Improved surgical techniques and multidisciplinary rehabilitation protocols
that involve coordination among surgeons, physical therapists,
anesthesiologists, and social services personnel have led to excellent knee
function and range of motion in a large percentage of patients following total
knee arthroplasty. Nevertheless, there remains a small number of patients with
persistent dysfunction that is difficult to
treat1-4.
Functional problems following total knee arthroplasty may be incapacitating as
a result of persistent
pain5,
instability6, and a
limited range of
motion7. It has been
shown recently that there is a direct correlation between a decreased range of
motion following surgery and a lower perceived quality of life as evaluated
with use of the Short Form-36 health survey
questionnaire8.
Continued dysfunction for any reason ultimately leads to decreased patient
satisfaction.
There is controversy about treatment methods for patients for whom initial
rehabilitation efforts are unsuccessful following total knee arthroplasty. The
reported efficacy of both noninvasive and invasive treatment modalities has
been variable, with the percentage of patients obtaining improvement ranging
from 0% to
90%3,9-12.
Patients who have continued dysfunction despite initial rehabilitation efforts
may require revision surgery. However, patients who have well-aligned,
well-fixed prosthetic components will likely not benefit from a complete
revision. Treatment of arthrofibrosis, scarring, soft-tissue contractures,
and/or other soft-tissue dysfunction should involve less invasive treatment
protocols before surgical options are considered. Nonoperative treatment
modalities for restoring the range of motion include intensive rehabilitation
protocols, static or dynamic splinting, injections, and application of serial
casts13.
Manipulation with the patient under anesthesia and invasive procedures,
including arthroscopic débridement, open débridement with or
without polyethylene exchange, and complete component revision, have been
utilized when initial nonoperative rehabilitation efforts have failed.
As a result of the variability in the functional limitations experienced by
patients and the inconsistency in the success of present, commonly used
rehabilitation modalities, additional treatment options need to be considered
and evaluated. We are reporting our experience with diagnosing the underlying
causes of dysfunction following total knee arthroplasty and the level of
success that we have had with use of various rehabilitation and treatment
modalities.
Diagnostic Methods
Patients for whom initial rehabilitation efforts following total knee
arthroplasty were considered to have failed because of continued pain and/or
functional limitations underwent a careful radiographic and clinical
evaluation. Radiographic studies were used to identify component loosening,
malalignment, or other problems (for example, retained bone cement) that
indicated the need for revision surgery. When infection was suspected,
aspiration and cultures were utilized to verify the diagnosis and to identify
the organism, thus ensuring proper antibiotic therapy and timely surgical
treatment14.
Patients for whom standard rehabilitation efforts had been unsuccessful and
who had no evidence of radiographic abnormalities were thoroughly assessed to
identify any underlying soft-tissue problems associated with their dysfunction
(Fig. 1). The range of motion
was assessed at all clinical evaluations prior to and following the total knee
arthroplasty. Standard flexion and extension measurements were performed with
use of a goniometer. Limb length was measured with a tape measure when a
limb-length discrepancy was suspected.
Muscle strength was measured with use of manual isokinetic strength testing
and graded on a scale of 0 to
515. A grade of 0
corresponded to no contraction of the muscle and 100% loss of strength,
whereas a grade of 5 indicated active movement against gravity with full
resistance and no loss of strength. All measurements were made by two of us
(A.B. and T.M.S.), who agreed with each other regarding the grade assessments
>95% of the time.
Video gait analysis was conducted with use of an eightcamera motion
analysis system (Motion Analysis, Santa Rosa, California) combined with two
central force-plates (Advanced Mechanical Technology, Watertown,
Massachusetts). This system was utilized to measure body kinematics, muscle
activation, and ground reaction force along a 10-m, level-surface walkway.
Twenty-six reflective markers (Helen Hayes marker set) were placed on both
sides of the patient at the hip, knee, ankle, and foot. After the setup of
markers for movement recording, electrodes were placed for surface
electromyography. The force-plates and electromyography system were
synchronized with the motion analysis system to measure ground reaction forces
and activities of the lower-extremity muscles. The subjects were instructed to
walk several passes at a self-selected speed until ten consistent force-plate
recordings were collected for each side. Throughout the observation period,
the sampling rate for the data points was 60 Hz. The data were processed with
use of standard OrthoTrak software (Motion Analysis). The analyzed parameters
were walking velocities, ground reaction vectors, joint moments, power, knee
flexion, and ranges of ankle dorsiflexion during the stance phase as well as
during the loading response.
