Patients who undergo total knee arthroplasty often experience
intense postoperative pain, particularly during efforts to mobilize
and strengthen the affected extremity1.
A reduction in postoperative pain after total knee arthroplasty
is associated with an increase in the range of motion of the knee,
faster mobilization, and a shorter hospital stay2,3.
The intensity of pain following total knee arthroplasty is predictive
of extended stays in rehabilitation hospitals4 and
aberrant gait patterns5,6. Pain
intensity and residual functional limitations are closely related
to patients’ perceptions of success7,8.
Although uncontrolled pain is an acknowledged impediment to postoperative
functional recovery9, strategies
to ensure patient comfort during rehabilitation have yet to be extensively
integrated into clinical practice. Typically, immediate-release
opioids are prescribed on an as-needed basis to control pain following
total knee arthroplasty. Often, a combined opioid-acetaminophen
or opioid-aspirin formulation is administered every four to six
hours on patient request. There have been well-documented problems
with this approach consequent to both patient and caretaker-related
barriers (for example, misinformation regarding addiction risk, patients’ reluctance
to trouble their caretakers, and delays in delivery of the analgesic
after it has been requested)10.
Given the expanding numbers of arthroplasties being performed, the
optimal postoperative management becomes increasingly important
as a medicoeconomic and public-health concern. If the adequacy of
pain control affects recovery of joint strength and range of motion,
restoration of functional autonomy, and/or postoperative
utilization of resources, the clinical and economic consequences of
pain control could be substantial.
Controlled-release opioid preparations provide a reliable means
of maintaining stable serum concentrations and avoiding the erratic
fluctuations that may characterize immediate-release formulations; they
also free patients from the onus of requesting as-needed pain medication.
We conducted a randomized, double-blind, placebo-controlled trial
to assess whether controlled-release opioids provide superior control
of postoperative pain, result in better functional recovery, and
reduce the duration of rehabilitation following unilateral total
knee arthroplasty in comparison with on-request, immediate-release
opioids.
The study was conducted at two affiliated freestanding acute-rehabilitation
facilities and was approved by a central institutional review board.
All subjects screened for study participation had been transferred
to a rehabilitation hospital within seven days following elective
unilateral total knee arthroplasty performed for the treatment of
osteoarthritis or rheumatoid arthritis. Patients were recruited
between February 1, 1997, and September 30, 1997. Eligible subjects
had to speak English, have rated their pain as moderate to very
severe on a 5-point Likert-type scale (1 = none, 2 = mild,
3 = moderate, 4 = severe, and 5 = very
severe), have been cleared to bear weight fully on the involved extremity
at the time of admission to the rehabilitation hospital, have no
history of substance abuse as assessed through administration of
the Drug Abuse Screening Test11,
and have no evidence of cognitive impairment (a score of >27
as determined with the Mini-Mental State examination described by
Folstein et al.12). There were
no exclusion criteria based on age, functional status before the
total knee arthroplasty, or pain severity or duration before the
arthroplasty.
A total of 135 patients were screened, and fifty-nine (44%)
were enrolled in the study. Reasons for nonenrollment included transfer
back to an acute-care institution due to medical instability (2%)
and the patient’s refusal to participate (54%).
Enrolled and nonenrolled subjects were similar with respect to race
(p = 0.99), sex (p = 0.51), and age (p = 0.44). The
median pain rating of the patients who were enrolled in the study
was 0.6 point higher than that of the patients who were not (p = 0.05).
Study subjects were approached on the day of admission to the
rehabilitation hospital and were screened for eligibility. After
subjects had provided written informed consent they were randomized
at a central pharmacy in blocks of ten. Patients were randomized
separately for the two participating facilities, since the site
of the rehabilitation-service delivery was believed to be a potential
confounder.
In the intervention group, patients received opaque white capsules
containing 10 mg of OxyContin (oxycodone) at 8:00 PM and opaque
blue capsules containing 20 mg of OxyContin at 8:00 AM. Patients
assigned to the control group received identical capsules containing
lactose on the same dosing schedule. Both groups had standing orders for
immediate-release oxycodone, 5 mg every four hours, as needed. The
study therapy was initiated on the evening following admission to
the rehabilitation hospital.
The starting dose of OxyContin (20 mg in the morning and 10 mg
in the evening) was arbitrary and deemed unlikely to be optimal
for all patients in the OxyContin group. Therefore, a blinded upward
titration of the OxyContin regimen based on the number of times
that the patient received on-request, immediate-release oxycodone
was adopted. Patients who received three or more on-request 5-mg
doses of immediate-release oxycodone on two consecutive days had
the OxyContin dose increased by 10 mg. They then received 20 mg of
OxyContin in the morning and 20 mg in the evening contained in opaque
blue capsules. The upward titration was continued to a possible
maximum OxyContin dose of 30 mg in the morning and 30 mg in the
evening. Patients in the placebo group underwent a similar titration
of study medication contingent on their utilization of rescue doses;
however, they continued to receive opaque capsules containing only
lactose.
