Although total hip arthroplasty remains the cornerstone of
surgical treatment of degenerative joint disease, dislocation continues
to be a relatively common complication of the procedure, second in
frequency only to late prosthetic loosening17.
Approximately 80 percent of all dislocations following total hip
replacement occur in a posterior direction, with a reported prevalence
of 0.7 to 5.5 percent following primary surgery and 5 to 20 percent
following revision4. The stability
of the artificial joint is influenced by numerous variables, including
soft-tissue laxity, component position, prosthetic features, surgical approach,
and comorbid conditions1,4,19.
It is generally appreciated that the range of motion of the artificial
hip prior to impingement increases with an increase in the diameter
of the femoral head because of the corresponding increase in the ratio
of the head and neck diameters18,19.
However, with the increasing recognition that osteolysis is a major
cause of long-term failure of total hip replacements, many authors
have advocated the routine use of femoral heads of smaller diameter
to reduce the volumetric wear rate6,8.
Nevertheless, it is widely believed that use of a head with a diameter
of twenty-two or twenty-six millimeters leads to an increased prevalence
of implant dislocation, especially when the operation was performed
through the posterior approach8,21.
Because the etiology of dislocation following total hip arthroplasty
is a multifaceted problem, we developed a cadaveric model of the
hip joint to allow experimental simulation of activities known to challenge
joint stability11,20. This model
allows for the systematic examination of the effect of individual
variables on prosthetic joint stability.
In our study, we tested the limits of a cadaveric hip joint in
flexion and adduction in a range of positions commonly associated
with rising from a chair. Using this model, we examined the effect
of the size of the femoral head of one design of prosthetic hip
joint on the range of motion prior to impingement and posterior
dislocation as well as on the site of impingement.
Six fresh cadaveric specimens were retrieved at postmortem dissection
from donors ranging in age from forty-two to eighty-six years (average,
sixty-one years) at the time of death. All specimens were dissected
free of soft tissue and prepared for implantation of a cementless
hip prosthesis with use of standard surgical technique. An uncemented femoral
component (Meridian; Howmedica, Rutherford, New Jersey) was implanted
in neutral position in relation to the neck of the native femur (approximately
15 degrees of anteversion), preserving leg length to within three
millimeters. The femoral prosthesis had a cylindrical neck of 11.8 millimeters
in diameter and a neck-shaft angle of 132 degrees. In every instance,
the femoral component was implanted with a modular head without
a skirt. A porous-coated acetabular component (Vitalock Cup; Howmedica)
with a neutral polyethylene liner was implanted into the acetabulum in
20 degrees of anteversion and 45 degrees of inclination. The bearing
surface of the cup was hemispherical and had a chamfered edge; the
center of rotation of the bearing surface was not displaced with
respect to the center of the external spherical contour of the component.
During implantation, the orientation of the cup with respect
to the pelvis was monitored by measuring the positions of standard
landmarks on the osseous pelvis and the implant. All measurements were
performed with a three-dimensional (3-D) digitizing arm (MicroScribe
3D; Immersion, San Jose, California). An anatomical coordinate system
was defined by the spatial coordinates of the anterior superior
iliac spines, the pubic symphysis, the ischial tuberosities, the
tip of the sacrum, and the midpoint of the fifth lumbar vertebral
body. The orientation of the acetabulum was defined by three landmarks
located on its osseous rim. The inclination of the acetabular component
was calculated as the angle between the transverse plane and the
rim of the acetabular component in the anterior-posterior projection.
The anteversion of the acetabular component was calculated as the
angle between the axis perpendicular to the frontal plane and the
plane of the acetabular component in the superior projection.
Before testing, each pelvis was mounted on the actuator of a
biaxial servohydraulic testing machine (Bionix; MTS, Minneapolis,
Minnesota) in an orientation simulating the position of the pelvis
with a person standing. The pelvis was positioned so that a plane
passing through both anterior superior iliac spines and the pubic
symphysis would be oriented vertically (Fig. 1). A custom fixture attached to
the superior surface of the fifth lumbar vertebra was used to adjust
the position and orientation of the pelvis with respect to the femur
such that the axis of rotation of the actuator passed through the
center of the acetabulum. The angle of hip flexion was measured
with use of pendulum rotational transducers attached to the distal
end of the femur. Hip flexion was increased at 2 degrees per second
from a starting position of 90 degrees until dislocation occurred.