Patient Cohort
On the basis of a thorough clinical evaluation performed with use of the
previously mentioned diagnostic methods, functional problems and stiffness of
the knee after total knee arthroplasty were diagnosed in 106 patients (108
knees). Initial rehabilitation efforts, which included manipulation under
anesthesia in some cases, had failed for all of these patients. The patients
had shown minimal improvement in terms of both function and pain relief at a
minimum of two months postoperatively. Common symptoms included quadriceps
fatigue pain, anterior knee pain, abnormal gait, back pain, limp, difficulty
with walking long distances, and an inability to participate in non-strenuous
sports or recreational activity. Patients also reported difficulty with
sitting in and rising from a chair, descending and ascending stairs, and
sexual relations. There were sixty-four women and forty-two men, and the mean
age was fifty-three years (range, nineteen to seventy-seven years). Seventeen
of the patients had the functional problems following revision total knee
arthroplasty, and all others had them after primary total knee
arthroplasty.
All individuals demonstrated gait abnormalities that were associated with
one or more biomechanical dysfunctions. The six most prevalent categories of
dysfunction were flexion contracture, knee flexion deficit, quadriceps
weakness, limb-length discrepancy, peroneal nerve entrapment, and functional
malalignment of the distal joints. Sixty-six patients (sixty-eight knees) were
diagnosed as having a knee flexion contracture; forty-three patients
(forty-four knees), muscle tightness; twenty-eight patients (twenty-eight
knees), quadriceps muscle weakness; seven patients (seven knees), a
limb-length discrepancy of 1 to 2.5 cm; four patients (four knees), functional
malalignment; and fourteen patients (fourteen knees), peroneal nerve
dysfunction. Some of the patients had more than one functional problem.
The specific rehabilitation and physical therapy modalities were customized
for each patient, and treatments with these modalities were not mutually
exclusive. Some of these treatments included the use of a customized knee
device (seventy-nine knees), a JAS device (Joint Active Systems, Effingham,
Illinois) (thirty knees), botulinum toxin injections (nine knees), electrical
stimulation (forty-one knees), and peroneal nerve release (fourteen knees). In
general, the physical therapy regimens lasted for three to six months (mean,
4.6 months) (Fig. 2). All
rehabilitation efforts failed for seven patients, who subsequently received
surgical treatment, including arthroscopic débridement, open
débridement with or without polyethylene exchange, and complete
component revision.
Nonoperative Treatment
Connective tissue is viscoelastic in nature, and under tension it responds
by reaching either the elastic or plastic deformation state. In elastic
deformation, tissue can revert to its original length after the force is
removed. In the plastic deformation state, tissue will maintain the newly
elongated length after elimination of the force. The goal of therapeutic
stretching is to achieve permanent length changes without requiring new tissue
growth by plastically deforming tissues that have stiffened or
shortened16.
Custom Knee Devices
Seventy-eight patients (seventy-nine knees) followed a daily outpatient
rehabilitation algorithm using a fitted custom knee device
(Fig.
3)17.
The device is hinged at the knee and has elastic bands (Thera-Band; The
Hygenic Corporation, Akron, Ohio), which produce a flexion moment. The
patients were instructed how to position the brace for maximum tolerated force
and were expected to utilize the custom knee device for thirty to forty
minutes two or three times a day. The brace was removed at all other
times.
All patients who were treated with the custom knee device had a knee
flexion deficit, which was defined as a flexion angle of <90° and which
resulted in functional deficits in the ability to rise from and sit in a
chair, sit for prolonged periods of time, ascend and descend stairs, and
engage in sexual activities. These problems were mainly related to tightness
of the rectus femoris muscle, patellar tracking problems, and tightness or
inflammation of the patellar tendon. In addition to the patients' use of the
custom knee device, physical therapists supervised inferior patellar, patellar
tendon, quadriceps tendon, rectus femoris, and knee joint mobilization efforts
to further increase knee flexion (Fig.
4).
