SPINE Volume 34, Number 17, pp 1751–1755
©2009, Lippincott Williams & Wilkins
The Natural History of Congenital Scoliosis
and Kyphosis
David S. Marks, FRCS, FRCS (Orth),*† and Saeed A. Qaimkhani, FRCS (Orth)*
Study Design. Review article.
Objective. To discuss natural history of congenital sco-
liosis and kyphosis.
Summary of Background Data. Review of previously
published literature on natural history of congenital spine
deformities.
Methods. Medline and google search for congenital
scoliosis, kyphosis, and kyphoscoliosis, congenital spine
anomalies, deformities, and pathologies, and congenital
vertebral anomalies, deformities, and pathologies was
performed.
Results. Congenital vertebral anomalies have poten-
tial to progress and careful assessment and monitoring is
essential and early intervention may be desirable.
Conclusion. Congenital vertebral anomalies invariably
result from disturbed asymmetric growth and can have
serious consequences.
Key words: congenital, scoliosis, kyphosis, parapare-
sis, defects of formation, defects of segmentation. Spine
2009;34:1751–1755
At present more is known about the anomalies produc-
ing congenital scoliosis and kyphosis than the underlying
structural changes in the vertebra associated with any
other type of scoliosis. The embryological injury to the
developing somites that produces the vertebral anoma-
lies is well defined and the pathologic processes behind
these injuries are increasingly being identified. Recent
studies on the genetic causes have implicated a number of
genes, particularly those involved in the regulation of
somitogenesis.1 Advances in prenatal imaging have fur-
thered our detection of anomalies.2 Simplifying and
stratifying syndromic congenital anomalies has allowed
for more meaningful comparison of treatment regimes,
particularly relevant to the emergence of new techniques
of nonfusion of spine and rib instrumentation.3 Focus on
the associated congenital rib anomalies in recent years
has allowed a more global appreciation of the problem of
congenital deformity and opened up the potential for
exciting new treatment here.4
From the *Royal Orthopaedic Hospital and †Birmingham Children’s
Hospital, Northfield, Birmingham, England.
The manuscript submitted does not contain information about medical
device(s)/drug(s).
No funds were received in support of this work. No benefits in any
form have been or will be received from a commercial party related
directly or indirectly to the subject of this manuscript.
Address correspondence and reprint requests to David S. Marks,
FRCS, The Royal Orthopaedic Hospital & Birmingham Children’s
Hospital, Bristol Road South, Northfield, UK B31 2AP; E-mail: david.s.
marks@talk21.com
Congenital Scoliosis
Prior to Winter et al,5 authors who distinguished patients
with congenital scoliosis tended to conclude that this was
a relatively benign condition. With the study by Winter
et al came an appreciation for the first time that the
natural history of congenital vertebral anomalies was not
equal and within congenital scoliosis there was a range of
risk of progression dependent on factors such as type and
site of anomaly.
Analysis
It was the study of McMaster and Ohtsuka,6 that first
concentrated on the natural history of congenital scolio-
sis and determined in detail the risk of progression re-
lated to 4 key areas: the type of anomaly, the site within
the spine, the patients age at presentation, and whether
the curves are single or multiple.
We will consider these deformities in 4 groups, failure
of formation, failure of segmentation, mixed defects, and
complex unclassifiable pattern.
Failure of Formation
Block Vertebra. These are often multiple, have little
growth potential, and progress at a slow rate (less than
1° per annum). The site they occur at in the spine does
not affect the outcome but if numerous, they can stunt
trunk height.
Wedge Vertebra. These are concentrated in the lower
thoracic and thoracolumbar regions and exhibit a vari-
able but relatively slow rate of progression at 1° to 2° per
annum.
Simple Hemivertebra (Incarcerated/Unsegmented). These
have little or no growth potential and hence there is min-
imal risk of progression. Typically the curves are less
than 30° at maturity.
Semi Segmented/Fully Segmented and Multiple Hemivertebra.
The risk of deterioration in curves resulting from these
anomalies is particularly dependent on the site, degree of
segmentation, and number of each. It is therefore useful
to consider these in some detail.
Upper thoracic hemivertebra progress at an average of
1° to 2° per annum before the age of 10 and 2° to 2.5°
after that. They tend to produce cosmetic deformity and
shoulder imbalance.
Lower thoracic have a more rapid progression at 2°
per annum prepuberty and 2.5° to 3° afterwards.