During all experiments, the hip joint was maintained in 10 degrees
of internal rotation. A starting position of 90 degrees of flexion
and 10 degrees of internal rotation was chosen because pilot studies
had shown that every component configuration tested was stable in
that position. Hip adduction was altered by rotating the pelvis
by means of the biaxial actuator of the testing machine. Neutral
adduction was defined by the position in which the femoral shaft was
perpendicular to the plane of the pelvis (Fig. 1).
A second fixture, located on a linear sliding table, was used
to load and orient the femur in contact with the pelvis. The axial
rotation of the femur was controlled with a rod fixed within the
distal portion of the medullary canal. The distal end of the rod was
secured to a sliding bearing that was modified to allow flexion
and axial displacement of the femur in a fixed amount of internal
or external rotation.
Physiological hip-loading was reproduced with a system of cables
simulating the seven major muscle groups acting when a person rises
from a low chair; these included the rectus femoris, hamstrings,
adductor longus, gluteus maximus, gluteus medius, iliopsoas, and
short external rotators. These muscles were selected on the basis
of electromyographic studies performed in healthy individuals during
sit-to-stand activities2,14. Cables
were attached to the femur at osseous landmarks corresponding to
the insertion site of each muscle group9 and
passed through guides mounted at the point of origin of each muscle
in the pelvis. Each cable was then routed over a pulley mounted
posterior to the pelvis and attached to a free weight. To facilitate testing
and to prevent fracture of the cadaveric specimens during dislocation
episodes, the force applied to each cable was restricted to 10 percent of
the predicted force of contraction of each muscle2,10,14.
Using the electromyographic data, the relative muscle activity
(activation coefficient in newtons per centimeter squared) during
the action of rising from a seated position was estimated. The force developed
by each muscle was calculated as the product of its physiological
cross section (in square centimeters) and its activation coefficient (its
force of contraction per unit of cross-sectional area) (Table I).
At the start of each experiment, the hip joint was mounted in
90 degrees of flexion; the pelvis was then displaced until the point
of complete dislocation of the artificial joint. Once dislocation
occurred, the femur and its mounting platform moved independent
of the pelvis in a posterior direction under the action of the loading
cables. To prevent damage to the cadaveric specimens and the loading
apparatus, the displacement of the femur was limited with motion
stops attached to the guide rails of the sliding platform.
During testing, each orthogonal component of the angular position
of the hip joint (flexion-extension, abduction-adduction, and external-internal rotation)
was monitored with rotational transducers mounted on the sliding
bearing and the distal portion of the femur. All data were continuously sampled
at twenty-four hertz with use of a computerized data-acquisition
system (Data Translation, Marlborough, Massachusetts).
To facilitate detection of impingement, the rim of the acetabular
liner, the osseous acetabular rim, and the proximal part of the
femur were coated with conductive foil, which was connected to an external
power source. With this modification, impingement was defined as
the point where contact was recorded between the implant and either
the liner or the osseous pelvis. Dislocation was defined as the
point at which loss of electrical continuity between the neck of
the femoral implant and the liner or the acetabulum occurred. With
continuous monitoring of the voltage between the femoral prosthesis
and the pelvis with the transducers, the positions at the instants
of impingement and dislocation of the femur were automatically recorded.
The entire experiment was performed at 10, 20, and 30 degrees
of adduction with femoral heads of four standard diameters (twenty-two,
twenty-six, twenty-eight, and thirty-two millimeters).
The range of motion of the hip in flexion increased with increases
in the diameter of the prosthetic head at adduction angles of 10
degrees (p = 0.001), 20 degrees (p < 0.001), and 30 degrees
(p = 0.003) (Fig. 2).
