Study Sites and Training
Thirty-two sickle-cell-disease centers agreed to participate in the trial
by keeping a monthly log of patients with sickle cell disease who might also
have osteonecrosis of the femoral head. A study protocol, an operations
manual, and standardized data-collection forms were developed for the
participating centers. The hip core decompression and the physical therapy
regimen were standardized for the study. Data regarding preoperative
transfusion, perioperative management, and radiographic assessment were
recorded at each center. There were site-visits to all centers to facilitate
training of physicians, surgeons, and physical therapists. Institutional
review board approval was obtained at each site, and a National Institutes of
Health (Heart, Lung, and Blood Institute)-sponsored Data Safety and Monitoring
Board monitored the study design and progress. The study database was designed
and managed at the Children's Hospital and Research Center, Oakland,
California, the coordinating center for the study. Data collection forms were
edited when received; error and data quality checks were built into the data
entry application. The data were imported into SAS software (SAS Institute,
Cary, North Carolina) for further cleaning, error checking, and reporting
monthly.
Inclusion Criteria
Patients with hemoglobin-SS, hemoglobin-SC, or
hemoglobin-S/ß-thalassemia who were ten years of age or older and had hip
pain and/or a decreased range of motion of the hip were screened with plain
radiographs and magnetic resonance images of both hips. Those with at least
one hip with Stage-I, II, or III osteonecrosis of the femoral head, as defined
by Steinberg et
al.44, were
eligible for enrollment. In the Steinberg system, Stage 0 indicates no
radiographic evidence of osteonecrosis of the femoral head; Stage I, findings
consistent with osteonecrosis of the femoral head on magnetic resonance
imaging but normal findings on plain radiographs; Stage II, sclerosis and
lucencies of the femoral head seen on plain radiographs; Stage III, evidence
of flattening of the femoral head on plain radiographs; and Stage IV, collapse
of the femoral head with degenerative changes in the joint space.
Exclusion Criteria
Patients who were less than ten years of age and those with a concurrent
condition other than sickle cell disease that predisposed them to the
development of osteonecrosis of the femoral head (e.g., a history of chronic
systemic steroid use or alcoholism) were excluded from the study.
Patients Screened
Four hundred and twenty patients with sickle cell disease from thirty-two
centers were screened for osteonecrosis of the femoral head of the hip; 176
met the inclusion criteria. Of the 244 patients who were excluded from the
study, one-third had advanced osteonecrosis of the femoral head (Steinberg
Stage IV) with collapse of the hip joint. Forty-six (26%) of the eligible
patients were enrolled in the study; the remaining patients were not enrolled
because they declined to participate in a randomized surgical trial or because
they had geographic or social issues that were thought to prevent their
compliance with the study design.
Enrollment and Randomization
From June 1998 through June 2002, forty-six patients (forty-six hips) from
fifteen centers were enrolled after they provided informed consent. One hip of
each patient was randomized to either Arm A (hip core decompression and
physical therapy) or Arm B (physical therapy alone) with matching on the basis
of the Steinberg stage assigned at the local center. (When patients had
bilateral disease, the more symptomatic hip was entered into the study.) A
weakly positive antinuclear antibody developed in an Arm-A patient one year
after that patient started the protocol, but, after consultation with
rheumatic disease specialists, a diagnosis of a connective tissue disorder was
not made. We opted to keep this patient in the study. After randomization,
eight patients (six in Arm A and two in Arm B) were withdrawn from the study;
six of them were withdrawn because the parents or patient declined to
participate and two were withdrawn at the discretion of the principal
investigator. This left thirty-eight patients (thirty-eight hips; seventeen in
Arm A and twenty-one in Arm B) from fifteen centers who completed the
treatment protocol.
Radiographic Staging of the Hip Disease
The two treatment arms were initially matched according to the Steinberg
stage of the study hip assigned by the participating centers at the time of
registration. On completion of the study, the coordinating center's radiology
committee, consisting of three radiologists who were blinded to the reports
from the participating centers, reinterpreted the baseline plain radiographs
and magnetic resonance imaging scans of the patients who had completed the
protocol. Thirty-eight bilateral plain radiographs and thirty-four bilateral
magnetic resonance imaging scans were available for review. Bone marrow edema,
a rare syndrome that presents like osteonecrosis of the femoral head, and
acute infarction were ruled out on the basis of the duration of symptoms at
the time of study entry or magnetic resonance imaging findings consistent with
a diagnosis of osteonecrosis of the femoral
head45-47.
