Between August 1963 and October 1972, seventy-four Salter innominate
osteotomies were performed in sixty-two patients. The technique
was precisely the same as that described by Salter in his first
report9. In all hips, one or two
Kirschner wires were used for internal fixation of the osteotomy
site. Postoperatively, a spica cast was applied and the hip was
immobilized for six weeks. When the osteotomy site appeared united
radiographically, the Kirschner wires were removed and active hip
motion was permitted. Two days later, partial weight-bearing with
the assistance of a physiotherapist was allowed. After about two
weeks, full weight-bearing was encouraged.
Patient Data
One female patient died of aplastic anemia 6.4 years postoperatively.
Therefore, the results of seventy-three osteotomies in sixty-one
patients could be evaluated. There were ten male and fifty-one female
patients. Five of the ten male patients and seven (14%)
of the fifty-one female patients had a bilateral procedure. The
average age at the time of the operation was 4.1 years (range, 1.3
to 8.8 years) for all patients, 4.8 years (range, 1.3 to 8.2 years)
for the boys, and 4.0 years (range, 1.3 to 8.8 years) for the girls.
Surgical Procedures
Thirty-seven right hips and thirty-six left hips were involved. Twenty-four
hips had had operative treatment before the index operation: five
had had an isolated intertrochanteric osteotomy; nine, an intertrochanteric
osteotomy and a closed reduction; two, an intertrochanteric osteotomy
and an open reduction; and eight, an isolated open reduction. Forty-eight hips
had the innominate osteotomy alone, twelve had the osteotomy combined
with an open reduction without an intertrochanteric osteotomy, two
had the osteotomy with an open reduction and an intertrochanteric
osteotomy, and four had the osteotomy with an intertrochanteric
osteotomy without an open reduction. Seven hips had an intertrochanteric
osteotomy 0.3 to 5.8 years after the innominate osteotomy.
The functional and radiographic results were evaluated by an unbiased
investigator (A.B.) who was not a member of the department. The
patients had a clinical examination and responded to a questionnaire,
which included the Merle d’Aubigné and Postel
score12 as well as the Harris
hip score13. Failure was defined
as a true revision or a Merle d’Aubigné score
of <13 points and/or a Harris hip score of <70
points. Radiographs of the pelvis made preoperatively, at six months postoperatively,
and at the time of the most recent follow-up were analyzed to determine
the grade of dislocation according to the system of Tönnis14, the CCD angle14,
the acetabular index15, the acetabular
angle of Sharp16, the ACM angle
of Idelberger and Frank17, the
center-edge angle of Wiberg1,
the migration percentage described by Reimers18,
the grade of avascular necrosis of the femoral head according to
the system of Tönnis14,
the severity of degenerative arthrosis according to the system of
Tönnis14, and the radiographic
result according to the Severin classification19.
Additionally, the summarized hip factor (SHF)20,21 was
calculated with use of the measurements of the depth of the acetabulum
(the ACM angle), lateral coverage of the femoral head by the osseous
acetabulum (the CE [center-edge] angle), and lateral
subluxation of the femoral head (MZ):
(Eq.)
Instead of calculating the summarized hip factor, nomograms can
be used14,20. The CCD angle, acetabular
index, acetabular angle, ACM angle, center-edge angle, and summarized
hip factor were graded according to the age-adjusted classification
system of the German Association of Orthopaedic and Trauma Surgery14, which enabled us to compare the
measurements for the different age-groups. Grade 1 indicated normal;
grade 2, slightly abnormal; grade 3, severely abnormal; and grade
4, extremely abnormal.
We performed survival analysis to January 1999 with use of true
revision as the criterion of failure. Survival curves with 95% confidence
intervals were calculated with use of the Kaplan-Meier method (JMP,
version 3.1.6.2; SAS Institute, Cary, North Carolina). The p value
for the noncrossing survival curves was calculated with use of the
log-rank test, and the p value for the crossing curves was calculated
with use of the Wilcoxon test. Correlation and significance between
variables were calculated with use of the chi-square test, Fisher test,
or t test.
The mean duration of follow-up was 30.9 years (range, 26.2 to
35.4 years). At the latest follow-up examination, the mean age of
the patients was thirty-five years (range, 27.6 to 40.6 years).
We were able to locate all sixty-one patients (seventy-three hips)
for clinical evaluation. Seven hips had had a reoperation and were
excluded from the clinical and radiographic follow-up evaluation.
