Background: Chronic use of opioid medications may lead to dependence or hyperalgesia, both of which might adversely affect perioperative and postoperative pain management, rehabilitation, and clinical outcomes after total knee arthroplasty. The purpose of this study was to evaluate patients who underwent total knee arthroplasty following six or more weeks of chronic opioid use for pain control and to compare them with a matched group who did not use opioids preoperatively.
Methods: Forty-nine knees in patients who had a mean age of fifty-six years (range, thirty-seven to seventy-eight years) and who had regularly used opioid medications for pain control prior to total knee arthroplasty were compared with a group of patients who had not used them. Length of hospitalization, aseptic complications requiring reoperation, requirement for specialized pain management, and clinical outcomes were assessed for both groups.
Results: Knee Society scores were significantly lower in the patients who regularly used opioid medications at the time of final follow-up (mean, three years; range, two to seven years); the opioid group had a mean of 79 points (range, 45 to 100 points) as compared with a mean of 92 points (range, 59 to 100 points) in the non-opioid group. A significantly higher prevalence of complications was seen in the opioid group, with five arthroscopic evaluations and eight revisions for persistent stiffness and/or pain, compared with none in the matched group. Ten patients in the opioid group were referred for outpatient pain management, compared with one patient in the non-opioid group.
Conclusions: Patients who chronically use opioid medications prior to total knee arthroplasty may be at a substantially greater risk for complications and painful prolonged recoveries. Alternative non-opioid pain medications and/or earlier referral to an orthopaedic surgeon prior to habitual opioid use should be considered for patients with painful degenerative disease of the knee.
Level of Evidence: Therapeutic Level III. See Instructions for Authors for a complete description of levels of evidence.
Osteoarthritis of the knee has a chronic course, with gradually increasing pain and decreasing quality of life prior to reaching the end stage of disease that may be amenable to arthroplasty treatment1. A number of guidelines and consensus statements have been released regarding the management of this disease2-4. While the American Academy of Orthopaedic Surgeons clinical practice guidelines on the treatment of osteoarthritis of the knee do not address the use of opioid medications5, most reports recommend that mild analgesics such as acetaminophen be used initially, followed by nonsteroidal anti-inflammatory drugs and opioids. Examples of prescribed opioids include: Tylenol #3 (codeine and acetaminophen), Percocet (oxycodone and acetaminophen), Percodan (oxycodone and acetylsalicylic acid), OxyContin (oxycodone timed-release formulation), Dilaudid (hydromorphone), Vicodin (hydrocodone and acetaminophen), and methadone. Surgical treatment, including total knee arthroplasty, is typically considered after nonoperative methods have failed.
The use of both natural and synthetic derivatives of opium has become well established for the palliative management of intense pain6. More recently, an increase in the use of these medications for the treatment of chronic musculoskeletal conditions has been reported7. Lee et al. reported that opioid medications have become the second or even the first-line treatment of choice for primary-care providers for the management of osteoarthritis-associated pain8. In 2009, the American Geriatrics Society released updated guidelines for the management of chronic pain in elderly patients9; these guidelines recommended against the use of nonsteroidal anti-inflammatory medications due to the high risk of gastric bleeding and suggested the use of opioid analgesics instead. However, several authors have raised concerns about the use of these medications, specifically with regard to the development of tolerance and hyperalgesia10 and the potential association of opioids with worse treatment outcomes in association with a variety of musculoskeletal conditions affecting the neck, lower back, and knee11-14.
The purpose of this study was to compare the perioperative course, complication rates, and clinical outcomes of patients who underwent total knee arthroplasty and were or were not treated with chronic preoperative oral opioid analgesic medications.
