1. Copyright 1997 by The Journal of Bone and Joint Surgery, Incorporated
The Prevalence of Back Pain in Children
Who Have Idiopathic Scoliosis*
BY NORMAN RAMIREZ, M.D.t, CHARLES E. JOHNSTON, II, M.DJ, AND RICHARD H. BROWNE, PH.D.*, DALLAS, TEXAS
Investigation performed at the Texas Scottish Rite Hospital for Children, Dallas
ABSTRACT: A retrospective study of 2442 patients
who had idiopathic scoliosis was performed to deter-
mine the prevalence of back pain and its association
with an underlying pathological condition. Five hun-
dred and sixty (23 per cent) of the 2442 patients had
back pain at the time of presentation, and an additional
210 (9 per cent) had back pain during the period of
observation. There was a significant association be-
tween back pain and an age of more than fifteen years,
skeletal maturity (a Risser sign of 2 or more), post-
menarchal status, and a history of injury. There was
no association with gender, family history of scoliosis,
limb-length discrepancy, magnitude or type of curve, or
spinal alignment. At the latest follow-up evaluation,
324 (58 per cent) of the 560 patients who had had back
pain at presentation had no additional symptoms.
Forty-eight (9 per cent) of the 560 patients who had
back pain had an underlying pathological condition:
twenty-nine patients had spondylolysis or spondylo-
listhesis, nine had Scheurmann kyphosis, five had a
syrinx, two had a herniated disc, one had hydromyelia,
one had a tethered cord, and one had an intraspinal
tumor. A painful left thoracic curve or an abnormal
neurological finding was most predictive of an under-
lying pathological condition, although only eight of the
thirty-three patients who had such findings were found
to have such a condition.
When a patient with scoliosis has back pain, a care-
ful history should be recorded, a thorough physical
examination should be performed, and good-quality
plain radiographs should be made. If this initial evalu-
ation reveals normal findings, a diagnosis of idiopathic
scoliosis can be made, the scoliosis can be treated ap-
propriately, and non-operative treatment can be initi-
ated for the back pain. It is not necessary to perform
extensive diagnostic studies to evaluate every patient
who has scoliosis and back pain.
Idiopathic scoliosis is usually considered to be pain-
less, as painful causes of spinal deformity have, by defi-
*No benefits in any form have been received or will be received
from a commercial party related directly or indirectly to the subject
of this article. No funds were received in support of this study.
tP.O. Box 990, Mayaquez, Puerto Rico 00681.
tTexas Scottish Rite Hospital for Children, 2222 Welborn Street,
Dallas, Texas 75219-3993.
nition, not been termed idiopathic. As far as we know,
the prevalence of back pain in children and adoles-
cents who have idiopathic scoliosis has not been es-
tablished. If the association of pain and scoliosis with
serious etiologies is indeed strong, as has been demon-
strated by studies of small populations and by anecdotal
reports23
''215
'2027
, then substantial back pain suggests a
non-idiopathic diagnosis.
It has been our impression that pain in the back is
a common symptom in children and adolescents who
are evaluated for spinal deformity. This investigation
was performed to determine the prevalence of back
pain and underlying pathological conditions in patients
who have idiopathic scoliosis and to identify factors that
can be used to predict the presence of such conditions.
Materials and Methods
The records of 3104 consecutive patients who were
seen in scoliosis clinics between 1980 and 1990 were
reviewed. Two thousand four hundred and forty-two
were either referred for or diagnosed as having idio-
pathic scoliosis, and these patients make up the present
study. Of the remaining 662 patients, 321 were excluded
because they had obvious non-idiopathic scoliosis and
341 were excluded because they did not have scoliosis
as defined by the Scoliosis Research Society (a Cobb
angle6
of 10 degrees or more).