In the thoracolumbar area, the rate of progression is
most rapid, ranging from 2° to 5° per annum prepuberty
1751
1752 Spine • Volume 34 • Number 17 • 2009
to 3.5° per annum following it and these curves have the
potential to produce significant trunk imbalance.
Lumbar hemivertebra have a slower rate of progres-
sion than their thoracic counterparts but nevertheless
still have rates of deterioration measuring about 1° per
annum. Puberty has little effect on the rate.
Lumbosacral hemivertebra are unique in that, al-
though they do not deteriorate at any greater rate than
lumbar ones, they are associated with a significant risk of
abnormal “take off” of the lumbar spine from the pelvis.
This can induce severe compensatory curves which be-
come fixed and decompensated.
For all lumbar/lumbosacral curves the degree of trunk
imbalance is determined by the growth potential.
Failure of Segmentation
Unilateral unsegmented bar: As with failure of forma-
tion, failure of segmentation defects varies in their po-
tential to cause deformity. The extent of the bar and the
site in the spine determine its nature.
In the upper thoracic spine, the rate of deterioration is
2° per annum before age often, 4° after. There is poten-
tial for significant shoulder elevation and head tilt to-
gether with the ability to produce long distal thoracic/
thoracolumbar curves which themselves can deteriorate
at twice the rate of the primary ones. In the lower tho-
racic region, the risk of progression is 5° per annum
before early puberty and 6.5° after it.
When encountered in the thoracolumbar region the
rates increase to 6° and 9°, respectively, but in the lum-
bar region these are about 5° per annum.
The clinical problem is the development of pelvic
obliquity and trunk imbalance, the magnitude and prob-
lems they cause being proportionate to the primary curve
and the structural secondary curves that evolve over
time.
Mixed Defects
Unilateral unsegmented bar and contralateral hemiver-
tebra: Most of these anomalies are located in the thoracic
spine; they are the most aggressive of all congenital sco-
liotic deformities with respect to their ability to deform.
The hemivertebra occur on the convex side and may be
multiple and remote from the bar. There is potential for
rapid deterioration (in excess of 14° per annum in some
thoracolumbar and lumbar cases) and the secondary
compensatory curves soon become fixed leading to trunk
shortening, apparent limb length discrepancy, and severe
cosmetic deformity.
Complex Unclassifiable Pattern
Complex mixed pattern anomalies do not follow any set
course and therefore are difficult to place within the
scheme of predicting progression untreated. Some have
significant potential to deform by asymmetrical growth
and tethering, whereas others produce marked loss of
height but little coronal or sagittal deformity. In general
the thoracolumbar apex curves do worse than the tho-
racic or lumbar.
Age at Presentation
Although by definition the anomaly is itself present from
the embryological period, the deformity it produces may
not be apparent for some considerable time after birth.
Curves which are evident before age 10 are in general
terms associated with a worse ultimate outcome due to
their potential for growth.
McMaster and Ohtsuka6 arbitrarily divided progres-
sion into 2 groups either side of age 10. This they did to
define a pre and post pubertal growth peak. They iden-
tified a mean rate of progression in deformities occurring
before age 10 as between 5° (for lower thoracic anoma-
lies) and 6° (for thoracolumbar) per annum. In the most
severe anomalies (unsegmented bar with contralateral
hemivertebra) the progression in childhood was up to 10°
per annum. Other authors have considered rates of pro-
gression in the first 6 years of life, however, the problem
with this is 2-fold; not all curves present before 10 and the
elapsed time to chart progress is relatively short. The au-
thors believe that the use of age ten in calculating progres-
sion risk is useful particularly in less aggressive curves.
As with other forms of scoliosis, deterioration in the
secondary and tertiary curves can also occur with growth.
Progression after skeletal maturity may result in the same
way even if the congenital abnormality remains stable.
Single or Multiple Curves
The majority of congenital scoliosis curves are single,
with secondary curves being initially compensatory and
ultimately becoming structural.
Where 2 or more congenital curves occur, the outcome
with respect to deformity is variable and depends primarily
on whether the convexities are unilateral or contra lateral.
If unilateral and remote, then the deformity may be pro-
found. If contralateral, then a balanced “kink” in the spine
may result with little clinical deformity.