Materials and Methods
Databases of patients who underwent operation at either of two specialized centers were used to identify all primary total knee arthroplasties that were performed for end-stage degenerative joint disease. Multiple sources of data were reviewed for each of these patients, including inpatient and outpatient charts and paper and computer-based office notes as well as repeated patient and family interviews, to identify subjects who had documented treatment with oral opioid medications for knee pain for a minimum of six weeks prior to the arthroplasty procedure. This threshold for chronic use was based on the report by Chu et al. on the development of hyperalgesia in human volunteers following one month of daily oral opioid medication use10. All opioid medications and dosages were converted to a morphine-equivalent dose with use of ratios published by Labby et al.15, with only those patients using a minimum equivalent dose of 20 mg/day included in the preoperative opioid group (equivalent to approximately four hydrocodone or three oxycodone tablets per day). The medication dosages were determined from several sources, including prescription records, clinic notes, and admission records. Forty-nine knees were identified in fourteen men and thirty-one women (including forty-one unilateral procedures and four bilateral arthroplasties) who had a minimum of twenty-four months follow-up. An additional three patients were identified who met the study criteria for chronic preoperative opioid use, but these patients were lost to follow-up following the initial postoperative visit and were not included in the study cohort. No perioperative complications were noted for these three patients.
The patients in the opioid group had a mean age of fifty-six years (range, thirty-seven to seventy-eight years) and a mean body mass index (BMI) of 34 kg/m2 (range, 19 to 58 kg/m2) at the time of the index arthroplasty. These patients used a mean morphine-equivalent dose of 58 mg/day (range, 20 to 300 mg) prior to surgery. The majority of patients were close to the minimum 20 mg/day morphine-equivalent dose threshold, with thirty-four of the forty-five patients reporting a daily dose of from 20 to 60 mg. The reported daily dose was in the range of 61 to 100 mg for six patients, 101 to 140 mg for four patients, and 300 mg for one patient.
This opioid cohort group was matched to a group of patients who had primary total knee arthroplasty at the same centers over a similar time period, but who were not treated with chronic opioids prior to surgery. Institutional review board approval was granted for the review and study of these patients at both centers.
The medical records of the patients in the opioid group as well as the patients in the matched non-opioid group, including inpatient and outpatient charts, were reviewed. The demographic characteristics, type of medical insurance, preoperative and perioperative histories, and clinical outcomes of the study patients were reviewed and compared with those of the matched patients.
Specific demographic characteristics reviewed included patient age, sex, BMI, and type of insurance. Preoperative and perioperative factors examined included the specific opioid medication(s) used and duration of preoperative use; preoperative diagnosis leading to degenerative joint disease; history of substantial knee trauma; number and type of previous knee surgical procedures; history of chronic back pain and/or back surgery; smoking status; neuroleptic, antidepressant, and/or anxiolytic medication use; alcohol abuse; systemic corticosteroid use; preoperative University of California Los Angeles (UCLA) activity scores; Knee Society objective scores; knee range of motion; and number of days until discharge following surgery. Clinical outcomes evaluated were the Knee Society objective score, knee range of motion, and the prevalence of arthroscopic and/or revision surgery for unexplained knee stiffness and/or pain. Serial radiographs of all patients who had additional surgical procedures were assessed for evidence of progressive radiolucency adjacent to the components or for component migration. Patients were also evaluated for the need for referral to a pain management specialist. This is routinely done at both study centers if patients continue to experience substantial pain ninety days following the most recent surgical procedure and the pain is unexplained by implant factors and requires narcotic medication.
Opioid-use patients were individually matched to patients on the basis of center (exact match), procedure type (unilateral or bilateral procedure; exact match), sex (exact match), age (± four years), and BMI (± 4 kg/m2). Matching was performed by sequential comparison of study patients to a prospectively collected database of patients treated at the same centers and made use of a randomized order of patients as well as filtering to include only those patients treated with a primary total knee arthroplasty. Once a match was identified, the patient's demographic and clinical data were reviewed to ensure that the matched patient did not fit our criteria for preoperative opioid use, and that a minimum of twenty-four months of follow-up data were available. The matched patient group included fourteen men and thirty-one women who had a mean age of fifty-seven years (range, thirty-seven to seventy-six years) and a mean BMI of 33 kg/m2 (range, 22 to 54 kg/m2). Similar to the study group, four patients had undergone bilateral arthroplasties under the same anesthetic. The mean preoperative Knee Society scores, knee motion, follow-up times, and preoperative diagnoses were similar for both groups (Table I).