We recorded the age at presentation and at the on-
set of the back pain, duration and severity of symp-
toms, gender, menarchal status, whether the pain was
localized or radiated, history of pain at night, activi-
ties associated with back pain, family history of scolio-
sis, history of injury, duration of follow-up, treatment,
results of diagnostic and imaging studies, resolution
of the pain, neurological findings, limb-length discrep-
ancy, shift in the alignment of the trunk, and presence
of deformities of the thoracic and lumbar ribs. Postero-
anterior and lateral radiographs of the spine were re-
viewed in order to determine the type of curve, the
Cobb angle6
, and the Risser sign25
. Maturity was de-
fined on the basis of the Risser sign: patients who had
a Risser sign of 0 or 1 were considered to be immature
and those who had a sign of 2 to 5, mature. Additional
diagnostic studies, such as blood tests, bone-scanning,
computerized tomography, magnetic resonance imag-
ing, electrodiagnostic studies, and myelography, were
364 THE JOURNAL OF BONE AND JOINT SURGERY

2. THE PREVALENCE OF BACK PAIN IN CHILDREN WHO HAVE IDIOPATHIC SCOLIOSIS 365
performed on the basis of the clinical findings.
The mean duration of follow-up was 3.1 years for all
2442 patients. The mean duration of follow-up was 3.2
years for the patients who did not have back pain and
2.7 years for the patients who did (p < 0.001).
Chi-square analysis was used to assess the associa-
tion between the prevalence of back pain and other
factors. The two-sample t test was used to compare the
means and to evaluate the association between under-
lying disease and other factors. Because of the large
number of patients, we expected to find a p value of less
than 0.05 between rates. When a non-significant result
was obtained, the p value was reported as it indicated
the absence of a substantial difference in rates. We be-
lieve that the value of the data is related to the clinical
importance of the finding rather than to the statistical
significance.
Results
At the time of presentation, 560 (23 per cent) of the
2442 patients who were presumed to have idiopathic
scoliosis reported a history of back pain; another 210
patients (9 per cent), who were managed with observa-
tion only and had not had pain initially, subsequently
reported back pain during the follow-up period. Of the
560 patients who had back pain initially, 174 were seen
primarily for the pain and were first diagnosed as having
scoliosis at the initial visit, 188 reported a history of
occasional back pain after the diagnosis of scoliosis was
made, and 198 reported frequent and constant back pain
after being diagnosed with scoliosis at another institu-
tion before our evaluation. Several factors were evalu-
ated to determine their value in the prediction of back
pain.
Age: The mean age of the 560 patients who had
back pain at the time of presentation was fourteen years
(range, six to twenty years), whereas the mean age of
the patients who did not have pain was thirteen years
(range, nine months to twenty-two years). The mean age
at the time of the onset of the pain was thirteen years
(range, three to nineteen years), and the mean dura-
tion of the pain before our evaluation was nine months
(range, three months to seven years). Pain was reported
at presentation by none of the fifteen patients who were
five years old or less, twenty-three (15 per cent) of the
152 patients who were six to ten years old, 388 (21 per
cent) of the 1815 patients who were eleven to fifteen
years old, and 149 (32 per cent) of the 460 patients who
were more than fifteen years old. This difference among
the prevalences was significant (p < 0.001).
Maturity: Four hundred and three (29 per cent) of
the 1392 mature patients and 157 (15 per cent) of the
1050 immature patients had back pain. There was a
significant association between maturity and the preva-
lence of back pain (p < 0.001).
Gender: Two thousand and fifty-one (84 per cent) of
the 2442 patients were female, and 485 (24 per cent) of
the female patients had back pain initially. Seventy-five
(19 per cent) of the 391 male patients had back pain
initially. The trend toward female preponderance was
not significant (p = 0.066).
Menarchal status: Of the 2051 female patients, 1293
(63 per cent) were postmenarchal. Three hundred and
sixty-three (28 per cent) of the 1293 postmenarchal pa-
tients were seen for back pain, compared with 122 (16
per cent) of the 758 premenarchal patients. This differ-
ence was significant (p < 0.001).
Associated activities: The back pain was associated
with sports activities in 199 patients (36 per cent) and
with sedentary activities in ninety-three patients (17 per
cent). It was not associated with any activity in the re-
maining 268 patients (48 per cent).
Family history: One hundred and nineteen patients
(21 per cent) who had back pain at presentation had a
family history of scoliosis, as compared with 392 (21 per
cent) of the 1882 patients who did not have pain (p =
0.782). A family history of scoliosis did not increase the
likelihood of back pain.
History of injury: Twenty-seven patients (5 per cent)
who had back pain initially and twenty-one patients (1
per cent) who did not have pain had a history of injury
(p = 0.018).
History of pain at night: Thirty-three patients (6 per
cent) had pain at night. Of these, four had an underlying
pathological condition.