With a thirty-two-millimeter head, dislocation occurred at 120.0 ± 4.2 degrees of flexion in 10 degrees of adduction, 112.7 ± 3.1 degrees in 20 degrees of adduction, and 103.8 ± 5.9 degrees in 30 degrees of adduction. When the head diameter was reduced from thirty-two to twenty-eight millimeters, the flexion angle at dislocation was reduced by only 1.2 ± 2.3 degrees in 10 degrees of adduction (p = 0.26), 2.7 ± 3.8 degrees in 20 degrees of adduction (p = 0.15), and 3.3 ± 7.1 degrees in 30 degrees of adduction
(p = 0.31). None of these differences was significant at the 5 percent
level.
Greater changes were observed when the head size was reduced
from twenty-eight to twenty-six millimeters. In this case, the loss
of flexion was 4.4 ± 1.7 degrees in 10 degrees
of adduction (p = 0.001), 4.9 ± 4.5 degrees
in 20 degrees of adduction (p = 0.046), and 3.3 ± 2.8
degrees in 30 degrees of adduction (p = 0.034). An additional reduction
in head size from twenty-six to twenty-two millimeters led to an
additional loss of flexion prior to dislocation of 3.1 ± 1.8 degrees in 10 degrees of adduction (p = 0.007),
3.9 ± 0.7 degrees in 20 degrees of adduction
(p = 0.00004), and 3.0 ± 2.1 degrees in 30
degrees of adduction (p = 0.013).
With a twenty-two-millimeter head, the angle of hip flexion at
dislocation was only 111.3 ± 1.6 degrees
at 10 degrees of adduction, whereas the corresponding values at
20 and 30 degrees of adduction were 101.2 ± 1.0
degrees and 94.2 ± 2.5 degrees, respectively.
With the hip in 10 degrees of adduction, the range of stable joint
motion increased by 3.1 ± 1.8 degrees with
a twenty-six-millimeter head (p = 0.008), 7.5 ± 1.7
degrees with a twenty-eight-millimeter head (p = 0.0001), and 8.7 ± 3.3 degrees with a thirty-two-millimeter head
(p = 0.001). Similar changes were observed in 20 and 30 degrees
of adduction (Fig. 3).
The primary mechanism for dislocation changed with the size of
the femoral head. Three different mechanisms of dislocation were
noted during the experimental runs (Fig. 4): impingement of the prosthetic
femoral neck on the cup liner (Group A), impingement of the osseous
femur on the osseous pelvis (Group B), and spontaneous dislocation
(Group C). A transition from prosthetic to osseous impingement occurred with
increasing diameter of the femoral head (Fig. 5) and adduction of the hip (Fig. 6). Whereas the
primary mechanism of dislocation with the twenty-two-millimeter
head was impingement between the prosthetic femoral neck and the
acetabular liner, the most frequent cause of dislocation with the
thirty-two-millimeter head was impingement between the osseous femur
(the lesser trochanter) and the pelvis (the ischium).
The diameter of the femoral head also affected the range of hip
flexion prior to impingement (Fig. 7). With use of the thirty-two-millimeter
head, the average range of flexion until impingement was 113.3 ± 2.4 degrees in 10 degrees of adduction, 106.3 ± 1.8 degrees in 20 degrees of adduction, and 100.2 ± 2.0 degrees in 30 degrees of adduction. Decreasing
the head size from thirty-two to twenty-two millimeters caused an
average decrease in the range of flexion prior to impingement of
8.6 ± 2.4 degrees in 10 degrees of adduction
(p < 0.001), 6.8 ± 2.9 degrees in 20
degrees of adduction (p = 0.002), and 8.1 ± 2.9
degrees in 30 degrees of adduction (p = 0.001) (Fig. 8). Decreasing
the head size from thirty-two to twenty-eight millimeters also led
to losses in the range of flexion prior to impingement: 1.7 ± 1.5 degrees in 10 degrees of adduction (p = 0.042),
3.0 ± 2.4 degrees in 20 degrees of adduction
(p = 0.027), and 2.3 ± 1.0 degrees in 30
degrees of adduction (p = 0.003).