The type of medical insurance used by the patients was stratified into one of four categories: (1) private insurance or self-pay, (2) Medicare, (3) Medicaid, and (4) Workers’ Compensation. The distribution of the different types of insurance was compared between the two groups and was found to be similar (Table I).
Once the initial matched groups were identified, the prevalence of various potential confounding risk factors, including previous knee arthroscopy, chronic back pain and/or surgery, smoking, and the use of psychiatric medications, were compared between the two groups and analyzed for significant differences in these factors between the two populations (Table I). The only factor found to be significantly different between the two groups was a higher rate of use of antidepressant and/or anxiolytic medications in the opioid group. However, when complication rates were compared between patients who did and did not take these medications, both groups had a similar rate of additional surgical procedures for unexplained knee stiffness and/or pain (12% versus 13%; p = 0.819).
Perioperative analgesia was provided through patient-controlled intravenous analgesia, continuous infusion of local anesthetic into an epidural catheter, or intermittent intravenous analgesia. The prevalence of these modalities in each group is described in Table II. Postoperatively, patients were transitioned to oral analgesia in preparation for discharge. The mean morphine-equivalent dose of oral opioid medications prescribed at the time of discharge was calculated and is described in Table II. The prevalence of different types of postoperative analgesia and the mean morphine-equivalent dosage of oral narcotics were similar between the two groups.
A matched t test or Fisher exact test was used where appropriate to compare the lengths of hospitalization, Knee Society scores, and range of motion between the two groups, while McNemar's chi square analysis16 was used to compare the prevalence of potential preoperative risk factors, additional surgical procedures, and referral for specialized pain management. A t test was performed to compare the mean preoperative range of motion in the study group between those knees that did and did not undergo postoperative manipulation under anesthesia. Pearson product moment correlation was calculated to test for correlation in the study group between the postoperative length of stay and the preoperative narcotic dosage as well as the preoperative Knee Society scores.
Sources of Funding
No external funding was received in support of this work.
Patients with a history of preoperative opioid medication use had a significantly longer duration of postoperative hospitalization compared with those who did not use preoperative opioids (p = 0.013). The opioid group was discharged from the hospital at a mean of 4.3 days (range, two to eight days) following the index arthroplasty, compared with 3.4 days (range, two to six days) for the non-opioid group (Table III). No significant correlation was identified in the opioid group between the preoperative narcotic dosage and postoperative length of stay (r = 0.163, p = 0.286).
A significantly higher number of patients in the opioid group required additional surgical procedures over the course of the study period for unexplained knee stiffness or pain (eleven versus none; p < 0.001; 95% confidence interval [CI] was 5 to 17 for the opioid group). There were five arthroscopic evaluations and debridements for unexplained knee pain and/or stiffness resistant to nonoperative treatment in the opioid group. Eight arthroplasties in the opioid group were revised for unexplained knee stiffness and/or pain despite being aseptic and with all components well fixed, including an arthroplasty in one patient who had previously undergone an arthroscopic evaluation but remained symptomatic and an arthroplasty in a patient who originally underwent bilateral total knee arthroplasties and had both knees revised. All patients who underwent revision for persistent stiffness had previously undergone manipulations under anesthesia (indicated in patients who had less than 95° of flexion and/or loss of more than 5° of extension at six to seven weeks following the index arthroplasty) without lasting benefit. A total of seventeen manipulations were performed in the opioid group (95% CI, 10 to 24), and nine in the non-opioid group (95% CI, 3 to 15). On the basis of the numbers, no significant difference in preoperative range of motion was identified between the patients in the study group who did and did not undergo manipulation under anesthesia (mean ranges of motion of 100° and 101°, respectively, p = 0.495). No progress in radiographic lucencies adjacent to the implants or evidence of component migration was identified in any of these patients prior to component revision. In contrast, no patients in the non-opioid group underwent these procedures for aseptic reasons.