Patients with a knee flexion contracture used a customized device to
increase knee extension, which was similar to the custom knee device used to
increase flexion and which was also applied for thirty to forty minutes three
times a day (Fig. 5). Knee
flexion contracture leads to problems with walking, anterior knee pain,
extensive patellar wear, and patellar tendinitis. An overall excellent result
was obtained in seventy-one knees (90%) treated with the custom knee device.
The mean Knee Society score at the time of final follow-up was 91 points
(range, 55 to 100 points), and the total arc of motion improved by a mean (and
standard deviation) of 24.7° ± 18.3°. After obtaining minimal
improvement with use of the custom knee device, three patients required
arthroscopic lysis of adhesions, one patient had an open débridement
with polyethylene exchange, and one patient had complete component revision.
Some patients chose not to undergo additional surgery and ended their
rehabilitation efforts despite continued range-of-motion deficits.
JAS (Joint Active Systems) Device
Twenty-nine patients (thirty knees) with a soft-tissue contracture after
total knee arthroplasty or trauma to the joint used the JAS (Joint Active
Systems) knee device to treat both flexion and extension limitations
(Fig. 6). The JAS device is a
patient-directed, bidirectional orthosis that can be used conveniently in a
home setting. It incorporates stress relaxation and static progressive
stretch, while simulating some of the manual techniques used by physical
therapists in the clinical setting. Patients with heterotopic ossification,
peroneal nerve palsy, or true osseous blocks were not treated with this
device. Fifteen men and fourteen women with a mean age of fifty-two years
began using the JAS device at a mean of 14.1 ± 9.7 weeks after the
index surgery or injury.
Patients were instructed regarding the application of the device and the
protocol for its use as described by Bonutti et
al.18. After they
placed the orthosis on the affected limb at the current limit of motion, the
patients were directed to increase the stretch to the extent that they could
tolerate by turning the ratchet on the device and holding that position for a
period of five minutes. They then increased the stretch to the extent that
they could tolerate and held the new position for another five minutes.
Patients were instructed to stretch to the extent that they could tolerate,
without causing excruciating pain, and they continued this incremental
stretching for the entire thirty-minute treatment session.
All but two patients who underwent rehabilitation with use of the JAS
device had a favorable outcome, with an increase in the range of motion, and
two patients made no progress in at least one direction. The patients gained a
mean of 7.4° ± 8.1° of additional extension and 15.1°
± 12.3° of additional flexion. The total mean increase in the
flexion-extension arc was 22.5° ± 16.3°, from a mean of
85.4° ± 22.2° before utilization of the JAS orthosis to a mean
of 107.9° ± 16.8° after a mean of 9.4 ± 7.8 weeks
(range, three to thirty-three weeks) of treatment. Of the two patients who did
not have an improvement in the range of motion, one required arthroscopic
lysis of adhesions and one had an open débridement without polyethylene
exchange.
Botulinum Toxin Injections
Eight patients (nine knees) who had rigidity of the hamstring and/or
gastrocnemius muscle as the primary cause of a continued flexion contracture
despite initial physical therapy were treated with botulinum toxin type-A
injections. These injections produce a neuromuscular blockade for
approximately three months. The underlying mechanism of action of botulinum
toxin is at the cellular level. By acting selectively on peripheral
cholinergic nerve endings, it leads to chemodenervation and local
paralysis19. The
temporary paralysis and reduction in functional muscle spasticity or muscle
tone provided by botulinum toxin promote better motor balance across joints;
improve walking ability; and, when combined with intensive physical
rehabilitation, allow the therapists to progress the range of motion beyond
previous
limitations20,21.
The decision regarding which muscles to inject was based on the patient's
reported feeling of tightness and spasm as well as on the findings of the
clinical examination. Patients who lost knee extension as the hip joint was
being flexed to 45° were considered candidates for botulinum injections
into the hamstring muscles. The hamstring muscles were palpated as the knee
was extended to determine whether the medial or lateral hamstrings required
treatment. The decision to inject the gastrocnemius muscle was based on a loss
of the range of knee extension as the ankle was dorsiflexed compared with the
range in the plantigrade position.
The mean time from the index surgery to treatment with botulinum toxin
injections was eighteen months (range, three to eighty-one months). Following
the primary total knee arthroplasties, we injected the medial hamstrings in
eight cases (seven patients) and both the medial and the lateral hamstrings in
one patient. Two patients also received injections into both heads of the
gastrocnemius muscle. All injections were administered by one of us (M.A.M.).