The reoperations included one acetabuloplasty at 5.4 years postoperatively,
one triple osteotomy at 25.9 years postoperatively, and five total
hip replacements at 19.7, 23.1, 24.4, 27.9, and 27.9 years postoperatively.
Two female patients (three hips), all with an excellent clinical result
(a Harris hip score of 100, 100, or 92 points), declined follow-up
radiographic evaluation. Clinical assessment of the end result was
therefore possible for sixty-six hips of the seventy-three index
hips, and radiographic assessment was possible for sixty-three.
Survival Analysis
When failure was defined as a true revision (acetabuloplasty, triple
osteotomy, or total hip replacement), the cumulative survival rate
for all seventy-three hips was 0.90 at 35.3 years (Fig. 1). Survival analysis
was performed to determine whether outcome could be predicted by
the age at the time of the operation, grades of dislocation at the
first radiographic examination (before any conservative or operative
treatment) and immediately before the index operation, preoperative
and postoperative grades for the acetabular index, preoperative and
postoperative grades for the summarized hip factor, preoperative
and postoperative grades of avascular necrosis, and technique of
reduction. Hips with grade-4 dislocation preoperatively had a significantly
worse cumulative survival rate (0.48; 95% confidence interval,
0.03 to 0.91) (Wilcoxon test, p = 0.0129) than did those
with grade-1 (0.97), grade-2 (0.90), or grade-3 (0.83). Hips with
a postoperative summarized hip factor of grade 4 had a significantly
worse cumulative survival rate (0.57; 95% confidence interval,
0.24 to 0.91) (Wilcoxon test, p = 0.0132) than did those
with grade 1 (1.00), grade 2 (0.88), or grade 3 (0.96). Hips with
grade-4 avascular necrosis preoperatively had a significantly worse
cumulative survival rate (0.60; 95% confidence interval,
0.17 to 1.00) (Wilcoxon test, p = 0.036) than did those
with grade-0 (0.97), grade-1 (0.94), grade-2 (0.80), or grade-3
(0.76). Hips with grade-4 avascular necrosis at six months postoperatively
(Fig. 2) had
a significantly worse cumulative survival rate (0.40; 95% confidence
interval, 0 to 0.83) (Wilcoxon test, p = 0.0004) than did
those with grade-0 (1.00), grade-1 (0.92), grade-2 (0.80), or grade-3
(0.88). The cumulative rate of survival after a Salter osteotomy
performed simultaneously with an open reduction (0.61; 95% confidence
interval, 0.34 to 0.89) (log-rank test, p = 0.0005) was
significantly worse than that after a Salter osteotomy preceded
by a separate open reduction (1.00) or a closed reduction (0.96;
95% confidence interval, 0.9 to 1.00). Of the fourteen
hips treated with a simultaneous Salter innominate osteotomy and
open reduction, seven had grade-3 and seven had grade-4 dislocation
at the first radiographic examination. Of the ten patients managed
with a separate open reduction, two had grade-2, six had grade-3,
and two had grade-4 dislocation. Of the forty-nine patients who
had a closed reduction, fixation alone, or no pretreatment, two
had grade-1; fifteen, grade-2; nineteen, grade-3; and thirteen,
grade-4 dislocation. The age at the time of the operation, the grade
of dislocation at the time of the first examination, the preoperative
and postoperative acetabular index, and the preoperative summarized
hip factor were not found to have a significant influence on the
cumulative rate of survival, with the numbers available.
Clinical Results
At the time of the latest follow-up examination, the mean Merle
d’Aubigné and Postel score for the sixty-six hips
that had not had a reoperation was 15.6 points (range, 9 to 18 points)
and the mean Harris hip score was 87 points (range, 42 to 100 points).
In addition to the seven true revisions, all eight hips with
a Harris hip score of <70 points and/or a Merle
d’Aubigné and Postel score of <13 points
were considered failures. Thus, fifteen (21%) of the seventy-three
hips failed. The preoperative grade of dislocation had a highly
significant influence on the rate of failure (chi-square test, p < 0.0001)
(Fig. 3),
whereas the grade of dislocation at the first radiographic examination
was less significant (chi-square test, p = 0.0388) and
the grade of dislocation after the index operation was not significant,
with the numbers available. The rate of failure was also significantly
influenced (chi-square test) by the preoperative (p = 0.0392)
and postoperative (p = 0.0072) grades of avascular necrosis,
preoperative (p = 0.0051) and postoperative (p = 0.0002)
grades for the summarized hip factor (Fig. 4), and technique of reduction (p < 0.0001)
(Fig. 5).
