors have used the critical load of a scoliotic curve as measure of its stability for different boundary conditions that simulate different orthotic design parameters. Clearly, the absolute values of critical loads will depend on the assumed value of the elastic limit in bending and on other simplifying assumptions used in the model to approximate the complex structure. However, because of the comparative evaluation of the two orthotic treatment modalities, these assumptions do not affect the results.
In the present study, the authors used a finite element model of the spineorthosis system to simulate the interaction of the spinal orthosis with the spine in the frontal plane. It is recognized the spinal curvature in the sagittal plane and the rotation about the longitudinal axis of the involved vertebral bodies are important considerations in the morphology of idiopathic scoliosis. Further, the analysis was based on estimates of the mechanical properties of the spine and considered no active neuromuscular involvement. However, the results of this analysis provide a biomechanical rationale for clinical observations regarding the outcome of orthotic treatment for scoliosis.
The results indicate that in primary thoracic curves, maximum stability with the CTLSO is achieved with the maximal pad load and continued lateral support at the apex of the thoracic curve in conjunction with an axillary sling force to minimize the neck ring reaction.
Maximum stability using a TLSO in these curves is obtained by placing the trimline at the superior endpoint of the curve with a pad at the apex of the thoracic curve. These results do not undermine the need for a lumbar pad to reduce a lumbar compensatory curve; however, they may help explain why single thoracic curves, treated with only a thoracic pad, reduce more than thoracic curves in the presence of a lumbar component where both thoracic and lumbar pads are used.
In using the CTLSO to stabilize primary lumbar curves, optimum stability is obtained with maximal pad support at the apex of the lumbar curve with an apical thoracic counterforce great enough to minimize the neck ring reaction. In primary lumbar curves, the optimum result with a TLSO is achieved with the trimline one level inferior to the apex of the compensatory thoracic component and maximal pad load and continued lateral support at the apex of the lumbar curve.
Comparisons show a properly fitted TLSO to be as effective as the CTLSO in stabilizing the primary lumbar curves. On the other hand, in primary thoracic curves the optimum stability achieved with a TLSO is approximately 25 percent less than that obtained with the CTLSO. This helps explain why many, but not all, thoracic curves are well-controlled with a TLSO. For patients in whom a TLSO is not able to control a primary thoracic curve, a CTLSO may provide enough additional support to stabilize the curve.
However, as the results show, correct placement of the pads is important. Optimum results with an orthosis can be achieved when the pad loads are applied at the apical level of the curve. Incorrect placement of the pads reduces the stabilizing effect. This effect is greater in the lumbar area as a mistake of even one level too superior reduces the benefit of both the thoracic and lumbar pads. The detrimental effect of incorrect pad placement on the stability of the curve observed in the present study is consistent with the observations of Andriacchi et al. (15) who noted that correction of the thoracic curve decreased when the pad was placed two levels inferior to the apex of the curve. Therefore, careful attention must be given to pad placement.
Incorrect pad placement can occur as a result of at least two fabrication and fitting mistakes. First, it is quite possible the lumbar pad is simply positioned too high. The apex of the lumbar curves is just above the iliac crest, and it is easy to place the pad too superior.
The second mistake is much more common. The waist diameter can be fabricated and fitted too small, causing the entire orthosis to move superior. When this occurs, the correctly placed pads are functionally one or more levels too superior, and the effect of the pads is diminished. In addition, if the patient has grown since the last visit and the previously well-fitted pelvic girdle has migrated superiorly, the curve correction also is diminished. Thus, the results of the biomechanical model underscore the need for meticulous fit in the waist and pelvis and the importance of proper pad placement and routine follow-up adjustments for growth.
Clinical considerations of cosmesis and geographical regionalization of orthoses are helpful in determining which orthosis may be more tolerable for a particular patient. Other clinical factors in selecting an appropriate orthosis are 1) selecting the orthosis with which a clinical team in a given region has the most experience and 2) selecting the orthosis that has provided the best outcomes. Factors such as curve stability, age, growth rate and progression risk factor should be primary considerations when selecting the appropriate orthosis.
For thoracic primary curves, a well-fitted CTLSO provides more stability but may not be necessary for an older patient with a 25-degree curve that already has a high initial critical load value due to its small magnitude and low risk of progression because the patient is near maturity. A CTLSO may be more appropriate for an 11-year-old patient with a 45-degree curve since the critical load value is small due to the curve's large magnitude; such a patient's upcoming peak height velocity growth puts him or her in the highest progression risk group for adolescent idiopathic scoliosis. For a 25- or 45-degree lumbar primary curve, a TLSO will be adequate, regardless of age growth or stability, since the CTLSO does not provide greater stability (critical load) than the TLSO.
The most appropriate orthosis is one that will provide a balance of the best outcome, the most comfort and the greatest amount of cosmesis.
This study was supported in part by funds from the Rehabilitation Research and Development Center, Hines VA Hospital, Hines, Ill. The authors thank Patrick Carrico for his help in preparing the illustrations.
AVINASH C. PATWARDHAN, PhD, works in the department of orthopedic surgery at Loyola University, 2160 S First Ave., Maywood, IL 60153, (708) 343-7200, ext. 5804, and also works at the Rehabilitation Research and Development Center at the VA Hospital in Hines, Ill.
THOMAS M. GAVIN, CO, is president and director of clinical services for BioConcepts Inc. in Burr Ridge, Ill., as well as a teaching associate at Loyola University Medical Center and a research orthotist at the VA Hospital in Hines, Ill.
WILTON H. BUNCH, MD, PHD, is a student at the Church Divinity School of the Pacific in Berkeley, Calif.
VICTORIA M. DVONCH, MD, works in the department of human restoration at Stanford University in Palo Alto, Calif.
RAY VANDERBY JR., PhD, works in the division of orthopaedic surgery at the University of Wisconsin in Madison, Wis.
KEVIN P MEADE, PhD, works in the Rehabilitation Research and Development Center at the VA Hospital in Hines, Ill., and in the department of mechanical, materials and aerospace engineering at the Illinois institute of Technology in Chicago.
MARK SARTORI works in the department of orthopaedic surgery at Loyola University in Maywood, Ill., and at the Rehabilitation Research and Development Center at the VA Hospital in Hines, Ill.