Early Onset Idiopathic
Scoliosis
Abstract
Children with early onset scoliosis typically present before age 5
years. Radiographic criteria help to distinguish progressive cases
from those that will spontaneously resolve. Severe
cardiopulmonary problems may occur in untreated progressive
cases. A comprehensive evaluation should be performed to identify
commonly associated conditions, such as plagiocephaly, congenital
heart disease, inguinal hernia, and hip dysplasia. For curves >20°,
magnetic resonance imaging of the neural axis is indicated to rule
out occult central nervous system lesions. Surgical management
should be considered when nonsurgical measures, including
bracing and casting, fail to arrest progression. Surgical methods
continue to evolve and are primarily directed at obtaining and
maintaining curve correction while simultaneously preserving
spinal and trunk growth.
T
reatment of children with pro-
gressive scoliosis occurring be-
fore age 5 years is challenging. Left
untreated, progressive curves may
produce significant thoracic defor-
mity, leading to deleterious effects
on the cardiopulmonary system. As
James et al
1
observed in 1959, pro-
gressive early onset scoliosis “devel-
ops rapidly and relentlessly, causing
the severest form of orthopaedic

nal growth, and onset of scoliosis is
uncommon.
3,4
More recently, the
term “early onset” has been used to
reflect the presence of scoliosis of all
etiologies by age 5 years; “late on-
set” is used to describe the appear-
ance of scoliosis after age 5 years.
5
These terms more accurately reflect
the physiologic basis of and clinical
experience with this condition. The
distinction between early and late
Bruce L. Gillingham, MD,
CAPT, MC, USN
Ryan A. Fan, MD, LT, MC,
USNR
Behrooz A. Akbarnia, MD
Dr. Gillingham is Director, Surgical
Services, Naval Medical Center, San
Diego, CA, and Assistant Professor of
Surgery, Uniformed Services University
of the Health Sciences, Bethesda, MD.
Dr. Fan is Resident, Orthopaedic
Surgery, Naval Medical Center, San
Diego. Dr. Akbarnia is Clinical Professor
of Orthopaedic Surgery, University of
California, San Diego, and Medical
Director, San Diego Center for Spinal

Knowledge of the normal growth
of the chest, spine, and pulmonary
system is essential in understanding
the effect of scoliosis on these struc-
tures. The number of alveoli in the
terminal respiratory unit rapidly in-
creases in number and volume dur-
ing the first year of life. The number
of alveoli increases more than 10-
fold between birth and adulthood,
primarily during the first 8 years of
life.
6
In addition, the number of re-
spiratory branches increases from 21
at age 3 months to 23 at age 8
years.
7
The significant increase in
growth of the lung parenchyma is
paralleled by corresponding growth
of the spine and chest. Two thirds of
the final sitting height is achieved by
age 5 years.
3
The growth velocity of
the T1 to L5 segment is greatest
from birth to age 5 years, averaging
>2 cm per year, with marked decel-
eration between ages 5 and 10 years.

In contrast with late-onset dis-
ease, early onset idiopathic scoliosis,
specifically infantile idiopathic
scoliosis, spontaneously resolves in
a large number of patients. Although
early series reported a low incidence
of resolution, later reports demon-
strate a much more favorable
outcome.
8-11
Magnetic resonance im-
aging (MRI) evaluation of the neural
axis was unavailable at the time of
these reports, however. In 1951,
James
8
identified a pattern of scolio-
sis he termed “infantile idiopathic
scoliosis,” involving primarily in-
fant boys with a left mid or lower
thoracic curve. Only 4 of 33 patients
(12%) resolved. In contrast, Lloyd-
Roberts and Pilcher
10
reported on
100 patients with structural curves
diagnosed within the first year of
life. Ninety-two percent of the
curves resolved spontaneously. In
their study of 99 infants, Ceballos et

obliquity, and hip adduction. This
proposed correlation between infant
positioning and early onset scoliotic
deformities is disquieting and war-
rants fur ther research in light of the
current trend in American pediatrics
to recommend positioning infants
supine to reduce the risk of sudden
infant death syndrome.
13
Several factors, such as age of on-
set; location, type and magnitude of
the scoliotic curve; associated devel-
opmental anomalies; sex; and fami-
Figure 1
cm/year
2.5
2
1.5
1
0.5
0
Age 1
Age 5 Age 10 Age 15
2
2
1.25
1.25
1.25
0.75

