Secondary Logo

Journal Logo

SECTION I SYMPOSIUM: Papers Presented at the Knee Society 2003

Ranawat Award Paper: Predicting Total Knee Replacement Pain

A Prospective, Observational Study

Brander, Victoria A. MD*†§; Stulberg, S. David MD***†‡; Adams, Angela D. BA***; Harden, R. Norman MD§; Bruehl, Stephen PhD; Stanos, Steven P. DO§; Houle, Timothy PhD§

Author Information
Clinical Orthopaedics and Related Research: November 2003 - Volume 416 - Issue - p 27-36
doi: 10.1097/01.blo.0000092983.12414.e9
  • Free


Total knee arthroplasty has proved successful for the treatment of joints compromised by OA or other disorders. In the United States, more than 250,000 TKAs are done each year, with success rates of 80% to more than 90% reported in various series 1,17,23 and the number of procedures is projected to reach 314,000 by the end of this decade. 1 In an aging population, more people are likely to experience knee dysfunction, fueling demand for the procedure, and increasing total costs for surgery and rehabilitation.

Causes of poor outcome include infection, poor prosthetic design and positioning, patellofemoral disorders, neurovascular complications, periprosthetic fractures, ligament imbalances, component loosening, and chronic pain. Although pain relief is the most important postoperative outcome, the natural history of postoperative pain has not been studied. Some patients persistently complain of unacceptable pain despite normal radiographs, unremarkable physical examinations, and lack of identifiable implant or soft tissue dysfunction. These patients have self-perceived poor outcomes and some experience prolonged and difficult recoveries. These patients may report that their pain is worse after surgery; many are judged by their surgeons to have acceptable results despite the patients’ complaints of pain. 7 Anecdotally, many surgeons think that psychologic factors play a strong role in these patients’ complaints of pain.

Traditionally, the results of TKA have been evaluated on the basis of the physician’s assessment of the pain, standard radiographic analysis of alignment and periprosthetic bone quality, and the functional ability of the patient. 16 Patients usually are informed by physicians that pain may persist approximately 3 months after surgery. Although such responses should be based on evidence-based conclusions, studies describing normal pain patterns after surgery or identifying patients at risk for unusual pain are lacking. To answer these questions about postoperative pain, we investigated normal pain recovery patterns and identified factors predicting high pain reports after TKA.


Patients and Procedures

Between 1998 and 2000, all consecutive, primary TKAs for OA done by one surgeon at an academic medical center were evaluated for enrollment. Patients had to meet the following criteria for inclusion in the study: diagnosis of degenerative arthritis, intact cognitive abilities, age older than 18 years, no diagnosis of depression or concurrent treatment with an antidepressant or anxiolytic, no concurrent musculoskeletal diagnosis that would impact the interpretation of pain (fibromyalgia, spinal stenosis, significant ipsilateral hip OA), and submission of a signed consent form. The investigation was approved and supervised by the University’s Office for the Protection of Research Subjects.

All patients followed a standardized, postoperative rehabilitation clinical pathway. On the first postoperative day, patients were allowed to bear full weight on the operative leg, and the operative leg was placed in a continuous passive motion machine that was used continuously when the patients were in bed. Patients received physical therapy twice each day. In the absence of complications, by the fourth or fifth postoperative day, patients were discharged to home or to a specific inpatient rehabilitation unit, depending on their level of independence and home support. All patients received home or outpatient physical therapy.

Study Design and Data Sources

Using a prospective, observational study design, we assessed the clinical and radiographic variables preoperatively and 1, 3, 6, and 12 months postoperatively. Data collected at each visit were results of the visual analog scale (VAS) 8 and McGill Pain Questionnaire-SF—Sensory, Affective, and Total scores, 12 Hospital for Special Surgery knee-rating scale, 10 Western Ontario and McMaster University Osteoarthritis Index (WOMAC), 2 Knee Society knee and function scores, 9 measures of depression and anxiety (Beck Depression Inventory, 18 State-Trait Anxiety Index, 22 Perceived Stress Scale), 15 and a standard clinical examination form (Joint Reconstruction and Implant Service Database [JRISDAT](Appendix 1)). All questionnaires were self-administered tests, with the exception of the JRISDAT, which was completed by the examining physician at the time of the patients’ followup.

Preoperative radiographs were analyzed and graded (0 to 4) by two readers, blinded to other subject data, using the Kellgren and Lawrence scale. 13,14,20 Radiographic evaluation of component alignment and fixation was done after surgery using the Knee Society roentgenographic evaluation system. 9 Complications were recorded by the examining physician at each visit and obtained by the research assistant during chart review.