Associated Congenital Anomalies
Many authors7–11 have sort to identify whether associ-
ated anomalies, such as those involving the genitouri-
nary tract, the Cardiovascular system or the Neurologic
Axis have any influence on the progression of untreated
congenital scoliosis. All have concluded that these asso-
ciated abnormalities do not adversely influence the be-
havior of the vertebral abnormalities. They may how-
ever, affect our ability to treat the congenital curves and
this should be borne in mind when considering where
and when to intervene surgically.
Congenital Kyphosis and Kyphoscoliosis
Congenital kyphosis is a sagittal plane deformity character-
ized by abnormal posterior convex angulation of a segment
of spine. Congenital kyphosis or kyphoscoliosis results
from developmental vertebral anomalies that impair longi-
tudinal growth anterior or anterolateral to the transverse
axis of vertebral rotation in the sagittal plane.12
Von Rokitansky13 in 1844 and Schulthess,14 in 1905,
described the typical appearances of thoracolumbar ky-
phosis without any scoliosis. Since then isolated case re-
 Congenital Scoliosis and Kyphosis • Marks and Qaimkhani 1753
ports appeared with moderate frequency, particularly in
German and French literature. Greig15 in 1916 was the
first to publish a case report in English literature and
described a case of congenital absence of the body of first
thoracic vertebra. Paraplegia was first mentioned as a
complication of congenital kyphosis by Lombard and Le
Genissel16 in 1938. Van Schrick17 in 1932 was the first to
differentiate 2 types of deformities as Group I (failure of
segmentation) and Group II (absence of vertebral body).
In 1955, James18 reviewed 16 patients with kyphosis at
thoracolumbar junction and 5 with the defect in upper
thoracic area. Three of the 5 with upper thoracic defects
developed paraplegia compared to 2 of 16 with thoraco-
lumbar curves. The onset of paralysis in 4 of the 5 pa-
tients was in later years of growth. This led James to issue
a strong plea for early intervention in the form of poste-
rior fusion to prevent this catastrophe.
Studies by Winter et al19 and McMaster and Singh12
have evaluated the natural history of congenital kyphosis
and kyphoscoliosis in great detail.
Analysis
Congenital kyphotic deformities are less common than con-
genital scoliotic deformities, but can have serious conse-
quences if left untreated. Paraplegia is more common with
failure of formation which gives a sharp angular kypho-
sis. It is also associated with kyphosis in the upper tho-
racic area as this is the part of the spinal cord with poor-
est collateral circulation, the so called “watershed” area
of the blood supply of the spinal cord. Paraplegia may
occur early but is more common during the adolescent
growth spurt with rapid increase in the untreated kypho-
sis deformity and is described as occurring after minor
trauma.19
The vertebral anomalies are classified as per the sco-
liotic ones: failure of formation, failure of segmentation,
mixed pattern, and unclassifiable. The failure of forma-
tion can be purely anterior, resulting in pure kyphosis, or
anterolateral resulting in kyphoscoliosis and these may
be multiple.
Winter et al19 classified these congenital anomalies on
the same lines as earlier described by VanSchrick17 and
also used by Lombard and Le Génissel.16
Type I
Congenital Failure of Vertebral Body Formation (Cen-
trum Hypoplasia and Aplasia). There are 5 commonly
recognized forms of centrum deficiencies.20 In order of
ascending severity in the generation of deformity if left
untreated, these are:
Wedge: Posterior hemicentrum (posterior hemivertebra).
Lateral hemicentrum: This probably results from uni-
lateral lack of vascularization.
Posterior quadrant Centrum (posterolateral quadrant
vertebra): Approximately a quarter of the centrum
proceeds to ossify in 1 posterior quarter. The con-
tralateral neural arch elements are normal in appear-
ance or somewhat hypoplastic but not absent. This
anomaly is notorious for causing sharply angulated
kyphoscoliosis.
Centrum aplasia (complete absence of vertebral body):
This is the most severe form of anterior deficit. Pedicle
roots persist at the base of the pedicles but they are not
part of the centrum.
Another type initially described by Von Rokitansky13
and further detailed by Fischer and Vandemark21 is but-
terfly (sagittal cleft) vertebra which consists of a partial
or complete failure of formation of the anterior and cen-
tral portions of the vertebral body, leaving 2 posterolat-
eral fragments of bone attached to the neural arch. These
residual pieces of bone are wedged anteriorly and medi-
ally and are separated by a sagittal cleft.