A significantly greater number of patients in the opioid group were referred for specialized management of intractable pain of unknown cause compared with the non-opioid group, with ten referrals in the former group and one in the latter (p < 0.001). One (17%) of the six chronic narcotic patients in the Medicaid/Workers’ Compensation group, who was insured through Workers’ Compensation, was referred for pain management, compared with nine (23%) of the thirty-nine patients who had one of the remaining types of insurance. In the non-opioid group, none of the five Medicaid/Workers’ Compensation patients required referral, while one (2%) of the remaining forty-five patients was referred. While the prevalence of referral does not appear to vary on the basis of the type of insurance, the low number of patients in the Medicaid and Workers’ Compensation subgroups precludes a meaningful statistical comparison.
Clinical outcomes as measured with use of the Knee Society objective knee score were significantly worse in the opioid group. At the time of final follow-up, the mean Knee Society objective score was 79 points (range, 45 to 100 points) in the opioid group, compared with a mean score of 92 points (range, 59 to 100 points) in the non-opioid group (p < 0.001). A moderately significant correlation was identified in the study group between higher preoperative Knee Society scores and a shorter postoperative length of stay (r = −0.537; p < 0.001). The mean knee range of motion at the time of final follow-up was similar between the two groups, with 107° (range, 30° to 130°) in the opioid group and 111° (range, 75° to 130°) in the non-opioid group (p = 0.223).
Recently, there has been increasing interest in the use of opioid medications for the management of chronic pain from nonmalignant disease, and this treatment approach has been advocated by a number of specialty societies for use in place of nonsteroidal anti-inflammatory medications4,9,17. This trend has been substantiated by a recent report that oral opioid analgesics are being increasingly prescribed by primary-care physicians8 despite reported associations between the use of opioid analgesia and poorer outcomes following the treatment of musculoskeletal conditions11-14. To the best of our knowledge, there have been only two reports evaluating the association between preoperative opioid use and outcome after total knee arthroplasty. Karbassi et al. reviewed 6364 primary total knee arthroplasties that had been performed over a five-year period18 and found that preoperative opioid users had lower mean Knee Society scores at a follow-up time of one year (74 points versus 80 points, range not reported), and were significantly more likely to be dissatisfied with the results of the procedure (odds ratio, 3.48; p < 0.001). In contrast, Fisher et al. identified seventy-four patients from a total of 1024 patients who had a total knee arthroplasty performed at a single center and who had knee pain or stiffness one year following the index arthroplasty, and those authors found that preoperative opioid medication use was not associated with an outcome of stiffness or pain (p = 0.110)13.
Other authors have investigated the association between opioid medications and the outcomes associated with treatment of other musculoskeletal conditions. Lawrence et al. reported a significantly lower prevalence of good-to-excellent clinical outcomes (defined as normal activity level and work status, mild or no pain, no more than occasional use of nonsteroidal anti-inflammatory medications) following cervical arthrodesis in patients who used opioid analgesia for a minimum of six months prior to surgery (51% versus 100%; p < 0.01)14. Kidner et al. reported that, in patients admitted for rehabilitation of a chronic musculoskeletal disorder, increased opioid medication dosages were associated with an increased rate of program noncompletion, a lower rate of return to work, and an increased likelihood of receiving disability or social security supplemental income one year following treatment11. Both of these reports are consistent with the present study, which found that the chronic use of preoperative opioid medications is associated with poorer clinical outcomes and a significantly higher prevalence of complications requiring a return to the operating room.