Limb-length discrepancy: Thirty-seven patients (7
per cent) who had back pain and 115 patients (6 per
cent) who did not have back pain had a limb-length
discrepancy of more than 1.5 centimeters. There was no
significant association between limb-length discrepancy
and the prevalence of back pain (p = 0.704).
Alignment of the trunk: There was a shift in the
alignment of the trunk (a plumb line from the seventh
cervical vertebra fell more than one centimeter lat-
eral to the gluteal cleft) in 220 patients. This shift was
seen in sixty-two (11 per cent) of the 560 patients who
had back pain and in 158 (8 per cent) of the 1882 pa-
tients who did not. The difference was not significant
(p = 0.052).
Abnormal neurological findings: Forty-seven (2 per
cent) of the 2442 patients had findings such as tightness
of the hamstrings, pes cavus, absent abdominal reflexes,
or dermatomal hypoesthesia at the time of the initial
examination. Twenty of these forty-seven patients (4
per cent of 560) had back pain and twenty-seven (1 per
cent of 1882) did not (p < 0.001). However, only five of
the twenty patients who had back pain and abnormal
neurological findings at the initial evaluation (the most
likely combination of clinical findings to be associated
with underlying disease) were found to have an under-
lying etiology for the back pain.
Type of curve: Eighteen of the seventy-seven left
thoracic curves, 165 of the 725 right thoracic curves,
sixty-six of the 288 left lumbar curves, eighteen of the
VOL. 79-A, NO. 3, MARCH 1997

3. 366 NORMAN RAMIREZ, C. E. JOHNSTON, II, AND R. H. BROWNE
TABLE I
OCCURRENCE OF BACK PAIN
ACCORDING TO THE TYPE OF CURVE
Type of Curve
Left thoracic
Right thoracic
Left lumbar
Right lumbar
Thoracolumbar
Double thoracic
Left thoracic and right lumbar
Right thoracic and left lumbar
Total
No.
77
725
288
80
390
73
77
732
2442
No. (Percentage)
with Pain*
18 (23)
165 (23)
66 (23)
18 (23)
94 (24)
15 (21)
23 (30)
161 (22)
560 (23)
*The difference in the prevalence of pain among the types of
curves was not significant (p = 0.882, chi-square test).
eighty right lumbar curves, ninety-four of the 390 thora-
columbar curves, fifteen of the seventy-three double
thoracic curves, twenty-three of the seventy-seven left
thoracic and right lumbar curves, and 161 of the 732
right thoracic and left lumbar curves were associated
with pain. The chi-square test showed no association
between the type of curve and the occurrence of back
pain (p = 0.882) (Table I).
Magnitude of the curve: The mean Cobb angle6
at
presentation for the different types of curves ranged
from 20 to 34 degrees (Table II). The two-sample t test
showed no association between the magnitude of the
curve and the occurrence of back pain at the time of
presentation (p > 0.05 for all curve types).
Etiology of pain: Forty-eight patients (9 per cent
of the 560 patients who had a history of back pain and
2 per cent of the 2442 patients who initially were pre-
sumed to have idiopathic scoliosis) were found to have
an underlying etiology for the back pain. The etiol-
ogy was spondylolysis or spondylolisthesis in twenty-
nine patients; Scheuermann kyphosis in nine; a syrinx
in five; a herniated disc in two; and hydromyelia, an
intraspinal tumor, and a tethered cord in one each.
There was no significant association between the prev-
alence of underlying disease and the age at the initial
evaluation, age at the onset of the back pain, duration
of the symptoms, gender, menarchal status, pain at
night, family history of scoliosis, limb-length discrep-
ancy, shift in the alignment of the trunk, or magnitude
of the curve.
Diagnostic studies: The underlying disease was iden-
tified on plain radiographs for thirty-one of the forty-
eight patients. Magnetic resonance imaging, which was
performed for fifty-four patients, revealed a syrinx in
three and a herniated disc in two of twenty patients who
had associated neurological findings, a syrinx in two and
a tumor in one of thirteen patients who had back pain
and a left thoracic curve without associated neurological
findings, a tethered cord in one of four patients who had
progressive scoliosis, and hydromyelia in one of seven-
teen patients who had back pain of long duration. A
bone scan was made for forty-eight patients. Five had
the scan because of pain associated with a left thoracic
curve without neurological abnormality, and all five had
negative findings. Forty-three patients had the scan
because of back pain of long duration. Of these forty-
three scans, six revealed spondylolisthesis and one,
spondylolysis. Six patients had a computerized tomog-
raphy scan and seven had a myelogram; all of these
studies were normal. One patient who had normal find-
ings on the myelogram was found to have a syrinx on
magnetic resonance imaging.