The patient was placed in a prone position, and a 25-gauge needle was used to
inject the muscle belly with a dilution of 50 units/mL. The injections into
the medial hamstring muscles consisted of 100 units of botulinum toxin type A
distributed evenly among four sites, those into the lateral hamstrings
consisted of 100 units of toxin distributed evenly among three sites, and
those into the gastrocnemius muscle consisted of a total dose of 50 units of
toxin injected into two sites in both the medial and the lateral heads of the
muscle belly (Fig. 7).
After receiving the injections, the patients started a rehabilitation
regimen that consisted of inpatient and outpatient treatment for eight
continuous weeks. Intensive physical therapy combined with careful knee,
quadriceps muscle, and inferior patellar mobilization and stretching of the
rectus femoris muscle and the patellar tendon was provided three, four, or
five times per week. The patients who received gastrocnemius injections were
encouraged to sleep with the foot in a splint to maintain maximum dorsiflexion
stretch on the muscle.
There were no serious complications directly associated with the botulinum
treatment. A transient flu-like condition developed in one patient and lasted
for two days. However, the patient recovered without additional difficulties
and showed no later effects. Another patient had redness and swelling at the
site of the injection for several days after treatment; this cleared up by one
week.
Eight of the nine knees treated with the botulinum injections achieved
extension within 5° of the neutral position. The range of motion increased
by a mean of 42° (range, 10° to 75°; p < 0.001) at a mean of
thirty-eight months (range, twenty-four to fifty-eight months) following the
injections. The Knee Society scores improved by a mean of 35 points (range, 15
to 53 points; p < 0.001) from the time prior to the botulinum treatment to
the final follow-up examination. All patients had a good or excellent clinical
outcome as indicated by a Knee Society score of =80 points, except for one
patient with a score of 76 points. That patient had a residual knee flexion
contracture of 10° but was satisfied with the result and declined further
treatment.
Electrical Stimulation
Forty patients (forty-one knees) who had an extension lag and were included
in the previously discussed cohorts (those treated with botulinum toxin
injections and/or a custom knee device) also received electrical stimulation
during their physical therapy sessions to augment strength training. The
stimulation was administered while the patient underwent soft-tissue
mobilization and exercises with use of a leg-press to activate the quadriceps
and inhibit the flexors (Fig.
8). Throughout the session, the electrical stimulation was set at
a 7 to 10-sec on-time, a 15 to 20-sec off-time, a frequency of 70 to 90
pulses/sec, a 400-µsec pulse duration, and 90 Hz.
Manipulation Under Anesthesia
Patients who had a range of motion of =90° at six weeks following
the total knee arthroplasty were candidates for manipulation under anesthesia
(Fig. 9). These patients showed
no improvement or a reduction in both flexion and extension despite initial
rehabilitation efforts. Like the other nonoperative treatments, manipulations
were not utilized for patients who had component malposition or failure,
incorrect component sizing, joint line displacement, or inadequate bone
resection. Manipulations were not performed more than three months following
total knee arthroplasty because of decreased effectiveness and the risk of
fracture after that time. Patients who continued to show a limited range of
motion after three months underwent extensive radiographic and clinical
evaluation and were treated with a customized rehabilitation regimen with use
of one or more of the previously mentioned nonoperative treatment
modalities.
The criteria used to select candidates for manipulation under anesthesia in
this study differed from those reported in other studies. Keating et al.
selected patients for manipulation as early as two months following total knee
arthroplasty, and the procedure was utilized as late as forty-four weeks
postoperatively12.
Fox and Poss treated patients as early as two weeks after
surgery22, and
Shoji et al. reported that their patients received manipulations at ten days
following
surgery23. The
range-of-motion criteria used in other studies also varied. For example,
Brassard et al. selected patients with a range of motion limited to
<75°24, and
Esler et al. treated patients in whom flexion had failed to increase to
>80° following the initial
physiotherapy25.
The reported outcomes of manipulations under anesthesia have also been
variable. In the study by Keating et al., patients had a mean increase in
flexion of 35° after a mean of five years of
follow-up12.