We refer to the range of flexion of the joint during subluxation
(impingement until frank dislocation) as the safety margin. With
a twenty-two-millimeter femoral head, the safety margin was 6.6 ± 2.3 degrees in 10 degrees of adduction (p = 0.0008),
2.3 ± 0.4 degrees in 20 degrees of adduction
(p = 0.00003), and 2.1 ± 1.4 degrees in 30
degrees of adduction (p = 0.013) (Fig. 9). With a twenty-six-millimeter
femoral head, the safety margin was 7.2 ± 1.4
degrees in 10 degrees of adduction (p = 0.00006), 4.3 ± 2.4 degrees in 20 degrees of adduction (p = 0.006),
and 2.2 ± 1.5 degrees in 30 degrees of adduction
(p = 0.015). With a twenty-eight-millimeter femoral head, the safety
margin was 7.3 ± 2.1 degrees in 10 degrees
of adduction (p = 0.0004), 6.7 ± 2.8 degrees
in 20 degrees of adduction (p = 0.002), and 2.9 ± 2.3
degrees in 30 degrees of adduction (p = 0.028). With a thirty-two-millimeter
femoral head, the safety margin was 6.8 ± 2.5
degrees in 10 degrees of adduction (p = 0.001), 6.3 ± 2.3 degrees in 20 degrees of adduction (p = 0.001),
and 3.7 ± 5.1 degrees in 30 degrees of adduction
(p = 0.137).
Although many theoretical and experimental studies performed ex
vivo have suggested that the size of the femoral head affects
the range of stable joint motion, retrospective clinical studies
have led to contradictory conclusions with respect to the relationship
between head size and dislocation after hip replacement5,7,8,19,23. Woo and Morrey examined
factors associated with dislocation after hip replacement in a retrospective
review of 3353 hip replacements performed at the Mayo Clinic over
a three-year period23. Operations
performed with hip prostheses of five designs with head sizes of
twenty-two, twenty-eight, and thirty-two millimeters were included
in the study. The risk of dislocation was most strongly associated
with previous surgery, a posterior surgical approach, and avulsion
of the greater trochanter following a trochanteric osteotomy. The prevalence
of dislocation was 2.9 percent (fifty-five of 1910) in patients
who had a twenty-two-millimeter head compared with 3.3 percent (sixteen
of 486) in patients with a thirty-two-millimeter head. Though this
difference was not found to be significant, a statistical analysis
of the power of the comparisons performed in that study indicates that
the rate of dislocation of the larger head could have been 2.4 percent
larger than that of the twenty-two-millimeter head (that is, it
could have been 5.3 percent [2.9 plus 2.4 percent]) and the difference
between the dislocation rates of the two devices would still not
have been significant, even though 5.3 percent is almost twice as
large as 2.9 percent. Similarly, a sample of 3720 patients would
be required to detect a 2 percent difference between the dislocation
rates of two different devices or methods of treatment with a power
of 80 percent. For this reason alone, it is highly unlikely that
the retrospective analysis reported by Woo and Morrey demonstrated
the true effect of prosthetic head size on the rate of dislocation.
In their retrospective analysis, Woo and Morrey also identified
more than ten confounding variables that appeared to affect the
risk of dislocation in their patient population. Thus, it appears
that the true contribution of femoral head size can be isolated
only through studies employing randomized clinical trials7.
Several investigators have performed biomechanical experiments
in an attempt to isolate the effect of single variables on joint
stability. These investigators have examined primarily the roles
of head size, extended liners, and skirted modular heads15. In most previous studies, the range
of motion of the joint has been measured to the point of impingement
under unloaded conditions with use of prosthetic components tested
in isolation or after implantation in cadaveric specimens. Using
a cadaveric pelvis mounted on a three-dimensional protractor, Amstutz
et al. assessed the range of motion of various prosthetic hip components1. With the hip joint maintained in
neutral abduction and internal rotation, the Charnley prosthesis allowed
only 80 degrees of flexion compared with 96 degrees with the M¸ller
prosthesis. This dramatic difference was attributed to the larger
head diameter (thirty-two millimeters) and head-neck ratio (1.98)
of the M¸ller prosthesis compared with the Charnley design (twenty-two
millimeters and 1.74). Chandler et al. utilized the same model to
compare the ranges of motion of natural and artificial hips3. They found that larger heads delayed
neck-socket contact, leading to an increased range of motion prior
to prosthetic impingement. They also noted that increasing the head
diameter caused a transition from impingement between the prosthetic neck
and acetabular liner to osseous impingement, with the greater trochanter
coming into contact with the pubic and iliac bones to limit internal
rotation and flexion. Once osseous impingement occurred, increasing
the head size did not lead to greater joint motion prior to impingement.