nary vasculature is the primary
cause of the ventilation defect seen
in patients with early onset scolio-
sis.
15
Although progressive distor-
tion of the pulmonary architecture
by increasing deformity of the spine
and thorax would be thought to
compress the developing alveoli,
this has not been observed. Histolog-
ic studies of lung tissue in patients
with early onset scoliosis demon-
strate alveoli that are normal in
shape and outline but diminished in
number. In addition, the number of
alveoli for a given lung volume was
diminished more than could be ac-
counted for by limitation of space.
16
This effect is directly related to the
age of onset. The most hypoplastic
lungs are found in patients with the
earliest onset of scoliosis.
17
Patients with scoliosis demon-
strate a restrictive pattern of lung
disease, with reduction in vital ca-
pacity and total lung capacity along
with increased residual volume.

Clinically, there is a far greater
risk of cardiorespiratory complica-
tion when scoliosis is apparent be-
fore age 5 years. Even in the absence
of associated disease, disabling or
life-threatening respiratory failure is
relatively common and is likely to
present at or before late middle
age.
17,19
Physical Examination
The evaluation of a child with sus-
pected spinal deformity should begin
with a comprehensive history and
physical examination. Given that
the presence of cognitive delay has
been shown to correlate with curve
progression, particular attention
should be paid to whether the child
has appropriately reached develop-
mental milestones.
20,21
It is also im-
portant to obtain a birth history; ear-
ly onset idiopathic scoliosis has been
noted to occur more frequently with
breech presentation and in prema-
ture, low-birth-weight males.
21
A thorough physical examination

is calculated in a similar manner. Concave − convex = rib-vertebra angle difference.
(Adapted with permission from Mehta MH: The rib-vertebra angle in the early
diagnosis between resolving and progressive infantile scoliosis. J Bone Joint Surg
Br 1972;54:230-243.)
Bruce L. Gillingham, MD, CAPT, MC, USN, et al
Volume 14, Number 2, February 2006 103
Lower extremity limb-length equal-
ity is verified to rule out nonstruc-
tural scoliosis, particularly in the
ambulatory child with a predomi-
nant lumbar curve.
Following this focused evaluation
of the scoliotic curve and its second-
ary manifestations, the patient is ex-
amined for conditions associated
with early onset idiopathic scoliosis.
Plagiocephaly, which is extremely
common, is found in most patients
with early onset idiopathic scoliosis.
Developmental dysplasia of the hip,
inguinal hernia, and congenital heart
disease are found at a higher frequen-
cy in patients with early onset idio-
pathic scoliosis than in children
without scoliosis.
20,21,23
Finally, a careful neurologic ex-
amination, including assessment of
muscle tone and reflexes, is imper-
ative to detect occult central ner-

Anteroposterior and lateral radio-
graphs with full-length spinal views
should be obtained to evaluate both
the Cobb angle and the RVAD.
14
These radiographs are also helpful in
ruling out other associated vertebral,
lumbosacral, and hip joint abnor-
malities. The RVAD (Figure 2) is use-
ful in predicting curve progression
when there is no overlap of the rib
heads on the apical vertebra (phase 1
relationship)
14
(Figure 3, A). An
RVAD >20° indicates a high likeli-
hood of curve progression; curves
with RVAD measuring <20° are
more likely to resolve. A phase 2 re-
lationship, in which a rib head over-
laps the apical vertebra, implies that
progression is certain; thus, the
RVAD is not measured
14
(Figure 3,
B). In the Mehta study,
14
46 of the 86
phase 1 curves resolved. Of those
that resolved, 83% had an RVAD

the thoracic curve, downward obliq-
uity of the twelfth rib on the con-
cave side of the curve, and vertebral
rotation in opposite directions in the
thoracic and lumbar vertebrae.
Recent attention has focused on
defining the role of special imaging
for patients with early onset scolio-
sis. In contrast to patients with late-
onset scoliosis, the incidence of
occult central nervous system ab-
Figure 3
AB
Convex
Concave
Convex
Concave
A, Phase 1 rib-vertebrae relationship demonstrating no overlap of the rib head and vertebral body. B, Phase 2 rib-vertebrae
relationship. The overlap of the rib head on the vertebral body is indicative of curve progression. (Adapted with permission from
Mehta MH: The rib-vertebra angle in the early diagnosis between resolving and progressive infantile scoliosis. J Bone Joint
Surg Br 1972;54:230-243.)
Early Onset Idiopathic Scoliosis
104 Journal of the American Academy of Orthopaedic Surgeons
normalities is high in otherwise
healthy-appearing patients with
early onset scoliosis. In a prospec-
tive study of 34 patients younger
than age 10 years who presented
with cur ves measuring >20° and a
normal neurologic examination,