Type II
Congenital Failure of Vertebral Body Segmentation
(Anulus Fibrosus Osseous Metaplasia). The outer layer
of anulus fibrosus retains growth potential in the peri-
chondrium and later the periosteum. At cessation of
bone growth, ossified anulus bridges 2 or more centra.
The bony transformation of the fibrous anulus is called
osseous metaplasia of this structure.20 Clinical detection
is difficult until ossification of the defective anulus occurs
in childhood.
Type III
Mixed. McMaster and Singh12 added the subgroups of
butterfly (sagittal cleft) and anterior and anterolateral
wedged vertebrae and differentiated between an anterior
and anterolateral unsegmented bar. McMaster also
added another type to already existing 3 types:
Type IV
Unclassifiable Anomalies. Winter et al19 observed thirty
patients without surgical treatment for an average of 6
years and the average amount of progression during this
time was 44° or about 7° per annum. Progression was
more rapid during the adolescent growth spurt. They
could not determine the relative amounts of progression
in the 3 types as there were very few cases of Type II and
Type III deformity. Type I lesions tended to produce
sharp angular kyphosis while Type II lesions caused long
sweeping curves. Most Type I deformities (80/86) were
in the thoracic or thoracolumbar area and were the most
severe and progressed most rapidly. Most Type II defor-
mities (14/19) were in the upper lumbar or thoracolum-
bar area. Although most of these were progressive, they
did not produce the severe deformities associated with
Type I defects. Most Type III deformities (12/18) were in
thoracolumbar region. These deformities progressed
more rapidly and became more severe than those associ-
ated with Type II lesions but were not as severe as those
associated with Type I lesions.
1754 Spine • Volume 34 • Number 17 • 2009
In series by Winter et al, Paraplegia occurred only in
association with Type I lesions especially those in upper
part of the thoracic spine. Of the 86 Type I lesions, 16
had paraplegia, 11 with high thoracic and 5 with thora-
columbar deformities. The paralysis was not seen at
birth and appeared to be related to either growth, or
increase in curve, or both. The usual time for onset of
paralysis was during the preadolescent growth spurt, oc-
curring spontaneously or in relation to minor trauma.
Although the age at onset of paralysis ranged from 4 to
19; the average age was 12 years. Once the paralysis
developed, it was always progressive although the rate of
progression and the severity of paralysis varied.
In McMaster and Singh’s12 study of 112 patients, 63
were observed untreated before skeletal maturity for a
mean of 6 years and 6 months and 41 were untreated at
skeletal maturity. Sixty-one percent had a Type I kypho-
sis, 21% had a Type II kyphosis, and 11% had a Type III
kyphosis. In 7% the kyphosis could not be classified be-
cause the patients were seen untreated at a stage when the
deformity was so severe that it made it impossible for the au-
thors to classify the vertebral anomaly precisely. The apex of
the kyphosis was more frequent between the 10th tho-
racic and the first lumbar vertebra. They found that pro-
gression of kyphosis or kyphoscoliosis continued through-
out growth and usually accelerated during the adolescent
growth spurt (after the age of 10 years) before stabilizing at
skeletal maturity.
The most common pattern of failure of vertebral body
formation in McMaster and Singh’s12 series, causing a
kyphosis or kyphoscoliosis, was posterolateral quadrant
vertebra (seen in 35% of cases). Other Type I anomalies
included butterfly (or sagittal cleft) in 13%, a posterior
hemivertebra in 7%, and anterior or anterolateral
wedged vertebra in 5%. A kyphosis due to posterolateral
quadrant vertebra progressed at a median rate of 2.5° per
annum before the age of 10 years and 5° per annum
thereafter. Ten patients had a spinal procedure at a mean
age of 12 years and 8 months when the mean kyphosis
was 81°. Kyphosis resulting from posterior hemivertebra
had only a slightly better prognosis, with those due to a
butterfly vertebra and a wedge vertebra being the least
aggressive. A kyphosis due to 2 adjacent vertebral anom-
alies progressed more rapidly and produced a more se-
vere deformity than did a similar single deformity.
Patients with Type II kyphosis in McMaster and
Singh’s12 series had an unsegmented bar of bone extend-
ing anteriorly across the intervertebral disc spaces, join-
ing a mean of 3.5 vertebrae. These patients had variable
prognosis depending on whether the unsegmented bar
lay symmetrically in the sagittal plane or anterolaterally.