Preoperative and postoperative clinical assessments according to the Hospital for Special Surgery clinical rating scale were done at each visit. This system includes a score for pain, function, and radiologic assessment. 10

The VAS system was used to evaluate pain at the TKA site before surgery and at each followup. The scale consists of a 100-mm horizontal line, ranging from 0 (no pain) to 100 (intolerable pain). Patients were instructed to mark a point on the line that matched their pain. With a ruler, the number of millimeters was measured and converted to the same number of points. 8 The McGill Pain Questionnaire-Short Form provided a standardized measure of the affective and sensory dimensions of pain. 12

The WOMAC is a self-administered health questionnaire designed for patients with hip or knee arthritis. It includes 24 multiple-choice questions, divided into three categories: pain, stiffness, and physical function. The answers are ranked on a Likert 5-point scale (5 = worst, 1 = best). The instrument’s reliability and validity have been established in patients with knee arthroplasties. 2,3

Radiologic evaluation using the Knee Society roentgenographic evaluation system was done using the long-leg AP and lateral radiographs taken with the patient standing, and patella skyline views. 9

The JRISDAT physician assessment form was used to record the physical and functional data obtained during the physician’s examination, such as knee flexion and extension, mediolateral and AP laxity, and strength (Appendix 1).

The State-Trait Anxiety Inventory provided a standardized measure of an individual’s tendency to experience anxiety symptoms. 22 We used the Beck Depression Inventory as a standardized measure of depressive symptoms. 18 The Perceived Stress Scale is a brief self-report questionnaire that assesses an individual’s perception of current life stress. 15

Statistical analysis was done using the SPSS for Windows statistical package (SPSS Inc, Chicago, IL). Independent variables (predictors) included the demographic, physiologic, psychometric, heightened pain (VAS > 40) and test data described. Dependent variables (outcomes) were pain (VAS and McGill Questionnaire-SF scores), Knee Society scores, and healthcare use. Because many patients contributed more than one score to an analysis (separate outcome scores for each knee), multilevel models were used to nest knee scores within each person to control within patient similarity in preoperative variables. Multiple regression analyses were used to evaluate the relationship between the predictors and the outcomes. Significance was set as p < .05. Regression diagnostics were done to determine the effect of outliers on the observed relationships.

For the analyses regarding pain with time, missing observations at any followup precluded a patient from consideration, which greatly reduced the overall sample size. Therefore, in the instance where a patient had missing data in two or less followups, the last observation was carried forward to allow the patient to be included in the conducted analyses. Fifty-one patients had one observation carried forward and 20 patients had two observations carried forward.


Patients and Procedures

The mean age of the patients was 66 years (range, 36–85 years; SD 10.5) for the 116 patients in this series; 64 (55.2%) were women, 99 (85%) were Caucasian, 13 (11%) were African-American, one (0.9%) was Asian, one (0.9%) was Hispanic, and two (1.7%) had other ethnic backgrounds. The mean body mass index was 30.4 (range, 20.5–56.4). The population was unusual in its education status: 80 of 110 patients reported a college or higher level of education (six did not respond).

One hundred forty-nine TKAs were done in 116 patients. All implants were press-fit and cemented. The patella was resurfaced using a cemented all-PE component. The PCL was sacrificed in 63 knees (42.3%). Patients were seen preoperatively and at 1, 3, 6, and 12 months postoperatively.

Of the 149 arthroplasties done, 41 were on the left side, 42 were on the right side, and 33 were bilateral. One patient had bilateral procedures, but because they were 1.5 years apart, they were entered as two primary operations. Although most patients had primary OA of the knee, one case resulted from osteochondritis desiccans, two from gout, and two from posttraumatic arthritis. One patient had joint disease related to acromegaly and one had disease related to HIV infection. Preoperative systemic comorbidities included history of cancer (seven patients), gastrointestinal illness (five patients), hepatobiliary (three patients), osteoporosis (one patient), vascular disease (14 patients), cardiovascular disease (60 patients), genitourinary disease (four patients), psychiatric illness (one patient), diabetes (five patients), obesity (10 patients), respiratory disease (six patients), and hematologic or HIV disease (two patients).

There were 20 complications (Table 1). One patellectomy and one revision operation were done. A patient had an acute patella fracture 6 months after surgery and had patellectomy. She was excluded from the regression analysis. Another patient had arthrofibrosis and had multiple arthroscopic debridement procedures and ultimately required knee revision surgery. He also was excluded from the analysis. One patient died of lung cancer 1 year after surgery, and another patient died of pneumonia 18 months after surgery. One patient was receiving workers’ compensation. This patient had significant postoperative pain, and although the patient did not have revision surgery within the study period, the patient did have a knee revision surgery planned for the treatment of intractable pain. This patient was included in the analyses.