Patients in both groups also had standing orders for a bowel
regimen (docusate sodium, 100 mg, and senna three times a day; lactulose,
20 g three times a day on request; and a Fleet enema once a day
on request), acetaminophen (325 to 650 mg every six hours on request),
and an antiemetic medication (Torecan [thiethylperazine],
10 mg every eight hours on request). All patients used a continuous-passive-motion
machine for two and one-half hours each evening at a starting range
of 80° to 100° of knee flexion, which was increased as tolerated.
Both groups participated in a standard, rigorous rehabilitation
program for three hours each day. The program consisted of range-of-motion
activities, progressive resistive exercises, and instruction in
transfers, walking, and negotiation of stairs and uneven surfaces.
All subjects had physical therapy orders for the use of topical
thermal modalities (ice packs and hydrocollators) and/or
electrical stimulation to be used on an on-request basis for the
alleviation of pain associated with the total knee replacement.
Data were collected at various time-points throughout each subject’s
stay in the rehabilitation hospital. At baseline, information was
collected regarding sociodemographic characteristics, the visual-analog pain
scores before and after the total knee arthroplasty, the degree
of arthritis in other joints, and the duration of pain prior to
the total knee arthroplasty.
During the follow-up period, visual-analog pain scores were recorded
immediately following each full weekday physical therapy session.
Subjects were requested to rate their pain "right now" and "at
worst during physical therapy." They also were asked to
rate the degree to which the pain interfered with their ability
to participate in physical therapy. The validity of this type of
interference score has been demonstrated through use of the Brief
Pain Inventory13,14.
Initial and final panels of physical performance variables were
collected at the first and eighth weekday physical therapy sessions
by the treating physical therapist. The active and passive ranges
of knee motion, quadriceps strength, distance that the patient could
walk safely in a three-minute interval, and selected Functional
Independence Measure scores15 were
determined by therapists to assess the patients’ functional
status, establish appropriate therapeutic goals, and gauge the rate
of recovery. The passive and active ranges of motion were determined,
with use of a standard goniometer, with the subjects in a sitting-supported
position to normalize the degree of hip flexion. Three values for
both the passive and the active range of motion were recorded, and
the highest value was used in the data analysis. Knee extensor,
or quadriceps, strength was measured in pounds with use of a Chatillon CSD400C
handheld dynamometer (Chatillon, Greensboro, North Carolina)16,17.
Functional Independence Measure scores for walking, sit-to-stand
transfers, and stair-climbing were included in the panel of variables
collected during the first and eighth weekday physical therapy sessions.
The treating physical therapists assigned Functional Independence
Measure scores to the patients on the basis of their observed performance during
therapy. The subjects’ speed of walking was determined
by recording the distance safely traversed during a three-minute
period. A carefully measured rectangular course circumscribing the physical
therapy gym was used for this purpose.
The Memorial Symptom Assessment Scale18 was
administered to the subjects following the sixth physical therapy
session. A subscale of the Memorial Symptom Assessment Scale was
found through factor analysis to be sensitive for detecting the
presence and severity of opioid-related side effects. Although the
full Memorial Symptom Assessment Scale instrument was administered, values
for this subscale were used during data analysis to determine whether
the OxyContin and control groups differed in the degree to which
they experienced and were distressed by opioid-induced side effects.
Length of stay was recorded at the time of discharge from the
rehabilitation hospital, as was the plan for any additional physical
therapy. Possible disposition plans consisted of transfer to a subacute-rehabilitation
facility, enrollment in home or outpatient physical therapy, or
no additional physical therapy.
Selected patient characteristics, preoperative and postoperative
visual-analog pain scores, and functional measures on the first
day of physical therapy were compared between the OxyContin and
placebo groups. Outcome measures that were compared included visual-analog
pain scores (at the end of physical therapy, the worst during physical
therapy, and the degree to which pain interfered with physical therapy)
and change in functional measures; the latter was calculated by
subtracting scores recorded at the first physical therapy session from
those recorded at the eighth. Length of stay in the hospital and
discharge plans were also compared. P values were calculated with
use of a two-tailed t test for continuous measures and a chi-square
test for dichotomous measures; a nominal value of p < 0.05
was used to establish significance.
The average age of the fifty-nine study subjects was sixty-five
years (range, forty-six to eighty-five years); twenty-nine (49%)
were male. The OxyContin and placebo groups were similar with respect
to demographic and clinical characteristics. Visual-analog pain
scores were consistently high in both groups during both the preoperative
and the postoperative period (average, 7.8 for both periods), and
neither the pain scores nor the functional measures at the first
physical therapy session differed between the two groups. Of the
fifty-nine patients enrolled in the study, twenty-nine were randomized to
receive OxyContin and thirty, to receive a placebo.
Seven patients (three in the OxyContin group and four in the
placebo group) discontinued taking the study medication; they continued
to be followed, however, for the outcomes of interest and were included
in all analyses. The three patients in the OxyContin group expressed
a desire to have greater control over their immediate-release medication, whereas
the four patients in the placebo group gave inadequate analgesia
as the reason for discontinuing the study medication. Outcome data
were unavailable for one patient in the placebo group; that patient
was lost to follow-up because of emergency admission to an acute-care
hospital.