28
Management
Nonsurgical
The treatment of children with
early onset scoliosis is based on an-
ticipated or actual curve progression
(Figure 4). Mehta’s prognostic crite-
ria have proved to be very helpful in
differentiating between resolving
and progressive curves. Curves with
Cobb angles <25° with an RVAD
<20° are at low risk for progression.
These patients may be observed and
should be reevaluated with serial ra-
diographs every 4 to 6 months. Ac-
tive treatment should be initiated
with curve progression >10°. Upon
curve resolution, follow-up at 1- to
2-year intervals until maturity is
prudent to ensure that there is no re-
currence during the adolescent
growth spurt. A recent long-term
study of resolving curves validated
the use of the RVAD and demon-
strated that there was no advantage
of supine plaster bed treatment over
physiotherapy in regard to either
time to resolution or functional out-
come.
29

Specialty referral for
nonorthopaedic
conditions
Continue with nonsurgical
orthopaedic management;
progression to Cobb angle
>25°, RVAD >20°,
or positive phase 2 rib
relationship
Significant
nonorthopaedic
findings
Absent abdominal
reflexes or
Cobb angle >20°
Serial observation
every 4-6 months
Good response
Yes
No
Possible removal of
instrumentation and
continued observation
Annual clinical
examination until
skeletal maturity
Progression of
curve
Neurosurgery
specialty evaluation

ferentially fitting thoracolumbar
brace. In addition, the immature rib
cage often deforms before significant
correction is transmitted to the
spine. Bracing is generally continued
for a minimum of 2 years until there
is no further evidence of progression,
as indicated by an unchanging Cobb
angle and RVAD.
Mehta and Morel
22
studied 21 pa-
tients with infantile idiopathic scoli-
osis who were treated nonsurgically.
They reported that with total correc-
tion before the prepubertal growth
spurt, there is no relapse during ad-
olescence. Without full correction,
however, small relapses may occur.
These patients may require surgical
intervention if further progression
occurs during the adolescent growth
spurt. They should be followed until
skeletal maturity.
Surgical
Fusion
Several surgical procedures have
been used to manage progressive
curves in skeletally immature pa-
tients. Early procedures focused on

cant improvement. However, 40%
were stationary or showed very little
improvement (<10°).
In a recent study reviewing long-
term results, Marks et al
32
found
that convex spinal epiphysiodesis
with or without Harrington instru-
mentation did not significantly re-
verse the established deformity.
Thirteen patients with infantile id-
iopathic scoliosis were treated with
anterior and posterior convex epi-
physiodesis alone (four received Har-
rington instrumentation 2 to 4 years
later); a further nine patients were
treated with convex epiphysiodesis
and concurrent Harrington instru-
mentation. Radiographic progres-
sion, mirrored by a deteriorating
clinical result, occurred in all but
one patient. The best results were
noted with epiphysiodesis and si-
multaneous Harrington instrumen-
tation placement, which controlled
but did not improve the degree of the
primary curve.
Single Rod Instrumentation
Instrumentation without arthro-

subsequently modi-
fied Harrington’s technique by limit-
ing subperiosteal exposure to the
area of hook placement. The hook
sites were not fused. The rod itself
was placed subcutaneously. Moe et
al
34
also modified the Harrington rod
to have a smooth, thicker central
portion to prevent scar adhesion to
threads and to allow sagittal con-
touring. Patients were placed in a
Milwaukee brace postoperatively;
the construct was lengthened when
>10° loss of correction occurred.
There was an average increase of
2.9 cm in the length of the instru-
mented portion for all 20 patients
and of 3.8 cm (compared with a pre-
dicted growth of 4.5 cm) in the 9 pa-
tients who went on to fusion. Nota-
ble decrease in curve magnitude was
reported in the two patients with
progressive early onset curves. Com-
plications occurred in 50% of pa-
tients. Rod breakage, although less
common with the modified thicker
rod, still occurred. The authors also
reported hook dislodgement from