The radiology committee agreed with the Steinberg stage assigned by the
local centers for 56% of the hips. In 73% of the instances in which there was
disagreement, the magnitude of the difference was only one stage. The stages
assigned by the committee are shown in
Table I and were used in all
outcome analyses of the thirty-eight patients (thirty-eight hips) who
completed the treatment protocols.
Patient Demographics
The mean age of the patients who participated in the study was twenty-six
years. Gender, type of hemoglobinopathy, ethnicity, laterality of the study
hip (right or left), and percentage of patients with bilateral disease did not
differ significantly between the two arms
(Table I).
Treatment Arms
The seventeen patients (seventeen hips) randomized to Arm A (hip core
decompression and physical therapy) were admitted one day preoperatively for
evaluation, hydration, and transfusion. Patients with hemoglobin-SS or
hemoglobin-S/ß0-thalassemia underwent a single preoperative transfusion
to bring the hemoglobin level to approximately 100
g/L48. For patients
with hemoglobin-SC or hemoglobin-S/ß+-thalassemia whose baseline
hemoglobin concentration was higher than 100 g/L, partial exchange transfusion
was performed to avoid raising the hemoglobin concentration to an unsafe
level48.
With the patient under general or spinal anesthesia, a lateral approach was
utilized and a guide pin was inserted into the lateral cortex of the femur and
was directed into the proximal part of the femoral neck up to but not into the
necrotic lesion as defined on preoperative magnetic resonance images of the
femoral head. Alignment of the guide pin was checked on both anteroposterior
and lateral fluoroscopic views to ensure that the pin was in the center of the
involved area of the femoral head with satisfactory position and depth. A
cannulated drill bit was then directed to the tip of the guide pin. After the
position of the drill bit was checked on anteroposterior and lateral
fluoroscopic views, the guide pin was removed, and a trocar was advanced under
fluoroscopic guidance into the lesion, with retrieval of bone for histologic
analysis but with at least 5 mm of bone left distal to the subchondral plate.
The core canal was gently curetted and irrigated prior to wound closure.
All operative notes were reviewed, and the surgeons filled out an
additional form indicating the diameter of the drill that had been used, the
location of the drill track, and whether perforation or any other
intraoperative complications had occurred. Pathology specimens were sent in
buffered formalin to the coordinating center for review.
After discharge, patients followed a standardized six-week protocol of
limited weight-bearing, physical therapy, and home exercises (see Appendix).
In the first six weeks, physical therapy included outpatient visits and a home
program of stretching and strengthening of the hip muscles, especially the
adductors. Physical therapy continued for an additional three months after the
initial six-week period (see Appendix).
Patients randomized to Arm B participated in the same program of physical
therapy for the same duration. Compliance with the touch-down weight-bearing,
physical therapy, and home exercise regimen was encouraged weekly by telephone
contact and was recorded on a form.
Outcome Measures
Assessment of the clinical status of a young population with complications
of sickle cell disease required modification of the Harris hip score, which
was developed primarily for older patients with osteoarthritis following total
hip replacement49.
The Children's Hospital Oakland Hip Evaluation Scale
(CHOHES)50 is a
100-point scale for assessing pain, function, strength, and range of motion
(see Appendix). While they were enrolled in the study, the patients kept a
pain log daily for the first six weeks and then weekly thereafter with the use
of a 10-point visual analog scale. Estimates of the reliability and validity
of the CHOHES for the evaluation of patients with osteonecrosis of the femoral
head and sickle cell disease were published
recently50.
Physical therapists from each participating center were trained in the use of
the CHOHES. The patients were examined with the scale at baseline and then
were scheduled for reevaluations at three-month follow-up intervals while they
were enrolled in the study. We defined clinical success as improvement in the
baseline CHOHES score of at least 15 points at the time of follow-up.
Complications of sickle cell disease, including hip pain, vaso-occlusive
episodes, acute chest syndrome, fever and infection, neurologic events, renal
disease, and cardiovascular disease, were also tracked throughout the study.
Acute chest syndrome, defined as a new pulmonary infiltrate with fever and
respiratory symptoms, is a common complication and cause of mortality in
patients with sickle cell disease and has been reported after surgical
procedures48.
Hip Survival
If a patient required additional surgical intervention on the study hip
(total hip replacement for patients treated in the surgical arm or hip core
decompression or total hip replacement for those treated with physical therapy
alone), the coordinating center was notified. At the close of the study, each
center was again contacted to ensure accurate estimates of the study hip
survival rate.