Of the patients, in 13% the bar did lay symmetrically
producing pure kyphosis and this usually progressed rel-
atively slowly (at a median rate of 1° per annum before
the age of 10 years). Only 2 patients who were followed
after the age of 10 years had a kyphosis that had pro-
gressed at a rate more than 2° per annum. Nine patients
were untreated at skeletal maturity and at that time a
mean kyphosis was 64°. An anterolateral bar producing
a kyphoscoliosis occurred in only 8% but had a much
worse prognosis. In 5 patients who were seen untreated
at skeletal maturity, the mean kyphosis was 92°.
A Type III kyphoscoliosis due to anterolateral unseg-
mented bar combined with contra lateral posterolateral
quadrant vertebrae in McMaster and Singh’s12 series
was least common occurring in 11% of patients but it
usually progressed most rapidly and produced the most
severe deformity. This type of kyphosis progressed at a
median rate of 5° per annum before the age of 10 years
and 8° per annum thereafter. Seven patients had arthro-
desis of the spine at a mean age of 13 years when the
kyphosis was 92°.
In McMaster and Singh’s12 study a progressive spastic
paraparesis of the lower limbs due to anterior compres-
sion of the spinal cord at the apex of the congenital ky-
phosis or kyphoscoliosis occurred spontaneously in 10%
(11/112) patients, all of whom were neurologically nor-
mal previously. Seven of these patients had a Type I
anomaly and 4 patients had anomalies that could not be
classified because of the severity of the angular kyphosis.
Neurologic complications did not occur in recogniz-
able Type II anomalies because they produced a smooth
kyphosis in which the abnormal vertebrae were stabi-
lized by the anterior failure of segmentation. However, a
posterolateral quadrant vertebra and a posterior hemi-
vertebra were unstable and tended to extrude backward
into the spinal canal, causing anterior compression of the
spinal cord at the apex of the angular deformity as it
became more severe. The apex of the kyphosis in major-
ity of the patients who developed neurologic complica-
tions was in the middle or caudad thoracic regions,
where the diameter of the spinal canal is narrowest and
the spinal cord has relatively poor blood supply.
The onset of neurologic deterioration occurred be-
tween the ages of 8 and 11 years in 4 patients, between
the ages of 14 and 18 years in 6, and at 28 years in 1. The
mean size of kyphosis at the onset of paraparesis was
111° but 1 patient who had a posterolateral quadrant
vertebra at the ninth thoracic level had only a 60 degree
kyphosis. Occult congenital intraspinal anomaly was
found only in 1 of 51 patients who had myelograms or
magnetic resonance imaging.
Conclusion
Congenital vertebral anomalies invariably result from
disturbed asymmetric growth and the deformities they
produce are essentially “local” spinal problems. They
have however, the potential to produce a profound
“global” effect on the spine.
The key to management of these anomalies is a thor-
ough analysis of the type, site, and potential for growth
which, when coupled with a knowledge of the potential
outcome of the untreated anomaly, will allow for formu-
lation of a treatment strategy. This may dictate observa-
tion (if appropriate) and/or timely surgical intervention
(either ablating or modifying growth). It is vital to have a
 Congenital Scoliosis and Kyphosis • Marks and Qaimkhani 1755
sound knowledge of the potential for neurologic injury
with kyphotic deformities, a feature not seen with their
coronal plane equivalents.
Determining the evolution of a particular patient’s
deformity cannot occur without an understanding of the
natural history of the various vertebral abnormalities.
To fail to appreciate the potential for deterioration
and hence miss the opportunity to intervene and mod-
ify the growth of these anomalies to prevent deformity
is inappropriate. This can lead to potentially cata-
strophic neurologic complications which otherwise
may be preventable.
Key Points
● Congenital scoliosis
Failure of formation: The risk of progression is re-
lated to type of anomaly, site within the spine, pa-
tient’s age, and number of curves. Scoliosis with
upper thoracic hemivertebra progress between 1°
and 2.5° per annum and produce cosmetic defor-
mity and shoulder imbalance while curve with
lower thoracic hemivertebra have more rapid pro-
gression between 2° and 3° per annum. Most rapid
progression is seen in thoracolumbar area.
Failure of segmentation: The risk of progression of
scoliosis is related to extent of unsegmented bar
and its site in the spine. Rate of deterioration is
between 2° and 6.5° per annum in thoracic spine,
between 6° and 9° in thoracolumbar region and
about 5° per annum in lumbar region.