fusion technique.” Fourteen patients
underwent staged procedures in
which the vertebrae at each end of the
curve were initially fused. Five to 6
months later, hook placement was
performed along with subperiosteal
rod placement. Finally, at a third pro-
cedure 6 to 8 weeks later, the upper
hook was distracted. Serial lengthen-
ing was performed until definitive fu-
sion at maturity. Four of the 14 pa-
tients had completed treatment at the
time of publication of their report,
with “most satisfactory” results.
In 1977, Luque and Cardoso
37
re-
ported on their technique for segmen-
tal spinal instrumentation (SSI) with-
out ar throdesis. In 1982, Luque
38
reported the results of adding sublam-
inar wiring to a Harrington rod in 47
paralytic patients. The immobilized
area grew an average of 4.6 cm, with
curve correction of 78%. Smooth,
L-shaped rods were subsequently sub-
stituted for the Harrington rod in a
construct that became known as the
Luque trolley. Initial enthusiasm

tive analysis of Luque trolley instru-
mentation with and without convex
epiphysiodesis in 26 patients. Curve
deterioration was observed in all eight
patients treated with the Luque trol-
ley alone. In curves managed with
combined convex epiphysiodesis and
Luque instrumentation, the Cobb an-
gle worsened in 7 of 13 patients, was
unchanged in 4, and improved in 2.
Growth of the instrumented spinal
segment was 49% of the curve pre-
dicted in patients treated with the
Luque trolley alone, and 32% of the
curve predicted in patients undergo-
ing the combined procedure.
More recently, Blakemore et al
43
reported periodic lengthening with a
submuscular rod with and without
limited apical fusion in 29 children
with scoliosis. Ten of the curves
were idiopathic. The single rod was
placed within the muscle above the
spinal periosteum. This approach
placed the rod closer to the spine for
better contour and alignment with-
out causing spontaneous posterior
fusion. Apical fusion was performed
in curves >70° and in those that were

nique, building on concepts formu-
lated by Asher.
45
We currently prefer
this technique. Subperiosteal dissec-
tion is limited to upper and lower
anchor sites (foundations) (Figure 5).
Hooks or screws are placed on both
sides of the spine in so-called claw
patterns over two to three spinal lev-
els to avoid hook crowding. Pedicle
screws seem to add stability to the
construct.
46
A transverse rod con-
nector is positioned adjacent to or in
the middle of the claw constructs at
both foundations. Foundation sites
may be fused with local bone graft
supplemented with synthetic graft.
Upper and lower contoured 3/16-
inch-diameter rods are placed subcu-
taneously on both sides of the spine.
The rods are joined on each side with
extended tandem connectors placed
at the thoracolumbar junction to
avoid disturbing sagittal balance.
44
Lengthening is perfor med with a
distractor designed to fit within the

Volume 14, Number 2, February 2006 107
est follow-up. Growth of the T1-S1
segment averaged 1.21 cm per year.
The seven patients who completed
treatment at an average age of 10
years 3 months achieved a total T1-
S1 length increase of 11.8 cm from
preoperative status to postoperative
final fusion (1.66 cm growth per
year). In 14 patients with thoracic
curves, the space available for lung
ratio, as described by Campbell et
al,
48
improved from 0.87 preopera-
tively to normal (1.00) at latest
follow-up or after final fusion.
47
Complications occurred in 11 of 23
patients between initial surgery and
final fusion. The complications in-
cluded three anchor (hook or screw)
displacements, two rod breakages,
two deep wound infections, four
superficial wound problems, one
crankshaft, and one junctional ky-
phosis requiring an extension of in-
strumentation.
These results indicate that the
dual rod technique is safe and effec-

tial correction and maintenance of
correction and allowed more growth
than did single rod instrumentation.
Although the numbers were small,
the least favorable outcomes were in
patients who underwent short apical
fusion. This technique appeared to
lead to stiffening of the curve, crank-
shaft phenomenon, less correction,
and a higher incidence of complica-
tions. The authors indicated that
apical fusion may not be helpful.
49
Other Emerging
Techniques
Although still evolving, current
surgical techniques using instru-
mentation with minimal or no ar-
throdesis in the treatment of early
onset idiopathic scoliosis are capable
of significant initial curve correction
and prevention of subsequent curve
decompensation. This allows defin-
itive fusion to be delayed until ado-
lescence. In addition, it is possible to
preserve nearly normal growth
within the area of instrumentation.
It is hoped that growing rod instru-
mentation may be removed at matu-
rity in some patients, avoiding fu-