These results are in agreement with the findings of our study.
The results of our experiments demonstrate that the size of the
femoral head over the range of twenty-two to twenty-eight millimeters
affects the posterior stability of the artificial hip. However, the
range of flexion did not significantly increase between the twenty-eight
and thirty-two-millimeter heads, primarily because joint motion
was limited by osseous, not prosthetic, impingement. It must also
be noted that 120 and 113 degrees of flexion achieved with the femur
in 10 and 20 degrees of adduction, respectively, are extreme positions
that are only rarely seen after total hip arthroplasty.
Whereas impingement between the neck of the femoral prosthesis
and the acetabular liner occurred most frequently with the twenty-two-millimeter
head, use of the thirty-two-millimeter head was associated with
an increased prevalence of impingement between the osseous femur
and the pelvis. Also, because of an increased frequency of osseous
impingement, the head size had less of an effect on the range of
motion in positions of increased hip adduction. These results suggest
that, with this particular prosthesis, a femoral head of twenty-eight
millimeters in diameter can increase the range of motion after total
hip replacement, which may be beneficial when additional factors compromise
joint stability. As it is known that femoral version affects the
range of motion of total hip components in directions associated
with posterior impingement, the results obtained in this experiment
may pertain only to prostheses with this specific geometry and femoral
version. This is especially true for the results obtained with the
twenty-two-millimeter head, the dislocation of which almost always
occurred secondary to prosthetic impingement.
Our experimental model had several unique qualities that allowed
us to isolate the effect of individual factors on the stability
and the range of motion of the prosthetic hip joint. Using this
model, we examined the stability of the hip in positions most commonly
associated with dislocation, with application of the muscle loads
normally acting on the hip joint. Previous experimental studies1,3 have not included muscle-loading
and have only addressed the range of motion of the prosthetic joint
to the point of impingement. In contrast, our simulation allowed
us to assess the influence of head size on joint stability without
assuming that the range of motion of the joint at impingement is directly
related to the position of the joint at dislocation. Additional
studies performed with this model have shown that other variables
(for example, femoral anteversion12)
strongly influence the range of motion of the hip prior to impingement
but have much less effect on the position of the joint at dislocation22. These studies suggest that future
experimental models simulating instability of normal and prosthetic
hips should incorporate muscle forces and should allow the joint
to subluxate and to freely dislocate once osseous or prosthetic
impingement occurs.
Some limitations of the model must also be addressed. Soft-tissue
tension has been recognized as an important factor in stability
of the artificial hip joint4,6,13,16,19,20.
Preparation of the cadaveric specimens required removal of the hip
joint capsule, which negated any effect that soft-tissue tension
may have had on hip stability. Tension produced in vivo from
surrounding soft tissues may allow a hip prosthesis to go through
a greater range of motion prior to dislocation. Also, in the clinical
setting, the artificial hip must be stable in both the anterior and
posterior directions. Because only posterior dislocation was evaluated
in this model, we could not assess the effect of a change in femoral
head size on the anterior stability of the artificial hip.
We recommend that the selection of the femoral head for total
hip arthroplasty be individualized to address each patient's risk
of dislocation, with consideration also given to excessive joint
wear. In younger and/or more active patients with a life expectancy
of more than twenty years, we recommend a smaller femoral head (for
example, twenty-six millimeters), with careful attention to other
technical factors influencing joint stability. Conversely, in more
sedentary, elderly patients with relatively weak abductors, use
of a twenty-eight-millimeter head will allow a greater stable range
of motion in comparison with a smaller head. In this study, we have
shown that, with this particular prosthesis, an additional increase
in head size to thirty-two millimeters does not significantly increase
the range of motion.