Statistical Methods
Baseline demographic data and rates of complications related to the
sickle-cell disease were compared between the two treatment groups with use of
chi-square analysis with the Yates continuity correction for categorical data
and the Student t test for continuous data. Only p values of =0.05 were
considered significant.
The primary outcome variable was clinical improvement in the CHOHES score
from baseline. With forty-six patients enrolled and the assumption of a
two-sided alpha of 0.05 and a standard deviation of 19, the power to detect a
15-point difference in mean change scores between the two treatment groups was
calculated as 0.75. With a 20% cancellation rate, the power is approximately
0.65. We also examined clinical improvement over time with use of a
random-effects model in which the dependent variable was a change in the
CHOHES score from baseline and the independent variables were the treatment
arm, baseline CHOHES score, and time. The secondary outcome variable was hip
survival, which was compared between groups with use of chi-square analysis
with the Yates continuity correction.
Compliance with Treatment Protocols
Compliance data were available for 89% of patients during the first six
weeks of the protocol. During this period, all patients were asked to visit a
physical therapist weekly, perform exercises daily as prescribed, and limit
weight-bearing on the study hip by using the prescribed gait aid. Overall, 78%
of the patients (twenty-nine of thirty-seven) utilized a gait aid and 85%
(twenty-nine of thirty-four) participated in physical therapy and home
exercises. Also, 70% (twenty-six of thirty-seven) had kept a daily pain diary
during the first six weeks.
Clinical Function
The average CHOHES score at baseline was 55 points in Arm A and 52 points
in Arm B. This difference was not significant. Follow-up CHOHES scores were
available for 89% of the patients, who had a mean of five follow-up
evaluations (range, one to fifteen). With use of each patient's last CHOHES
score and with exclusion of the scores from patients who underwent additional
surgical intervention, the mean improvement (and standard deviation) after an
average of twenty-seven months was 18.1 ± 19.8 points in the group
treated with hip core decompression and physical therapy and 15.7 ±
19.6 points in the physical therapy group. With the numbers studied, the
difference was not significant. In univariate analyses, an analysis of
variance of clinical improvement versus stage of disease at study entry did
not demonstrate significance. At the last assessment, clinical improvement
according to our criteria (a change of =15 points compared with the
baseline CHOHES score) was seen in 57% of the patients in Arm A (eight of
fourteen) and 50% of those in Arm B (ten of twenty); this difference also was
not significant with the numbers studied.
Hip Survival and Orthopaedic Complications
The overall hip survival rate was high in both groups (fourteen [82%] of
seventeen in Arm A and eighteen [86%] of twenty-one in Arm B), and the
difference between the two treatment arms was not significant with the numbers
studied. The Kaplan-Meier survivorship curves are shown in
Figure 1. The log-rank test for
the comparison of the two curves did not demonstrate a significant difference
(p = 0.8). The average time of active participation in the study was 3.1 years
(range, two months to 6.7 years) in Arm A compared with 3.0 years (range, two
months to 5.1 years) in Arm B. Six patients had additional surgical
intervention on the study hip (Table
II). In Arm A, three of the seventeen patients subsequently
underwent a total hip replacement of the study hip, at fifteen, sixteen, and
forty months after the hip core decompression, because of recurrent pain. All
three patients showed evidence of osteonecrosis of the femoral head on biopsy
and had some initial relief of the symptoms; however, despite stable findings
on the radiographs, pain returned to the baseline level by one year after the
hip core decompression. In Arm B, three (14%) of the twenty-one study hips
underwent hip core decompression, at ten, thirteen, and forty-seven months
after the start of the conservative treatment protocol (physical therapy
alone), because of increasing pain. Hip survival rates did not differ
significantly across the baseline Steinberg stages (90% for Stage I, 82% for
Stage II, and 82% for Stage III).
Complications in Contralateral Hips
Four patients of the seventeen in Arm A underwent a surgical procedure on
the contralateral hip. Three of these hips had collapse of the femoral head
(Stage-IV disease) at the time of study entry and underwent total hip
replacement. The fourth non-study hip had Stage-III disease, and a hip core
decompression was performed. One hip replacement in a non-study hip dislocated
on two separate occasions three and four months later. One patient in Arm B
underwent hip core decompression on the non-study hip. This patient, who had
bilateral Stage-II osteonecrosis of the femoral head at the time of study
entry, initially had some pain relief but requested a hip core decompression
of the study hip at nine months and of the non-study hip at twelve months
after starting the treatment protocol. Three patients in Arm A fell on the
study hip and one fell on the non-study hip after completing the six-week
treatment protocol; no fractures or long-term sequelae were documented.