Radiographs and photographs of a girl aged 5 years 10 months old with infantile idiopathic scoliosis who was followed for 7
years and eventually had final fusion. Preoperative anteroposterior (A) and lateral (B) radiographs of the spine. Anteroposterior
(C) and lateral (D) radiographs taken after the initial surgery. Anteroposterior (E) and lateral (F) radiographs taken 5 years after
initial surgery. G and H, Clinical appearance at 5-year follow-up. Note the correction of scoliotic curve and the linear scar (G).
Patient kyphosis fell within normal limits (~50°) at the time of radiographs and clinical photographs.
Bruce L. Gillingham, MD, CAPT, MC, USN, et al
Volume 14, Number 2, February 2006 109
dures. The ideal implant would re-
quire minimally invasive insertion;
would be durable, rarely requiring re-
vision or replacement; would have a
minimal effect on adjacent tissue;
and, if required, would be easily in-
corporated into the definitive fusion.
One intriguing possibility is the
development of an implant that can
be lengthened by remote control. In
1998, Takaso et al
50
reported on the
development of a rod containing a
direct-current motor attached to a
radio-controlled receiver . Successful
serial correction of experimental
scoliosis was achieved in beagles.
The main problems encountered
were the relatively large size of the
rod (16 mm) and the receiver, which
necessitated placement in the ab-
dominal cavity.

48,52
This device was designed to manage
thoracic deformities resulting from
absent and fused ribs in congenital
and syndromic conditions, such as
Jeune’s asphyxiating thoracic dystro-
phy and Jarcho-Levin syndrome.
This effort has resulted in a broader
and deeper understanding of the cen-
tral role that the spine plays in the
architecture and function of the
chest wall and thorax. This new
awareness emphasizes the impor-
tance of evaluating thoracic volume
in addition to the standard assess-
ment of anteroposterior and sagittal
spinal alignment. Restoring this
“fourth dimension,” thereby maxi-
mizing the potential for pulmonary
development, is emerging as an im-
portant goal in the treatment of pa-
tients with congenital spine and rib
anomalies or with thoracic insuffi-
ciency syndrome.
48
Summary
Treating progressive early onset idio-
pathic scoliosis is challenging. Un-
treated curves may cause significant
disturbance of normal trunk and

Figure 7
Graphic representation of the improvement in length of implant, T1-S1 length, and
Cobb angle magnitude in the patient shown in Figure 6.5+10=5years 10
months; 13+6=13years 6 months.
Early Onset Idiopathic Scoliosis
110 Journal of the American Academy of Orthopaedic Surgeons
References
Evidence-based Medicine: Evidence-
based studies are identified in Ak-
barnia et al (reference 47, a multi-
center study on dual growing rods)
and Thompson et al (reference 49, a
comparison of single and dual grow-
ing rod techniques).
Citation numbers printed in bold
type indicate references published
within the past 5 years.
1. James JI, Lloyd-Roberts GC, Pilcher
MF: Infantile structural scoliosis.
J Bone Joint Surg Br 1959;41:719-735.
2. James JI: Idiopathic scoliosis: The
prognosis, diagnosis,and operativein-
dications related to curve patterns
and the age at onset. J Bone Joint
Surg Br 1954;36:36-49.
3. Dimeglio A: Growth of the spine be-
fore age 5 years. J Pediatr Orthop B
1993;1:102-107.
4. Dickson RA: Early-onset idiopathic
scoliosis, in Weinstein S (ed): The Pe-