Treatment: Three hundred and ninety-two of the
560 patients who had had back pain initially were man-
aged with observation for the scoliosis and exercises
and medication for the pain. At the latest follow-up
examination, 208 (53 per cent) of these patients had
no back pain. One hundred and twelve patients had
an operation to prevent progression of the curve, and
seventy-nine (71 per cent) of them had relief of the
back pain after the operation. Fifty-six patients were
managed with a brace to prevent progression of the
curve, and thirty-seven (66 per cent) of them reported
relief of the pain. Thus, 116 (69 per cent) of the 168
patients who were managed with either a brace or an
TABLE II
BACK PAIN AND MAGNITUDE OF THE CURVE
Type of Curve
Left thoracic
Right thoracic
Left lumbar
Right lumbar
Thoracolumbar
Double thoracic
Left thoracic and right lumbar
Right thoracic and left 1umbar
No.
18
165
66
18
94
15
23
161
Pain at Presentation
Magnitude of Curve6
(Degrees)
Mean
20
28
21
20
24
25
24
34
Range
10-50
10-86
10-58
10-65
10-65
14-46
10-52
10-82
No.
59
560
222
62
296
58
54
571
No Pain at Presentation
Magnitude
Mean
20
28
20
21
23
27
26
33
of Cur'ver
' (Degrees)
Range
10-68
10-122
10-68
10-63
10-75
11-56
10-53
10-105
P Value*
0.96
0.89
0.52
0.76
0.45
0.38
0.64
0.43
*As determined with the two-sample t test.
THE JOURNAL OF BONE AND JOINT SURGERY

4. THE PREVALENCE OF BACK PAIN IN CHILDREN WHO HAVE IDIOPATHIC SCOLIOSIS 367
operation had relief of the pain, whereas only 208
(53 per cent) of the 392 patients who were managed
with observation had resolution of the pain (p < 0.001).
It is possible that stabilization of a progressive curve
relieves back pain resulting from altered spinal bio-
mechanics secondary to scoliotic deformity. At the lat-
est follow-up examination, 211 (52 per cent) of the
403 mature patients had resolution of the back pain,
compared with 113 (72 per cent) of the 157 imma-
ture patients. Maturity appeared to significantly de-
crease the likelihood that back pain would resolve
(p < 0.001).
Discussion
Back pain in children has been regarded as a rare
phenomenon and is associated with serious underlying
disease. The differential diagnosis of back pain21
in chil-
dren usually includes conditions such as spondylolysis
or spondylolisthesis; traumatic, infectious, systemic, in-
flammatory, and neoplastic causes; and conversion dis-
order451116
"'242
'J
. Idiopathic scoliosis has been discounted
as a cause of back pain because scoliosis in children
is considered to be painless. However, the true preva-
lence of back pain in idiopathic scoliosis is essentially
unknown. It is commonly accepted that every child who
has scoliosis and reports back pain should be examined
thoroughly, especially if the patient has a painful left
thoracic curve3828
. However, Schwend et al.2f
' showed
that imaging studies are not necessary for every child
who has a left thoracic curve. Appropriate imaging stud-
ies are indicated for selected patients7
-9131430
. Some au-
thors2027
have recommended a bone scan, on the basis
of the belief that the most common cause of pain with
scoliosis is the osteoid osteoma and osteoblastoma com-
plex. However, in the present study, the most common
underlying pathological conditions were spondylolysis
and spondylolisthesis, followed by Scheuermann ky-
phosis, lesions of the spinal cord, a herniated disc, and
tethered cord. None of the patients had an osteoid os-
teoma. The underlying disease was identified on con-
ventional radiographs for thirty-one of the forty-eight
patients.