Keating et al. also reported no significant difference in the improvement in
flexion between patients who had been treated within twelve weeks following
surgery and those who had had the manipulation at more than twelve weeks. In
contrast, a study by Yercan et al. suggested that the timing of the procedure
does affect the final
outcome26. They
found the results in patients who had received the manipulations less than
three weeks postoperatively to be better than those in patients who had been
treated between three weeks and three months postoperatively. They reported an
improvement of 47° in the range of motion at the time of final follow-up.
In contrast to both of these studies, Fox and
Poss22 reported
that manipulations ultimately did not improve the range of motion, which
initially increased by 37° but the increase was only 17° one week
later. Furthermore, at one year, there was no difference in the range of
motion between the patients who had received manipulations and those who had
not. As a result of the variability in selection criteria and outcomes
reported in the literature, a prospective study to further evaluate the
efficacy of this procedure is currently being conducted at our
institution.
Surgical Treatment
Functional problems, persistent pain, and arthrofibrosis after total knee
arthroplasty are debilitating complications. The underlying etiology is often
multifactorial, and little information regarding these problems is
known27. Although
the etiology of these complications is poorly understood, several surgical
procedures have been proposed as viable treatment modalities. Open operative
procedures such as a complete revision of the prosthesis or a simple component
exchange have been described. More recently, and with the use of arthroscopic
surgery, there have been numerous attempts to address these types of
complications with minimally invasive techniques.
Peroneal Nerve Release
Peroneal nerve release with both proximal and distal decompression at the
fibular head was performed after the total knee arthroplasty in fourteen
patients (fourteen knees). All patients had had constant burning or shooting
pain in the dorsum of the foot, which was made worse by physical therapy. In
addition, electromyographic and nerve conduction velocity studies demonstrated
electrophysiological abnormalities of the peroneal nerve in all fourteen
patients. The possible causes of the peroneal nerve palsy appeared to be
direct traction on the nerve or tension on the surrounding soft tissues
created during the surgery. Other, indirect mechanisms of injury might have
included compression from tight dressings or possibly a combination of these
factors.
The mean age of the patients was sixty-three years. None had an underlying
neurological disorder. In each case, the peroneal nerve was released at two
possible entrapment points with use of a standard five-step surgical technique
(Fig. 10). First, a short
oblique incision was made at the level of the neck of the fibula in line with
the course of the nerve. Next, the superficial fascia of the leg was incised
and divided over the peroneal nerve. The third step was to divide the fascia
over the muscles medially. At this stage, the underlying fascial band that
passes over the nerve was exposed and then sectioned to release the first
potential entrapment point. The final step involved isolating and then cutting
the intermuscular septum between the anterior and lateral compartments. This
last step releases the second common point of entrapment. In some patients,
the anterior compartment fascia was also released longitudinally in a
subcutaneous fashion.
Following peroneal nerve release, all patients received intensive physical
therapy three, four, or five times a week for eight weeks, with use of either
the custom knee device or the JAS device. At the end of these eight weeks,
thirteen of the fourteen patients had achieved full extension within 5°.
They were completely symptom-free within six weeks after the surgery. One
patient had a persistent knee flexion contracture of 15°. Thirteen
patients reported a return to normal gait and recovery of ankle dorsiflexor
function. Also, sensation in the cutaneous distribution of the superficial and
deep peroneal nerves returned in all cases.
Arthroscopy
Arthroscopy at the site of a prosthetic knee is a technically challenging
procedure, but various reports have shown promising success rates. Williams et
al. reported on arthroscopic release of the posterior cruciate ligament in ten
stiff painful knees that had undergone posterior cruciate ligament-sparing
total knee
arthroplasty28. At
a mean of twenty months after the arthroscopic procedure, the mean increase in
knee flexion was 30.5° (range, 10° to 50°). Eight patients
reported satisfaction and decreases in pain and stiffness, whereas two
patients went on to have a revision total knee arthroplasty. Diduch et al.
studied the efficacy and safety of arthroscopy for diagnosing and treating
symptoms in forty knees that had undergone total knee
arthroplasty29.