When only hips with a summarized hip factor of grade 4 were considered,
five of eighteen hips without an intertrochanteric osteotomy and
eight of twenty-one with an intertrochanteric osteotomy failed.
The addition of an intertrochanteric osteotomy was not found to
have a significant influence on the failure rate (chi-square test,
p = 0.4956), with the numbers available.
Complications
Two hips had displacement of the bone graft, and two other hips
had dislocation of the Kirschner wire without displacement of the
bone graft. No additional surgery was performed in any of the four
hips, and, six months postoperatively, the summarized hip factor
was grade 2 in two hips and grade 3 and grade 4 in one hip each.
At the time of the latest follow-up, the four hips had a mean Harris
hip score of 85 points (range, 62 to 97 points). In addition, three
of them were graded Severin class I and one was graded Severin class
II. One hip had a redislocation six weeks postoperatively, when the
spica cast was removed. The parents of this girl wanted no additional
treatment for her. At the time of the latest follow-up, the result
was poor (Severin class VI and a Harris hip score of only 64 points).
After removal of the spica cast, two other hips had a flexion and
abduction contracture, which normalized after a short period of
physiotherapy.
Radiographic Evaluation
When the patients were seen for the first time at our clinic,
the dislocation was classified as grade 1 in two hips, grade 2 in seventeen,
grade 3 in thirty-two, and grade 4 in twenty-two. Operative or conservative
treatment before the index operation led to a clear improvement;
the dislocation was classified before the index operation as grade
1 in thirty-nine hips, grade 2 in twenty-one, grade 3 in six, and
grade 4 in seven. On the radiographs made six months postoperatively,
the dislocation was classified as grade 1 in sixty-eight hips, grade
2 in four, and grade 4 (redislocated) in one. The mean migration
percentage according to Reimers improved from 52.4% (range, 12% to
100%) preoperatively to 18.5% (range, 0% to
100%) after removal of the spica cast. At the time of the
latest follow-up examination, the mean migration percentage was 20.3% (range,
0% to 100%).
The mean acetabular index improved from 32° (extremely abnormal)
preoperatively to 20.1° (normal to slightly abnormal) postoperatively
and to 11.1° (normal) at the time of the latest follow-up. The mean
distance of decentering20 improved
from 12.1 mm preoperatively to 6.3 mm postoperatively. The grades
for the preoperative, postoperative, and final follow-up radiographic
measurements are shown in the electronic appendix to this paper.
Preoperatively, thirty-nine hips (53%) had avascular
necrosis: sixteen (22%) had grade 1, eight (11%)
had grade 2, ten (14%) had grade 3, and five (7%)
had grade 4. Six months postoperatively, forty-four hips (60%)
had avascular necrosis: twenty-four (33%) had grade 1,
seven (10%) had grade 2, eight (11%) had grade
3, and five (7%) had grade 4. No hip without radiographic
signs of avascular necrosis (grade 0) preoperatively had radiographic
signs of avascular necrosis at six months postoperatively.
At the latest follow-up examination, fifty (79%) of
the sixty-three hips that had complete radiographic follow-up were graded
as Severin class I; six, as class II; two, as class III; three,
as class IV; one, as class V; and one, as class VI. Twenty-two hips
had no evidence of osteoarthrosis, twenty-five had grade-1 osteoarthrosis,
nine had grade-2, and seven had grade-3 according to the radiographic
staging system of Tönnis14.
Up to the age of about ten years, children with developmental dysplasia
of the hip should always first be treated with complete and concentric
reduction of the femoral head into the depth of the true acetabulum.
Several types of operative procedures to stabilize the reduced hip
in older children have been described. Currently, acetabuloplasties
with use of the techniques of Lance or Pemberton and the pelvic
osteotomies of Salter or Chiari are most often performed14.
Tönnis reported that computerized tomographic studies
performed after Chiari osteotomies revealed that only about 33% of
the hips had adequate coverage of the femoral head in the sagittal
plane, whereas >80% had good coverage in the coronal
plane22. Additionally, the higher
inclination of the medially shifted original acetabulum causes a
reduction of the functional area of the original articular cartilage
contact. In a study of hip dysplasia by the German Association of
Orthopaedic and Trauma Surgery, 4357 hip joints and the results
of 4078 operative procedures (2483 intertrochanteric osteotomies,
611 modified Lance acetabuloplasties, 262 Pemberton acetabuloplasties,
287 Salter innominate osteotomies, and 435 Chiari osteotomies) were
evaluated at a mean of 5.8 years14,22.