Complications of Total Knee Arthroplasty

The use of resources was evaluated by the outcomes of length of acute care hospital stay, transfer to inpatient rehabilitation, length of rehabilitation hospital stay, number of home health visits by a physical therapist, and number of outpatient visits to a physical therapist. The average length of stay in acute care was 4.8 days (range, 3–10 days; SD, 1.4). Sixty patients who had 71 TKAs (48%) were discharged from the acute care hospital to home with home physical therapy, and 56 patients who had 78 procedures (52%) were transferred to inpatient rehabilitation. The average length of stay in a rehabilitation facility was 12.7 days (range, 3–45 days; SD, 7.6). Nineteen patients (19 knees; 12.8%) had closed manipulation under general anesthesia.

Test Scores

The mean Knee Society knee score was 45.4 points (range, 8–89 points SD, 16.3) preoperatively and 89.7 points (range, 45–100 points; SD, 12.5) at last followup. The mean Knee Society function score was 46.1 points (range, 0–90; points SD, 18.8) preoperatively and 75.0 points (range, 5–100 points; SD, 21.3) at last followup. The mean flexion was 105° (range, 0°–135°; SD, 24.1°) preoperatively and 112° (range, 0°–140°, SD, 15.4°) at last followup. The Knee Society roentgenographic system showed no radiolucencies in any of the knees. In all patients, postoperative alignment was within the acceptable postoperative range (5°–10° standing mechanical axis).

Normal postoperative pain recovery was evaluated with the VAS. The VAS (SD) values at the preoperative and 1-, 3-, 6-, and 12-month visits were 52.6 (24.4), 36.8 (21.8), 25.4 (21.3), 20.5 (20.1), and 16.6 (21.0), respectively (Fig 1). Differences were observed with time as pain significantly decreased at all followups, F (3.05, 366.5) = 84.7, p < .001 (using the Greenhouse-Geiser correction). Significant pain (VAS > 40) was reported by 72.3% of patients preoperatively, 44.4% of patients at 1 month, 22.6% of patients at 3 months, 18.4% of patients at 6 months, and 13.1% at of patients at 12 months. No differences based on anesthesia type, weight, age, or gender were identified (Fig 2).

Fig 1
Fig 1:
The graph shows average pain scores (VAS) after TKA at the preoperative (preop.) and 1-, 3-, 6-, and 12-month visits. Significant differences were observed at all points, p < .05.
Fig 2
Fig 2:
The graph shows percentage of patients with significant pain (VAS > 40) after TKA at the preoperative (preop.) and 1-, 3-, 6-, and 12-month (mo) visits. Approximately one in eight patients has significant pain 1 year after surgery.

We evaluated the influence of several preoperative variables on postoperative pain. Preoperative depression and anxiety symptoms predicted more pain at 1 year after surgery (Table 2). Regression diagnostics were done on the observed relationships, and even after excluding all outliers, midrange anxiety scores remained a strong predictor of pain at 1 year after surgery (p < .03) (Fig 3). Significant symptoms of depression were reported by only a few patients and were strongly associated with high levels of pain at 1 year. Patients with the greatest preoperative levels of depressive symptoms complained of the worst postoperative pain 1 year after surgery (Fig 4). However, after removal of outliers, midrange depressive symptoms were not predictive of pain at 1 year (p > .05).

Correlation Between Preoperative Variables and Visual Analog Scale Pain at Followup
Fig 3
Fig 3:
The graph shows scatter plot of State-Trait Anxiety Index (STAI) anxiety scores and pain 1 year after surgery.
Fig 4
Fig 4:
The graph shows scatter plot of Beck Depression Inventory (BDI) depression scores and pain 1 year after surgery.