Arthroscopy was used successfully to diagnose the cause of the symptoms in
97.5% of the patients, and arthroscopic treatment included removal of
impinging tissue or loose bodies. At an average of 19.9 months, the rates of
clinical success were 82% for procedures done to treat "clunks,"
60% for those used to remove impinging synovial or soft tissue, and 63% for
those used to treat arthrofibrosis. Similarly, Jerosch and Aldawoudy evaluated
the efficacy of arthroscopic management of knee stiffness after total knee
arthroplasty30. In
their series of thirty-two knees, twenty-five demonstrated improvements in
both the range of motion and the Knee Society scores. In the present study,
four patients underwent arthroscopic removal of scar tissue. These patients
had an improvement in the range of motion ranging between 10° and 20°.
At a mean of sixteen months following the arthroscopy, no patient had required
an additional surgical procedure and the mean Knee Society score was 89
points.
Arthrotomy and Débridement
When a patient presents with severe stiffness of the knee, an arthrotomy
with synovectomy, removal of scar tissue, lateral retinacular release,
posterior capsular release, posterior cruciate ligament release, and/or
exchange of a single component can be a reasonable approach. Babis et al.
reported the results in seven knees in which stiffness after total knee
arthroplasty had been treated with an arthrotomy, débridement of scar
tissue, and isolated exchange of the tibial polyethylene
insert9. At the time
of final follow-up, two knees had been revised, four knees were severely
painful, and one knee was moderately painful. The authors concluded that open
débridement and exchange of the polyethylene insert alone was not an
effective treatment approach in their hands. Conversely, Maloney, in a comment
on the evaluation and management of
arthrofibrosis11,
pointed out the risk of damage to the articulation during arthroscopy to
remove scar tissue and found open débridement and resection of the
posterior cruciate ligament to be a good treatment option. However, he did not
report the actual number of patients in his series. Yercan et al. reported the
results in seven patients who had been treated with an arthrotomy, which had
been combined with component exchange in two
cases26. Although
five of the seven knees showed improvements both in the range of motion and in
terms of pain relief, the authors concluded that open arthrolysis and isolated
revision of a component does not correct a limited arc of motion but leads to
reliable pain relief. We performed an arthrotomy with débridement of
scar tissue and exchange of the polyethylene insert to correct oversized
components in two patients. Both patients demonstrated improvements in the
Knee Society scores and the total arc of motion.
Revision Total Knee Arthroplasty
Patients with persistent stiffness and pain after total knee arthroplasty
who do not benefit from intensive physical therapy may require a revision
total knee arthroplasty. Revision arthroplasty is the preferred treatment for
patients with malaligned or oversized components and/or late-onset stiffness.
However, studies of revision arthroplasties performed to treat stiffness after
total knee arthroplasty have yielded mixed results. Haidukewych et al.
reported that, of fifteen patients (sixteen knees) with well-fixed and
well-aligned components who had been treated with revision arthroplasty, ten
were satisfied with the outcome and showed modest improvements in the Knee
Society pain and functional scores as well as in the total arc of
motion10. In a
study by Christensen et al., eleven patients who had been treated with
revision arthroplasty because of stiffness after total knee arthroplasty had
mean improvements in the Knee Society pain and functional scores of 39.6 and
53.7 points, respectively, and a mean improvement in the arc of motion of
43.5°31. Three
patients required manipulation under anesthesia because of recurrent
stiffness, and one required revision surgery because of loosening of the
femoral component. Nicholls and Dorr reported that, of thirteen revision
arthroplasties (in twelve patients) performed to treat stiffness, five
achieved an excellent or good result, four achieved a fair result, and four
achieved a poor
result32. Four
knees required manipulation because of recurrent stiffness postoperatively.
Despite these suboptimal clinical outcomes, eleven of the twelve patients
reported a high degree of satisfaction with the result of the procedure
because of pain relief. Ries and Badalamente reported the results of six
revision total knee arthroplasties in five patients who had had stiffness
after the primary
arthroplasty33. The
arc of motion increased from a mean of 36° before the revision surgery to
a mean of 86° after it. In a study of fifty-six knees in fifty-two
patients who had undergone revision surgery because of stiffness, Kim et al.
reported that the mean Knee Society pain and functional scores improved from
15 to 47 points and from 39 to 87 points,
respectively2.
Thirty-five knees (63%) had a decreased flexion contracture, and forty-five
knees (80%) had increased flexion. Fifty-two knees (93%) demonstrated a gain
in the total arc of motion. Only one patient underwent revision arthroplasty
in our series. This patient had an improved but still limited range of motion
and reported a high degree of satisfaction because of persistent pain
relief.