There are several potential reasons why chronic preoperative opioid use may result in a longer time to discharge, including poorer progression toward rehabilitation goals, a higher prevalence of in-hospital complications such as ileus, and difficulty with postoperative pain control as a result of hyperalgesia. All three of these potential reasons for increased length of stay were observed in the present study but were not specifically controlled for, as they may be associated with opioid medication use. In contrast, the type of surgery (unilateral versus bilateral arthroplasty) was controlled for in the matching strategy to ensure that this did not confound the data regarding length of stay. No patients in the present study had an increased length of stay because of delays in obtaining transportation or placement in a rehabilitation facility. While the results of the present study demonstrate an association between chronic preoperative opioid use and a longer time to discharge, further work is necessary to elucidate the specific reasons for this finding.
The authors acknowledge a number of limitations of this study. Because of the lack of prospective randomization, it is possible that differences in factors other than the use of opioid medications may have been responsible for some of the differences in outcomes seen between the two groups. Furthermore, because patients were self-administering the opioid medications at home, and because we were not able to review patients’ outpatient pharmacy records, it is difficult to exactly track the dosages of opioids used by the patients. It is possible that patients were procuring medications from multiple independent sources, diverting some medications to other individuals, or successfully concealing opioid use, resulting in an erroneous estimate of actual medication utilization. However, opioid use was confirmed by at least two separate data sources, and it is unlikely that patients were not taking at least the minimum doses reported. While the present report suggests that chronic preoperative narcotic use is associated with poorer postoperative outcomes, the study design did not make it possible to accurately identify the influence of potential risk factors such as BMI, unilateral versus bilateral surgery, preoperative activity levels, type of anesthesia used (regional or general), poor patient tolerance of non-opioid analgesics such as nonsteroidal anti-inflammatory drugs, and differences in social situation and/or occupation beyond the type of insurance used. A post hoc analysis of various potentially confounding factors found that the only significant difference was in the number of patients who were taking antidepressant and/or anxiolytic medications prior to surgery, although the prevalence of additional surgical procedures was similar between patients who did and did not use these medications. The mean age of the patients in the study group in the present study was fifty-six years, which is younger than the mean age for all patients undergoing primary total knee arthroplasty in the United States19. Several patients had a history of knee trauma or osteonecrosis, which have been associated with early progression to end-stage degenerative disease. However, the proportion of patients in both groups with secondary osteoarthritis was similar, suggesting that the poorer outcomes in patients who used chronic oral narcotics preoperatively are not specifically associated with the preoperative diagnosis.
In conclusion, the use of chronic oral opioid analgesia prior to total knee arthroplasty was associated with an increased rate of complications requiring return to the operating room and poorer clinical outcomes. It may be beneficial to investigate and define limits of preoperative opioid dosage and duration to improve the postoperative outcomes of patients with degenerative disease of the knee who cannot or prefer not to be immediately treated surgically. Although conjectural, surgeons might consider recommending an opioid withdrawal program with the assistance of pain management specialists prior to surgery in an attempt to improve postoperative outcomes. Also, patients and primary-care physicians should be encouraged to seek orthopaedic consultation prior to beginning opioid therapy, and it may be beneficial to limit preoperative analgesia to non-opioid and/or anti-inflammatory medications, giving consideration to more expeditious knee arthroplasty once these medications are no longer effective in controlling symptoms. However, larger and more rigorous studies should be conducted to better elucidate the relationship between opioid use and the outcomes of total knee arthroplasty.
Note: The authors acknowledge Qais Naziri, Aaron Johnson, Christopher Costa, and Patti Litke for their assistance in collecting and assembling the data for this manuscript.
Investigation performed at the Rubin Institute for Advanced Orthopedics, Baltimore, Maryland, and the Bonutti Clinic, Effingham, Illinois
A commentary by Thomas Parker Vail, MD, is linked to the online version of this article at jbjs.org.
Disclosure: None of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of any aspect of this work. One or more of the authors, or his or her institution, has had a financial relationship, in the thirty-six months prior to submission of this work, with an entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. No author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.
- Copyright © 2011 by The Journal of Bone and Joint Surgery, Incorporated