The study, which confirmed the findings in other series23, showed that the results were better
after a Salter osteotomy or an acetabuloplasty than after a Chiari
osteotomy.
When the acetabular index and the center-edge angle of Wiberg
are measured, the extent of primary correction after a Lance or
Pemberton acetabuloplasty may be found to be even better than that
after a Salter osteotomy14. In
the early postoperative period after these procedures, the acetabular
roof mostly shows positive development. After an acetabuloplasty,
however, there are strong indications of severe growth disturbances
in the acetabular roof during puberty24,25.
This problem may result from violation of the centers of ossification,
which are essential for the growth of the acetabular rim in adolescence.
Thus, the Salter innominate osteotomy seems to be preferable for
operative treatment of developmental dysplasia of the hip.
The rationale for the Salter osteotomy is stabilization of the reduced
hip in the position of function by redirecting the maldirected acetabulum.
In our study, the coverage of the lateral part of the femoral head
(the acetabular index) increased a mean of 12° between the preoperative
and postoperative evaluations and another 9° between the postoperative
evaluation and the time of the latest follow-up; that is, the mean
grade for the acetabular index improved from 3.2 to 1.5 and remained
at this level at the time of the latest follow-up (Figs. 6-A, 6-B, 6-C, and 6-D). This finding
is similar to that of Rab26, who
developed a mathematical model for the Salter osteotomy that predicted
a theoretical improvement in the acetabular index of about 10°,
and to the results reported in other studies27-30 (see
Appendix). The improvement in the acetabular index after a single-stage
bilateral Salter osteotomy may be even better than that after a
unilateral osteotomy. Ochoa et al. noted that the side on which
the osteotomy was performed second had better correction; therefore,
they recommended that the first osteotomy should be performed on
the hip with less dysplasia31.
Salter and Dubos, however, considered bilateral simultaneous osteotomy
to be contraindicated and recommended that the osteotomy on the
second hip be performed approximately two weeks after the first32. Although the mean grade for the
acetabular index remained constant between the postoperative radiograph
and that made at the time of the latest follow-up in our patients,
the mean grade for the summarized hip factor improved from 3.4 preoperatively
to 2.3 postoperatively and to 1.7 at the latest follow-up examination.
The positive biomechanical influence of a well-centered, stabilized
hip on secondary maturation and normal development of the hip has
been documented in most studies with long-term follow-up27,31,33-36.
The technical complications in our series were similar to those
in other studies. Migration of the Kirschner wires, displacement
of the bone graft, and loss of position of the osteotomy site are
relatively rare complications that can be treated effectively31,32,35,37. Salter and Dubos reported
that 5.6% of the hips in their series redislocated and
14.3% resubluxated32.
Other investigators also have reported a redislocation or resubluxation
rate of between 2% and 14%29,31,34,38.
The rate of these complications after a Salter osteotomy is similar
to that after an acetabuloplasty7.
Avascular necrosis may be the most severe complication after conservative
or operative treatment of developmental dysplasia of the hip39, and there is a high risk that it
will develop after closed reduction and conservative treatment40,41. The prevalence of avascular
necrosis after a well-performed Salter osteotomy has been reported
to range from 2.2% to 10%27,29,37,42,43.
After a combined open reduction and Salter osteotomy, the rate of
avascular necrosis has been reported to increase by up to 13%22,42. In the study by the German Association
of Orthopaedic and Trauma Surgery14,
the preoperative rate of avascular necrosis was 25%. Avascular
necrosis developed after a Salter osteotomy in 9.3% of
the hips without preexisting avascular necrosis14.
In the study by Gulman et al.44,
a good or excellent result was achieved in 93.4% of the
hips with grade-1 avascular necrosis, 75% of those with
grade-2, 0% of those with grade-3, and 50% of
those with grade-4. In our series, a good or excellent result according
to the Severin classification was noted in 93% of the hips
with grade-0 avascular necrosis, 81% of those with grade-1,
100% of those with grade-3, and 50% of those with
grade-4. The cumulative survival rate was significantly worse in
patients with grade-4 avascular necrosis (p = 0.0004) (Fig. 2).
Similar to our experience (Fig. 5), Mellerowicz et al. found that
between 63% and 70% of the hips that had not had
a simultaneous open reduction had normal radiographic results after
long-term follow-up, whereas only 31% of the hips that
had had a combined Salter osteotomy and open reduction had a normal
result36. Our data confirmed Salter’s
theory that avascular necrosis is a complication of the open reduction
rather than of the osteotomy7,32.