Safety of Hip Core Decompression
No intraoperative or immediate postoperative complications were reported in
the seventeen patients enrolled in Arm A (hip core decompression). The average
hospital stay, including both the admission and the hospital discharge days,
was 4.6 days (2.8 days postoperatively), with a range of two to ten days. All
but one patient received a transfusion of packed red blood cells prior to the
surgery in order to increase the hemoglobin level and decrease the percentage
of hemoglobin-S cells; two underwent exchange transfusion, and the remainder
received a single transfusion. Fourteen patients received eight hours of
intravenous hydration preoperatively. General anesthesia was used for fourteen
patients, two of whom also received a regional block, and three patients were
managed with a regional (spinal) block alone. The mean duration of anesthesia
was 2.1 hours, and the average surgical time was 1.1 hours. The average blood
loss was reported to be 48 mL, and no patient required a transfusion
intraoperatively. Sixteen patients were given perioperative antibiotics, begun
in the operating room, and sixteen were maintained on oxygen in the immediate
postoperative period. None of the patients participating in this study
received routine anticoagulation or were treated with anticoagulation therapy
in the postoperative period. Only two patients received epidural narcotics as
part of their postoperative pain management regimen.
During the thirty-day postoperative period, one patient in Arm A was
readmitted to the hospital because of ongoing bilateral hip pain (on day 6).
Postoperative infection was documented in one patient in Arm A
(Table II). Drainage from the
site of the hip core incision occurred on the forty-second postoperative day;
the site was drained, and the patient was treated with antibiotics as an
outpatient; there were no more sequelae. No episode of acute chest syndrome
occurred during this period.
Biopsy Results
Twelve of the biopsies showed evidence of osteonecrosis of the femoral head
on histologic examination, one biopsy specimen was not interpretable because
of artifact, and four showed no osteonecrosis of the femoral head.
Complications of Sickle Cell Disease Reported During the Study
Period
The majority of patients in both treatment arms experienced complications
related to the sickle cell disease. The most frequent of these were
vaso-occlusive crises and hip pain. The sickle cell disease-related
complications reported to the coordinating center during the study period are
listed in Table II. With the
number of study subjects available, there was no significant difference in the
rates of complications between the two treatment arms. There were three
deaths, 3.1, 3.5, and 5.2 years after study entry. Two of the patients were
hospitalized for a sickle cell pain crisis; acute chest syndrome developed,
and there was a cardiorespiratory arrest. The third patient was also admitted
for sickle cell pain and was subsequently found unresponsive in her hospital
bed; the cause of death was unknown.
To our knowledge, this is the first randomized prospective study comparing
hip core decompression combined with physical therapy with physical therapy
alone for the treatment of symptomatic osteonecrosis of the femoral head in
patients with sickle cell disease. With a mean duration of follow-up of three
years, this study demonstrated a good prognosis for the treatment of
osteonecrosis of the femoral head in patients with sickle cell disease, even
when physical therapy is used alone, and it confirmed the safety of core
decompression in these patients. We found no significant difference in
outcomes between the two treatment arms and found both regimens to be safe.
There were no intraoperative complications and no cases of acute chest
syndrome in the first thirty days postoperatively. However, a superficial
infection did develop at the incision site in one Arm-A patient six weeks
after the hip core decompression. The average operative times and the minimal
blood loss were similar to those reported for patients without sickle cell
disease51.
Intertrochanteric hip fractures, which were reported in some studies of hip
core
decompression27,51,52,
were not seen in our study. The biopsy evidence of necrosis was comparable
with that documented in previous
reports20,27.
Previous studies of hip core decompression for osteonecrosis of the femoral
head have included older patients with etiologic factors including steroid
use, ethanol abuse, and renal disease; only a few patients with sickle cell
disease have been included in those reports. In studies of hip core
decompression for diagnoses other than sickle cell disease, as many as 56% of
hips required total hip replacement, some within a year after the
decompression21,28,51.
High failure rates after hip core decompression have been consistently
reported in hips with Steinberg Stage-III
disease8,51-54,
and the survival rates of nonoperatively treated hips have generally been
lower than that of hips treated with
decompression21,24,55.
In our randomized prospective study of Steinberg Stage-I, II, or III
osteonecrosis of the femoral head in patients exclusively with sickle cell
disease, we found a mean three-year hip survival rate of 82% in the hips
treated with hip core decompression and physical therapy and 86% in those
treated with physical therapy alone. The survival rate was independent of the
baseline stage of hip disease.