Heightened preoperative pain is an independent risk factor for poor outcome. Patients with greater preoperative pain (VAS > 40) were no different from the overall group in terms of age, gender, BMI, or other demographics. The mean preoperative flexion in the patients with greater pain was 103° (range, 0°–130°; SD, 26.4°) not significantly different from the 111° (range, 35°–135°; SD, 17.7°) in patients with less preoperative pain (VAS < 40; 36 knees). The mean Knee Society knee score for the patients with greater pain was 42.6 points (range, 11–81 points; SD, 14.6) preoperatively and 89.7 points (range, 55–100 points; SD, 12.1) postoperatively. The mean Knee Society function score was 43.9 points (range, 0–90 points; SD, 17.7) preoperatively and 71.9 points (range, 5–100 points SD, 22.8) postoperatively for the patients with greater pain. Despite no significant difference in the preoperative Knee Society function score between patients with a preoperative VAS greater than 40 or VAS less than 40, there is a significant difference at 6 months and 1 year (Table 3). Therefore, greater preoperative pain predicted worse function as described by the Knee Society function scores 1 year after surgery. Greater preoperative pain also predicted the use of more home physical therapy visits (p < 0.01) and longer inpatient rehabilitation stays (p < 0.05). Patients with greater preoperative pain had more postoperative manipulations than patients with less preoperative pain (Table 4).

Comparison of Knee Society Scores
Comparison of Resource Use by Groups with Preoperative VAS > 40 or < 40


Many patients experience significant pain 6 to 12 months after TKA despite an absence of clinical or radiographic evidence of abnormalities as identified on the standard outcome instruments. Clinical experience has identified a large percentage of patients who have some persistent pain and stiffness, but these data are confounded by little correlation between surgeons’ assessment of pain (the parameter used in most outcome reports) and patients’ self-assessments. 5,11,23

The current study outlined the typical pain recovery pattern after TKA. In most cases, using the VAS as the patient self-reported pain intensity, pain after TKA reduces to 1/2 its preoperative level by 3 months after surgery. This description is an easily understandable concept for patients and can be used as an answer to the common question, how long will pain last after surgery? However, at 3 months after surgery, approximately 1/4 of patients with otherwise normal TKA outcomes continue to describe significant pain (VAS > 40). By 1 year, approximately one in eight patients still complain of significant pain, despite an absence of clinical or radiographic abnormalities. Surgeons can accurately tell patients that although most TKA pain declines rapidly between 6 and 12 weeks after surgery, many still have significant pain at 3 months and approximately one of eight patients complain of significant pain at 1 year. The pattern of recovery is longer for some individuals.

Psychosocial variables seem to play a significant role in the pain response, in confirmation of many surgeons anecdotal suspicions 6,22 Pain and depression are associated with a greater use of healthcare services as patients and their physicians search for answers. 4,20 We found that patients with preoperative depression and anxiety complained of more pain after surgery and used more postoperative resources. Because patients were excluded from this study if they had a diagnosis of or were being treated for depression, the study patients had previously undiagnosed depression. Therefore, we think that untreated depression preoperatively, or even higher than normal depressive symptoms, is an independent risk for severe postoperative pain and may explain a subset of those patients with unexplained pain after surgery. Identification and treatment of depression and anxiety before surgery therefore may be an important strategy to improve outcome after TKA surgery.

We found that patients who are likely to require prolonged pain management can be identified before surgery. The study results identified heightened preoperative pain (VAS > 40) as a risk factor for manipulations after surgery. As a consequence of their heightened preoperative pain, these patients may have compensatory techniques of joint protection, muscle guarding, or activity limitation that adversely affects their postoperative ability to achieve flexion, resulting in a greater use of services (inpatient and outpatient rehabilitation), including manipulations. Identifying these patients before surgery and instituting aggressive pain management treatment, including drugs, physical therapy, and patient education may be a strategy to reduce poor outcomes.

The economic impact of severe, unexplained pain after TKA is profound. Frustrated physicians and patients pursue diagnostic tests, extended physical therapy, and even manipulations. Simple, straightforward, validated tests are needed to identify patients at risk for a poor prognosis before surgery; and cost effective, outcome-based preoperative and postoperative treatment protocols need to be developed to improve overall TKA outcomes. We found elevated preoperative VAS to be predictive of heightened postoperative pain. This scale is incorporated easily into the preoperative and postoperative visits. The depression and anxiety scales used in this study (Beck Depression Inventory and State-Trait Anxiety Index) are quick, self-administered pen and paper tests that are highly validated in the psychologic and the pain literature. Perhaps administering these or other simple screening tools should be used routinely before TKA surgery, analogous to the psychologic tests that are administered to patients before spinal surgery. Surgeons should be aware of patients at risk, including patients with high psychologic distress scores, so that surgeons can supervise these patients’ recovery and use of services more closely. For example, patients with heightened preoperative anxiety might be required to attend preoperative classes focusing on education and self-efficacy as a method to improve their postoperative outcome. Patients who have elevated preoperative scores on depression measures might be required to have at least a screening visit with a psychologist to assess the presence of severe depression and initiate treatment programs if indicated. Perhaps more careful attention should be given to the aggressive management of preoperative pain as a method to improve postoperative outcome and reduce the use of resources after surgery. Development of office-based assessment and management protocols or triaging patients at risk to tailored interdisciplinary pain programs may provide a method to reduce this excess morbidity.