The patients in the OxyContin group requested an average of 1.9
doses of rescue medication per day—that is, immediate-release
oxycodone (5 mg per dose) to control pain that was inadequately managed
by the study medication—whereas those in the placebo group
requested an average of 2.6 doses per day (p = 0.02). As
a consequence, the dose was titrated to the highest dose of study
medication for only 7% (two) of the twenty-nine patients in
the OxyContin group compared with 43% (thirteen) of the
thirty in the placebo group. Total daily consumption of oxycodone
(controlled-release plus immediate-release) by the patients randomized
to OxyContin therapy was more than four times higher than that by
the patients in the placebo group (54.4 and 12.9 mg, respectively;
p < 0.001). Comparison of the scores on the Memorial Symptom Assessment
Scale on the sixth day of the study, however, revealed no difference
between the two groups with regard to opioid-related side effects; the
scores averaged 3.8 and 3.9 in the OxyContin and placebo groups,
respectively (p = 0.830).
The pain scores at the eighth physical therapy session were uniformly
lower for the patients in the OxyContin group (Table I). With the
numbers available, no significant differences were detected between
the OxyContin and placebo groups with regard to physiologic or functional
measures at the first physical therapy session. The patients treated
with OxyContin had greater improvement than the placebo group in
all functional measures; the difference between groups approached
or achieved significance for all seven measures.
The patients in the OxyContin group were discharged from the
rehabilitation hospital at an average of 2.3 days earlier than those
in the placebo group (13.0 compared with 15.3 days; p = 0.013). The
duration of the acute hospitalization was not factored into the
determination of the length of stay in the rehabilitation hospital.
The patients who received the placebo were nominally more likely
to be discharged to home physical therapy or transferred to a subacute-rehabilitation
facility.
Our findings indicate that patients who receive preemptive treatment
experience less pain, recover knee strength at an accelerated rate,
and utilize fewer health-care resources. The patients in the OxyContin
group received controlled-release opioid in addition to rescue doses
of immediate-release opioid; they also received a higher total average daily
dose of opioid than did the patients in the placebo group. The extent
to which these differences contributed to the study results is less
important than the fact that the preemptive approach to pain management
involved rapid titration to meet each patient’s opioid
requirement. This analgesic strategy differs markedly from the current
practice of using as-needed opioid or nonsteroidal anti-inflammatory
drugs for pain management, which can result in unrelieved pain and
delayed functional recovery8.
The intervention strategy employed in this study—using
rescue doses to relieve pain that was inadequately relieved by a
controlled-release opioid preparation and to guide its upward titration—is supported
by extensive clinical experience in the management of cancer-related
and postoperative pain. Nonsteroidal anti-inflammatory drugs and adjuvant
analgesics may be used as valuable supplements to opioid analgesia.
In our experience, however, the use of nonsteroidal anti-inflammatory drugs
or adjuvants in isolation rarely controls pain adequately during
aggressive mobilization after total knee arthroplasty.
One of the most important benefits afforded by controlled-release
opioid preparations is the maintenance of relatively constant serum
opioid levels. The constant analgesia provided by stable serum oxycodone
levels may have facilitated the intervention group’s recovery
in a variety of ways. Formal therapy sessions lasted only a total
of three hours per day; enhanced ongoing pain control may have allowed
the patients in the intervention group to remain active outside
of formal physical therapy sessions. Enhanced baseline analgesia
also may have diminished central sensitization, the permissive effect
of nociceptive impulses on the central nervous system’s
response to additional noxious stimuli19.
Better constant analgesia may have reduced the development of aberrant
biomechanical patterns20 in the
intervention group and may have facilitated the recovery of normal
movement. Finally, a growing body of literature supports the role
of peripheral opioid receptors in mediating local inflammatory responses21,22. The exuberant soft-tissue inflammation
that typically develops following total knee arthroplasty may have
been attenuated by stable serum opioid levels. The degree to which
these potential benefits were conferred by the use of controlled-release oxycodone
must remain speculative.
Our sample was derived from a population referred for acute inpatient
rehabilitation after total knee arthroplasty; this distinguishes
this series from the growing number of patients who are discharged directly
to home following arthroplasty. Many of our patients reported severe
pain at the arthroplasty site and likely fall within a subset of
patients experiencing greater pain after total knee arthroplasty. Patients
who consented to participate in the study had higher pain scores
than those who did not consent, suggesting that our study attracted
patients for whom pain was a substantial problem. It is unclear if
our findings are generalizable to a population of patients less
troubled by pain.
Despite these limitations, we believe that our study has important
clinical and economic implications. As financial pressures build
to limit the use of rehabilitative services following total knee
arthroplasty, timely recovery of function with effective control of
symptoms becomes increasingly important. Our results indicate that
preemptive pain control with a controlled-release opioid appropriately
titrated to each patient’s needs during rehabilitation
accelerates functional recovery and reduces the utilization of health-care
resources.