In addition to hip survival, clinical improvement in terms of pain and
function has been evaluated in some studies of hip core decompression for
osteonecrosis of the femoral head. Although we found no comparable reports on
patients with sickle cell disease, Mont et al. calculated rates of clinical
improvement of 64% after hip core decompression compared with 23% after
nonoperative treatment in their meta-analysis of twenty-four individual
reports on the treatment of osteonecrosis of the femoral head in other patient
populations22. The
definition of clinical improvement in these reports varied. In our study, pain
and function were assessed clinically with use of the
CHOHES50, the
validity and reliability of which were previously estimated to be 0.87 and
0.90. Physical therapists in many institutions were able to use this scale
effectively, which provided additional evidence of the face validity of the
CHOHES, given the correlation with pain diary data. Clinical improvement was
defined as a 15-point change from the baseline score and was demonstrated in
53% (eighteen) of the thirty-four patients for whom the data were available in
this study. The clinical response rate was equally high in the patients
treated with hip core decompression and physical therapy and those treated
with physical therapy alone, an observation that was supported by the
multivariate model, which did not show a significant treatment-arm effect. Of
note, the baseline radiographic stage was also not a predictor of clinical
improvement, and the rate of agreement between the staging of the
osteonecrosis of the femoral head performed by experts and that done at the
local centers in our study was comparable with the previously reported
variability in the classification of
osteonecrosis56.
Limitations of the Study
It is possible that our inability to demonstrate a difference in outcome
between the two groups was due to a type-2 statistical error. The sample size
required to detect this 2.4-point difference in mean change scores with a
power of 80% and a type-1 error of 0.05 would be more than 1000 in each group.
Although we are encouraged by the favorable outcomes in this study, given the
limited number of patients and the average follow-up period of only three
years, conclusions about the prognosis for patients with osteonecrosis of the
femoral head and sickle cell disease who are treated with these two treatment
regimens may be premature. Additional studies are needed to characterize the
long-term outcome of osteonecrosis of the femoral head in patients with sickle
cell disease and to compare these two interventions.
In conclusion, to our knowledge this is the first prospective randomized
trial of hip core decompression and physical therapy for the treatment of
early osteonecrosis of the femoral head in patients with sickle cell disease.
These results confirm other pilot data regarding the safety and efficacy of
hip core decompression in patients with sickle cell
disease57. We
believe that this is also the first study to evaluate physical therapy and
home exercises in addition to limited weight-bearing for the treatment of
osteonecrosis of the femoral head in patients with sickle cell disease. We
think that the mean improvement of the CHOHES scores (18.1 points in the group
treated with hip core decompression and physical therapy and 15.7 points in
the physical therapy group) was clinically relevant in both treatment groups.
Our data suggest that physical therapy alone is reasonable as a first
treatment modality for early stages of osteonecrosis of the femoral head in
patients with sickle cell disease, especially those who are at high risk for
perioperative complications. We caution that both interventions require close
follow-up with daily telephone calls to encourage compliance with the therapy
regimen. We believe that the CHOHES is a valuable tool for the evaluation of
pain, function, and physical disability in patients with sickle cell disease
who have osteonecrosis of the femoral head and that it can be effectively used
in the ongoing assessment of this population.
A table showing the physical therapy protocol and a figure showing the
CHOHES scoring form are available with the electronic versions of this
article, on our web site at
(go to
the article citation and click on "Supplementary Material") and on
our quarterly CD-ROM (call our subscription department, at 781-449-9780, to
order the CD-ROM). ?
Note: The authors acknowledge the following investigators who
participated in the National Osteonecrosis Trial in Sickle Cell Anemia:
Kenneth Ataga and Rupa Redding-Lallinger (University of North Carolina), Mauro
Grossi (Children's Hospital of Buffalo), Charles Daeschner (East Carolina
University), Nancy Olivieri (Hospital for Sick Children), James Casella (Johns
Hopkins Hospital), Michael Jeng (St. Jude's Children's Hospital), Abdullah
Kutlar (Medical College of Georgia), Zakia Al-Lamki (Sultan Qaboos
University), Charles Pegelow (University of Miami), Gloria Ramirez (St. Luke's
Hospital), Donald Mahoney and Kenneth McClaine (Baylor College of Medicine),
Bea Files and Charlton Davis (Children's Health Care Atlanta), Lakshamanan
Krisnamurti (Children's Hospital of Pittsburgh), and Clark Brown (Medical
University of South Carolina).