APPENDIX 1. Joint Reconstruction and Implant Service Database (JRISDAT) physician assessment form is shown.




1. American Academy of Orthopaedic Surgeons: Primary total hip and total knee arthroplasty projections to 2030. Available at: Accessed November 1, 2002.
2. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW: Validation study of WOMAC: A health status instrument for measuring clinically important patient-relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. J Rheumatol 15:1833–1840, 1988.
3. Bombardier C, Melfi CA, Paul J, et al: Comparison of a generic and a disease-specific measure of pain and function after knee replacement surgery. Med Care 33(Suppl):131–144, 1995.
4. Chang RW, Pellissier JM, Hazen GB: A cost-effectiveness analysis of total hip arthroplasty for osteoarthritis of the hip. JAMA 275:858–865, 1996.
5. Coyte PC, Hawker G, Croxford R, Allard C, Wright JG: Variations in rheumatologists’ and family physicians perceptions of the indications for and outcomes of knee replacement surgery. J Rheumatol 23:730–738, 1996.
6. Creamer P, Hochberg MC: The relationship between psychosocial variables and pain reporting in osteoarthritis of the knee. Arthritis Care Res 11:60–65, 1998.
7. Davis MA, Ettinger WH, Neuhaus JM, Barclay JD, Segal MR: Correlates of knee pain among US adults with and without radiographic knee osteoarthritis. J Rheumatol 19:1943–1949, 1992.
8. Dolan P, Sutton M: Mapping visual analogue scale health state valuations onto standard gamble and time trade-off values. Soc Sci Med 44:1519–1530, 1997.
9. Ewald FC: The Knee Society knee arthroplasty roentgenographic evaluation and scoring system. Clin Orthop 248:9–12, 1989.
10. Gore DR, Murray MP, Sepic SB, Gardner GM: Correlations between objective measures of function and a clinical knee rating scale following total knee replacement. Orthopedics 9:1363–1367, 1986.
11. Harris WH, Sledge CB: Total hip and total knee replacement. N Engl J Med 323:801–807, 1990.
12. Holroyd KA, Holm JE, Keefe FJ: A multi-center evaluation of the McGill Pain Questionnaire: Results from more than 1700 chronic pain patients. Pain 48:303–311, 1992.
13. Kellgren J, Lawrence J: Radiological assessment of osteoarthrosis. Ann Rheum Dis 16:494–501, 1957.
14. Kessler S, Guenther KP, Puhl W: Scoring prevalence and severity in gonarthritis: The suitability of the Kellgren & Lawrence scale. Clin Rheumatol 17:205–209, 1998.
15. Koh KB, Park JK, Kim CH, Cho S: Development of the stress response inventory and its application in clinical practice. Psychosom Med 63:668–678, 2001.
16. Lieberman JR, Dorey F, Shekelle P, et al: Differences between patients’ and physicians’ evaluations of outcome after total hip arthroplasty. J Bone Joint Surg 78A:835–838, 1996.
17. Meding JB, Keating EM, Ritter MA, Faris PM: Total knee arthroplasty after high tibial osteotomy: A comparison study in patients who had bilateral total knee replacement. J Bone J Surg 82A:1252–1257, 2000.
18. Morley S, Williams CA, Black S: A confirmatory factor analysis of the Beck Depression Inventory in chronic pain. Pain 99:289–295, 2002.
Munin MC, Kwoh CK, Glynn N, Crossett L, Rubash HE: Predicting discharge outcome after elective his and knee arthroplasty. Am J Phys Med Rehabil 74:294–301, 1995.
20. Murphy Jr WA, Altman RD: Updated osteoarthritis reference standard. J Rheumatol Suppl 43:56–59, 1995.
Subbarao JV, Blair SJ: Reflex sympathetic dystrophy syndrome. Phys Med Rehabil 9:31–49, 1995.
    22. Thomee P, Thomee R, Karlsson J: Patellofemoral pain syndrome: Pain, coping, strategies and the degree of well-being. Scand J Med Sci Sports 12:276–281, 2002.
    23. Zavadak KH, Gibson KR, Whitley DM, Britz P, Kwoh CK: Variability in the attainment of functional milestones during the acute care admission after total joint replacement. J Rheumatol 22:482–487, 1995.
    © 2003 Lippincott Williams & Wilkins, Inc.