The more closely the dysplastic anatomy could be restored to normal
by the Salter osteotomy, the higher the percentage of successful
long-term results (Fig. 4). It has been well documented,
however, that the pathoanatomy of acetabular dysplasia in adolescents
and young adults is not uniform. Murphy et al.45 found
not only a maldirection of the acetabulum but also an individual,
complex global dysplasia that could not be completely corrected
by reorientation, which may explain why 65% of the hips
in our series had some degree of radiographic osteoarthrosis at
the time of follow-up. On the other hand, we found that the anatomical
result of a well-normalized acetabulum could be preserved for three
decades. This observation is similar to that of Murray5, who found that an anatomically normal
hip in an adult is extremely unlikely to undergo spontaneous osteoarthrotic change,
and to that of Gallien et al.37,
who found that poor results had a tendency to get worse and good
results had a tendency to persist.
Malvitz and Weinstein reported the functional and radiographic
results for a group of patients with congenitally dislocated hips
who had had only closed treatment41.
The grade of dislocation before treatment and the mean duration
of follow-up (thirty years) were similar to those in our series.
Although the mean functional results were quite similar to those
in our series, the hips in their series had radiographic evidence
of more severe dysplasia. A total hip replacement had been performed
in 11.8% of the hips in their series compared with 6.8% in
ours. They found that the percentage of hips with an excellent functional
result decreased from 94% after twenty to twenty-nine years
of follow-up to 57% after thirty to thirty-nine years.
In our study, the mean Harris hip score was 86 points after 26.2
to 29.9 years of follow-up and 87 points after 30.2 to 35.3 years.
Malvitz and Weinstein reported that the prevalence of severe degenerative changes
in their patients increased significantly with the duration of follow-up
(0% at less than twenty years, 4% at twenty to
twenty-nine years, 29% at thirty to thirty-nine years,
and 42% at forty years or more) (p < 0.001)41. The prevalence of deterioration
in the hips in our study increased from 8% at 26.2 to 29.9
years to 12.8% at 30.2 to 35.3 years, but the finding was
not significant, with the numbers available. These results indicate
that patients managed with a Salter innominate osteotomy may have
a better result than those managed with a closed reduction alone.
With a true revision as the end point, the survival curve demonstrated
that most revisions became necessary more than twenty years after
the index operation (Fig. 1). Therefore, the duration of follow-up
after treatment of developmental dysplasia of the hip should be
more than twenty-five years. As we do not definitely know how many hips
in our series will eventually fail in the next few years, we must
follow our patients over a lifetime because the worst enemy of a
good result is an adequately long follow-up.
We were able to draw five conclusions from our study. First, we
found that, when an open reduction is needed, the higher probability
of a better long-term result associated with an open reduction performed
separately before a Salter osteotomy should be weighed against the
morbidity associated with the additional administration of general
anesthesia and the additional immobilization in a spica cast. More
data are necessary before we can evaluate the influence of the timing
of open reduction and Salter osteotomy on the clinical outcome. Second,
we also found that the grade of dislocation at the time of the first
examination and the grade immediately before the Salter osteotomy
have a significant influence on the long-term result. Therefore,
the quality of conservative or operative treatment before the Salter
osteotomy is an essential factor. Third, the more closely that the
dysplastic anatomy is restored to a normal configuration, the greater
the chance for an excellent or good long-term result. The summarized
hip factor is a helpful tool for assessment and prognosis. Fourth,
avascular necrosis that occurs after treatment of developmental
dysplasia of the hip is associated with a high risk of osteoarthrosis. There
seems to be a slightly increased risk of avascular necrosis when
open reduction and the Salter innominate osteotomy are performed
simultaneously than when these procedures are staged. Fifth, when
an acetabulum can be well normalized without the development of
avascular necrosis, good long-term results, lasting for up to thirty
years, can be expected.
Note: The authors thank Mrs. Pietsch-Breitfeld, Institute for Medical
Information Processing, Eberhard-Karls-Universität, Tübingen,
for statistical advice.
Tables showing the grades of the preoperative, postoperative, and
final follow-up radiographic measurements in the present study and
the results of comparative studies from the literature are available
with the electronic versions of this article, on our web site at
www.jbjs.org (go to the article citation and click on "Supplementary
Material") and on our CD-ROM (call our subscription department,
at 781-449-9780, to order the CD-ROM).