Several investigators have attempted to establish
the prevalence of back pain in children, but it is difficult
to compare the results because of the methodological
differences. For example, Fairbank et al.9
reported back
pain in 115 (26 per cent) of 446 students (of different
ages) who were evaluated with use of a questionnaire
only. Mierau et al.22
considered a positive straight-leg-
raising test to be indicative of back pain and found a 23
per cent (sixty-one of 267) prevalence in a group of
young children and a 33 per cent (forty-five of 135)
prevalence in a group of adolescents. Balague et al.1
, also
using a positive straight-leg-raising test as the crite-
rion, found back pain in 572 (33 per cent) of 1715
students; however, only 14 per cent (eighty) of them
sought medical advice and only 1 per cent (six) had
radiographs or treatment. Turner et al.29
found that only
2 per cent of all children referred to their institution
were referred for back pain. In our study, 560 (23 per
cent) of the 2442 patients reported a history of back
pain at the initial examination, and 210 additional pa-
tients reported back pain during the period of follow-up
for the scoliosis. If these latter patients are included, the
over-all prevalence of back pain in association with pre-
sumed idiopathic scoliosis followed without treatment
is 32 per cent, which is similar to the prevalence in the
general pediatric and adolescent population reported by
others1
-922
'23
.
The frequency of association of underlying disease
and back pain as reported in several other studies has
differed markedly from the prevalence in our study.
Turner et al.29
found that 50 per cent of sixty-one pa-
tients had discrete abnormalities, King et al.1719
found
that 63 per cent of fifty-four patients had positive find-
ings, and Hensinger14
reported that eighty-five of 100
patients had positive findings including kyphosis or sco-
liosis. Feldman et al.10
, at the Annual Meeting of the
Scoliosis Research Society in 1994, reported no demon-
strable pathological condition in 78 per cent of pa-
tients who had back pain. In the present study, only 9
per cent (forty-eight) of a large, unselected group of 560
patients who had scoliosis and back pain (2 per cent of
all 2442 patients initially diagnosed with idiopathic sco-
liosis at our clinics) had a demonstrable pathological
condition.
A patient who has scoliosis and back pain should be
assessed with an evaluation that includes the recording
of a complete history, physical examination, and good-
quality plain radiographs, as is done for any other pa-
tient who has scoliosis. If the findings of this initial
evaluation are normal, a presumptive diagnosis of idio-
pathic scoliosis can be made. The scoliosis can then be
treated appropriately (with observation, bracing, or
operative intervention), while the pain is treated non-
operatively (with avoidance of provocative activities
and use of medication with exercises) unless the magni-
tude of the deformity indicates a need for operative
stabilization. In the present study, a left thoracic curve
or an abnormal neurological finding was most pre-
dictive of underlying disease. There was no association
between the underlying diagnosis and any other demo-
graphic or physical finding. Our study shows that it is
unnecessary to perform extensive diagnostic studies on
every child who has scoliosis and back pain. We found
plain radiographs to be helpful in the diagnosis of the
underlying condition in most of the patients. In se-
lected patients, magnetic resonance imaging was the
next most useful study for the identification of tumors
and lesions of the spinal cord that could not otherwise
be diagnosed326
.
NOTII: The authors thank Sonia Vazquez. Sylvia Bilker, and Pal Ford-Williams for
collecting the data; Judy Guidry and Diane Ramey for the statistical analysts; and Holly Wise
for preparing the manuscript.
VOL. 79-A, NO. 3, MARCH 1997

5. 368 NORMAN RAMIREZ, C. E. JOHNSTON, II, AND R. H. BROWNE
References
1. Balague, E; Dutoit, G.; and Waldburger, M.: Low back pain in schoolchildren. An epidemiological study. Scandinavian J. Rehab. Med.,
20:175-179,1988.
2. Banna, M.; Pearce, G. W.; and Uldall, R.: Scoliosis: a rare manifestation of intrinsic tumours of the spinal cord in children. J. Neurol.,
Neurosurg., and Psychiat., 34: 637-641,1971.
3. Barnes, P. D.; Brody, J. D.; Jaramillo, D.; Akbar, J. U.; and Emans, J. B.: Atypical idiopathic scoliosis: MR imaging evaluation. Radiology,
186: 247-253,1993.
4. Bodner, R. J.; Heyman, S.; Drummond, D. S.; and Gregg, J. R.: The use of single photon emission computed tomography (SPECT) in
the diagnosis of low-back pain in young patients. Spine, 13:1155-1160,1988.