J Bone Joint Surg Am 1980;62:863-
875.
12. Mau H: Does infantile scoliosis re-
quire treatment? J Bone Joint Surg
Am 1968;50:881.
13. Kattwinkel J, Brooks J, Keenan ME,
Malloy M: Infant sleep position and
sudden infant death syndrome (SIDS)
in the United States: Joint commen-
tary from the American Academy of
Pediatrics and selected agencies of the
Federal Government. Pediatrics
1994;93:820.
14. Mehta MH: The rib-vertebra angle in
the early diagnosis between resolving
and progressive infantile scoliosis.
J Bone Joint Surg Br 1972;54:230-243.
15. Muirhead A, Conner AN: The assess-
ment of lung function in children
with scoliosis. J Bone Joint Surg Br
1985;67:699-702.
16. Davies G, Reid L: Effect of scoliosis on
growth of alveoli and pulmonary ar-
teries and on right ventricle. Arch
Dis Child 1971;46:623-632.
17. Branthwaite MA: Cardiorespiratory
consequences of unfused idiopathic
scoliosis. Br J Dis Chest 1986;80:360-
369.
18. Kafer ER: Idiopathic scoliosis: Gas ex-

J Bone Joint Surg Br 1995;77:762-767.
25. Gollogly S, Smith JT, Campbell RM:
Determining lung volume with three-
dimensional reconstructions of CT
scan data: A pilot study to evaluate
the effects of expansion thoracoplasty
on children with severe spinal defor-
mities. J Pediatr Orthop 2004;24:
323-328.
26. Ferreira JH, de Janeiro R, James JI: Pro-
gressive and resolving infantile idio-
pathic scoliosis: The differential diag-
nosis. J Bone Joint Surg Br 1972;54:
648-655.
27. Gupta P, Lenke LG, Bridwell KH: In-
cidence of neural axis abnormalities
in infantile and juvenile patients with
spinal deformity: Is a magnetic reso-
nance image screening necessary?
Spine 1998;23:206-210.
28. Dobbs MB, Lenke LG, Szymanski DA,
et al: Prevalence of neural axis abnor-
malities in patients with infantile id-
iopathic scoliosis. J Bone Joint Surg
Am 2002;84:2230-2234.
29. Diedrich O, von Strempel A, Schloz
M, Schmitt O, Kraft CN: Long-term
observation and management of re-
solving infantile idiopathic scoliosis:
A 25-year follow-up. J Bone Joint

cence. Orthopaedic Transactions
1978;2:271.
37. Luque ER, Cardoso A: Treatment of
scoliosis without arthrodesis or exter-
nal support: A preliminary report.
Orthopaedic Transactions 1977;1:37.
38. Luque ER: Paralytic scoliosis in grow-
ing children. Clin Orthop 1982;163:
202-209.
39. Mardjetko SM, Hammerberg KW, Lu-
bicky JP, Fister JS: The Luque trolley
revisited: Review of nine cases requir-
ing revision. Spine 1992;17:582-589.
40. Rinsky LA, Gamble JG, Bleck EE: Seg-
mental instrumentation without fu-
sion in children with progressive
Bruce L. Gillingham, MD, CAPT, MC, USN, et al
Volume 14, Number 2, February 2006 111
scoliosis. J Pediatr Orthop 1985;5:
687-690.
41. Patterson JF, Webb JK, Burwell RG:
The operative treatment of progres-
sive early onset scoliosis: A prelimi-
nary report. Spine 1990;15:809-815.
42. Pratt RK, Webb JK, Burwell RG, Cum-
mings SL: Luque trolley and convex
epiphysiodesis in the management of
infantile and juvenile idiopathic scoli-
osis. Spine 1999;24:1538-1547.
43. Blakemore LC, Scoles PV, Poe-

2005;30:S46-S57.
48. Campbell RM, Smith MD, Mayes TC,
et al: The characteristics of thoracic
insufficiency syndrome associated
with fused ribs and congenital scolio-
sis. J Bone Joint Surg Am 2003;85:
399-408.
49. Thompson GH, Akbarnia BA, Kostial
P, et al: Comparison of single and dual
growing rod techniques followed
through definitive surgery: A prelim-
inary study. Spine 2005;30:2039-
2044.
50. Takaso M, Moriya H, Kitahara H,
et al: New remote-controlled grow-
ing-rod spinal instrumentation pos-
sibly applicable for scoliosis in young
children. J Orthop Sci 1998;3:336-
340.
51. Betz RR, Kim J, D’Andrea LP, Mulca-
hey MJ, Balsara RK, Clements DH: An
innovative technique of vertebral
body stapling for the treatment of pa-
tients with adolescent idiopathic
scoliosis: A feasibility, safety, and
utility study. Spine 2003;28:S255-
S265.
52. Campbell RM, Smith MD, Hell-
Vocke AK: Expansion thoracoplasty:
The surgical technique of opening-