Mixed defects: Most aggressive of all congenital
scoliosis with potential for rapid progression.
● Congenital kyphosis
Failure of formation: Most of these kyphotic defor-
mities are in the thoracic or thoracolumbar area
and produce sharp angular kyphosis and progress
between 2.5° and 5° per annum.
Failure of segmentation: An anterolateral bar oc-
curs rarely but has much poor prognosis than an
anterior bar.
Mixed: These lesions progress most rapidly and
produce the most severe deformity.
Progressive spastic paraparesis: This occurred
spontaneously in 10% patients with kyphosis most
of whom were either failure of formation defects or
unclassifiable particularly if apex was in middle or
caudad thoracic regions. Neurologic complications
did not occur in pure failure of segmentation de-
fects. Early surgical intervention is needed to pre-
vent this serious complication.
References
1. Erol B, Kusumi K, Lou J, et al. Etiology of congenital scoliosis. UPOJ 2002;
15:37– 42.
2. Kalache KD, Bamberg C, Proquitté H, et al. Three-dimensional multi-slice
view: new prospects for evaluation of congenital anomalies in the fetus. J
Ultrasound Med 2006;25:1041–9.
3. Kusumi K, Turnpenny PD. Formation errors of the vertebral column. J Bone
Joint Surg 2007;89-A(suppl 1):64 –71.
4. Campbell RM, Smith MD, Mayes TC, et al. The characteristics of thoracic
insufficiency syndrome associated with fused ribs and congenital scoliosis.
J Bone Joint Surg 2003;85-A:399 – 408.
5. Winter RB, Moe JH, Eilers VE. Congenital scoliosis: a study of 234 patients
treated and un-treated. J Bone Joint Surg 1968;50-A:1– 47.
6. McMaster M, Ohtsuka K. The natural history of congenital scoliosis: a study
of 251 patients. J Bone Joint Surg 1982;64-A:1128 – 47.
7. Reckles LN, Peterson HA, Weidman WH, et al. The association of scoliosis
and congenital heart defects. J Bone Joint Surg 1975;57-A:449 –55.
8. Farley FA, Phillips WA, Hersenberg JE, et al. Natural history of scoliosis in
congenital heart disease. J Paediatr Orthop 1991;11:42–7.
9. MacEwen GD, Winter RB, Hardy JH. Evaluation of kidney anomalies in
congenital scoliosis. J Bone Joint Surg 1972;54:1451– 4.
10. Hensinger RN, Lang JE, MacEwen GD. Klippel–Feil syndrome: a constella-
tion of associated anomalies. J Bone Joint Surg 1974;56:1246 –53.
11. Basu PS, Elsebaie H, Noordeen MH. Congenital spinal deformity: a compre-
hensive assessment at presentation. Spine 2002;27:2255–9.
12. McMaster MJ, Singh H. Natural history of congenital kyphosis and kypho-
scoliosis: a study of one hundred and twelve patients. J Bone Joint Surg
1999;81:1367– 83.
13. Von Rokitansky KF. Handbuch der Pathologischen Anatomie. Vol 2. Vi-
enna, Austria: Braumüller und Seidel; 1844.
14. Schulthess W. Die Klinische Pathologie der RÜckgratsverkrÜmmungen. In:
Joachimsthal G, eds. Handbuch der Orthopädischen Chirurgie. Vol 2. Jena,
Louisiana: Veriag von Gustav Fischer, 1905:903.
15. Greig DM. Congenital kyphosis. Edinb Med J 1916;16:93–99.
16. Lombard P, Le Génissel. Cyphoses congénitales. Rev Orthp 1938;25:
532–50.
17. Van Schrick FG. Die angerborene Kyphose. Zeitschr Orthop Chir 1932;56:
238 –59.
18. James JI. Kyphoscoliosis. J Bone Joint Surg Br1955;37:414 –26.
19. Winter RB, Moe JH, Wang JF. Congenital kyphosis: its natural history and
treatment as observed in a study of one hundred and thirty patients. J Bone
Joint Surg 1973;55-A:223–74.
20. Tsou PM. Embryology of congenital kyphosis. Clin Orthop 1977;128:18 –25.
21. Frederick JF, Vandemark RE. Sagittal cleft (butterfly) vertebra. J Bone Joint
Surg 1975;27:695– 8.