5. Bunnell, W. P.: Back pain in children. Orthop. Clin. North America, 13: 587-604,1982.
6. Cobb, J. R.: Outline for the study of scoliosis. In Instructional Course Lectures, The American Academy of Orthopaedic Surgeons. Vol. 5,
pp. 261-275. Ann Arbor, J. W. Edwards, 1948.
7. Cochran, T., and Nachemson, A.: Long-term anatomic and functional changes in patients with adolescent idiopathic scoliosis treated
with the Milwaukee brace. Spine, 10:127-133,1985.
8. Coonrad, R. W.; Richardson, W. J.; and Oakes, W. J.: Left thoracic curves can be different. Orthop. Trans., 9:126-127,1985.
9. Fairbank, J. C; Pynsent, P. B.; Van Poortvliet, J. A.; and Phillips, H.: Influence of anthropometric factors and joint laxity in the incidence
of adolescent back pain. Spine, 9:461-464,1984.
10. Feldman, D. S.; Wright, J. G.; and Hedden, D. M.: Chronic back pain in children and adolescents. Orthop. Trans., 19: 592,1995-1996.
11. Fraser, R. D.; Paterson, D. C; and Simpson, D. A.: Orthopaedic aspects of spinal tumours in children. J. Bone and Joint Surg., 59-B(2):
143-151,1977.
12. Grantham, V. A.: Backache in boys — a new problem? Practitioner, 218: 226-229,1977.
13. Hall, J., and Nachemson, A.: Debate: scoliosis. Spine, 2: 318-324,1977.
14. Hensinger, R. N.: Back pain in children. In The Pediatric Spine, pp. 41-60. Edited by D. S. Bradford and R. N. Hensinger. New York,
Thieme, 1985.
15. Jackson, R. P.; Simmons, E. H.; and Stripinis, D.: Incidence and severity of back pain in adult idiopathic scoliosis. Spine, 8: 749-756,1983.
16. King, H. A.: Back pain in children. Pediat. Clin. North America, 31:1083-1095,1984.
17. King, H. A.: Evaluating the child with back pain. Pediat. Clin. North America, 33:1489-1493,1986.
18. King, H.: Back pain in children. In The Pediatric Spine: Principles and Practice, pp. 173-183. Edited by S. L. Weinstein. New York, Raven
Press, 1994.
19. King, et al.: A prospective study of back pain in children [abstract]. J. Pediat. Orthop., 6: 367,1986.
20. Mehta, M. H., and Murray, R. O.: Scoliosis provoked by painful vertebral lesions. Skel. Radiol., 1: 223-230,1977.
21. Micheli, L.: Low back pain in the adolescent: differential diagnosis. Am. J. Sports Med., 7: 362-364,1979.
22. Mierau, D.; Cassidy, J. D.; and Yong-Hing, K.: Low-back pain and straight leg raising in children and adolescents. Spine, 14: 526-
528, 1989.
23. Nissinen, M.; Heliovaara, M.; Seitsamo, J.; Alaranta, H.; and Poussa, M.: Anthropometric measurements and the incidence of low back
pain in a cohort of pubertal children. Spine, 19:1367-1370,1994.
24. Papanicolaou, N.; Wilkins, R. H.; Emans, J. B.; Treves, S.; and Micheli, L. J.: Bone scintigraphy and radiography in young athletes with
low back pain. Am. J. Radiol., 145:1039-1044,1985.
25. Risser, J. C: The iliac apophysis: an invaluable sign in the management of scoliosis. Clin. Orthop., 11:111-119,1958.
26. Schwend, R. M.; Hennrikus, W.; Hall, J. E.; and Emans, J. B.: Childhood scoliosis: clinical indications for magnetic resonance imaging.
J. Bone and Joint Surg., 77-A: 46-53, Jan. 1995.
27. Taylor, L. J.: Painful scoliosis: a need for further investigation. BMJ: British Med. J., 292:120-122,1986.
28. Thompson, G. H.: Back pain in children. J. Bone and Joint Surg., 75-A: 928-938, June 1993.
29. Turner, P. G.; Green, J. H.; and Galasko, C. S.: Back pain in childhood. Spine, 14: 812-814,1989.
30. Weinstein, S. L.; Zavala, D. C; and Ponseti, I. V.: Idiopathic scoliosis. Long-term follow-up and prognosis in untreated patients. / Bone
and Joint Surg., 63-A: 702-712, June 1981.
THE JOURNAL OF BONE AND JOINT SURGERY