To evaluate health-related quality of life (HRQOL) in patients with osteoarthritis undergoing total knee replacement (TKR); identify the influence of sociodemographic, clinical, intraoperative, and postoperative variables on HRQOL; and determine patient perceptions at 7 years.
We conducted a prospective study with 7 years of followup. HRQOL measures (Western Ontario and McMaster Universities Osteoarthritis Index [WOMAC] and Short Form 36 [SF-36]); sociodemographic, clinical, intraoperative, inpatient, and postoperative data; patient perceptions of TKR outcomes; and physical activity at 7 years were determined. Associations were analyzed using linear regression models.
Of 146 eligible patients, 112 (86 women, mean age 67.3 years) completed followup data. There were significant differences between pre- and postoperative WOMAC pain, stiffness, and function scores (P < 0.001). Variables retained in each of the models explained 14–32% (adjusted R2) of variability of the WOMAC dimensions. Obesity and postdischarge complications were associated with worse scores in all WOMAC dimensions (P < 0.05). Eighty-six percent of patients were satisfied with TKR, 80% would undergo the operation again, and 56% did regular physical activity and had better WOMAC scores (P < 0.05, except for stiffness [not significant]). Mean ± SD SF-36 scores for men and women at 7 years were 55.1 ± 27.1 and 39.5 ± 22.9 for physical function, 71.2 ± 36.5 and 51.5 ± 42.7 for physical role, 66.2 ± 26 and 55.6 ± 28.9 for bodily pain, and 60.7 ± 17.1 and 50.7 ± 21.2 for general health, respectively.
WOMAC dimension scores, especially pain, significantly improved at 7 years and were negatively influenced by obesity and postdischarge complications. HRQOL measures may help identify an increased risk of negative outcomes after TKR.
Knee osteoarthritis (OA) is one of the major causes of pain and disability in developed countries, particularly in aging populations (1–3), with the consequent loss of health-related quality of life (HRQOL) (4, 5). In Spain, the estimated prevalence of symptomatic knee OA is 10.2% (1, 2). Therefore, the demand for total knee replacement (TKR) is increasing as patients gain considerable pain relief and increased mobility and HRQOL (6).
Sociodemographic characteristics, including sex, age, and weight, and clinical and surgical factors have been reported to influence outcomes in some studies, but not in others (7, 8). A recent review of total joint arthroplasties and outcomes suggests that indications and contraindications for TKR are poorly defined, and little is known about the determining factors (9).
Reports assessing outcomes after TKR vary in the perspective of the evaluation and the variables and data analysis used. Some reports have determined long-term HRQOL outcomes or examined the magnitude and meaningfulness of improvements/changes following orthopedic surgery. There remains a lack of evidence (which is needed to guide patient selection and clinical decision making) and difficulties in establishing the best indication for TKR (8, 9). As stated by Altman et al, “Despite the success of total joint replacement of the hip and knee over the last 30+ years, the criteria for when to perform such surgery are not clear. The level of pain needed to indicate surgery is unclear” (10).
Although most reports assessing the effectiveness of TKR have focused on surgical and technical aspects, patient-reported outcomes provide an alternative source of information. Disease-specific or generic HRQOL instruments may provide valuable information and are widely used to evaluate health outcomes in clinical practice and, increasingly, to estimate improvements in pain and function after surgery (11). In knee OA, the disease-specific Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) questionnaire, version LK 3.0 (12) is recommended for monitoring functional outcomes in TKR (7). The most common generic measure of HRQOL in any disease or population is the Medical Outcomes Study 36-item Short Form Health Survey (SF-36) questionnaire, which allows age- and sex-standardized comparisons to be made with normal population values (NPVs) (13).
The objectives of this study were to evaluate HRQOL in patients with severe OA undergoing TKR; identify the influence of sociodemographic, clinical, intraoperative, and postoperative variables on HRQOL; and determine patient perceptions of TKR at 7 years.
PATIENTS AND METHODS
A prospective 7-year cohort study was carried out in the Rheumatology and Orthopedic Surgery (Knee Unit) Services of the Hospital Clínic Provincial, Barcelona, Spain, which is an urban tertiary care center. Participants were enrolled between January and November 2000. The study received approval from the hospital ethics committee.
Consecutive patients of all ages admitted to the Knee Unit for primary TKR with a diagnosis of severe knee OA (according to Kellgren and Lawrence criteria) (14) were included when they agreed to participate in the study and gave informed consent. Exclusion criteria were functional illiteracy or psychopathology severe enough that the patient could not fully participate in the study procedures. Patients who agreed to participate were informed of the need to attend a hospital interview to determine the outcome of TKR, and that they would be reminded by telephone or letter.
All of the patients underwent TKR. Surgical procedures and all care and treatment, including rehabilitation, were standardized according to hospital protocols (15). All patients were fitted with a standard (nonconstrained) prosthesis. At discharge, all patients received a booklet with advice on wound care, signs of possible complications, and rehabilitation exercises.
Variables determined and assessment intervals.
Patients were interviewed the day before surgery (baseline). At baseline, self-reported disease-specific HRQOL was measured by the Spanish version of the WOMAC questionnaire (16), which contains 3 dimensions: pain, stiffness, and function. A total score combining the 3 dimensions may be used. All WOMAC data were normalized to a 0–100 scale for each WOMAC dimension, where 0 = the best health status and 100 = the worst health status (12, 16). An independent researcher was present to provide aid, if necessary, for patients answering the questionnaire.
At baseline, a structured questionnaire recorded information on 1) sociodemographic characteristics (age, sex, education level, type of cohabitation, and occupational status); 2) disease duration (years since diagnosis); 3) chronic musculoskeletal pain unrelated to knee OA using the question: “Do you have any other rheumatic pain, in addition to your knee?” (yes/no); 4) number and type of preexisting comorbidities at baseline (self-report), according to the items of the regional Catalan Health Survey 2006 (17); 5) body mass index (BMI) (18), obtained by hospital measurements; and 6) previous prostheses (contralateral knee).
Medical records provided information on intraoperative surgical data (immediate postoperative) and inpatient medical data at discharge. To determine the degree of intraoperative difficulty (intraoperative surgical data), a multi-item questionnaire was used by surgeons to identify surgical problems and the degree of the intraoperative difficulty supposed by more than one variable of surgical difficulty in the same intervention. The items were chosen for their relevance after a literature search and use in previous studies (19, 20). The degree of intraoperative difficulty, defined as the sum of possible problems that could occur during TKR surgery, was scored as: 0 = no problems, 1 = slight problem (1 or 2 problems during the intervention), 2 = medium problem (3 or 4 problems), and 3 = substantial problem (>4 problems) (20). Inpatient clinical data included number and type of inhospital complications before discharge: problem healing of surgical wound (yes/no), signs of deep venous thrombosis (yes/no), immediate infection (yes/no), ambulatory status at discharge (climbing and descending stairs [yes/no]), and correct alignment of the lower extremity (deviations in the anteroposterior axis inferior to 3°/goniometer) by long-leg coronal radiograph findings (yes/no). The length of hospital stay in days was also recorded.
At 7 years postoperatively, the WOMAC questionnaire was evaluated. In addition, generic HRQOL was evaluated using the Spanish version of the SF-36 questionnaire (with respect to functional state: physical function, bodily pain, physical role, and general health) with a range of 0–100, where higher scores indicate better HRQOL (13).
At 7 years, we asked about 1) pharmacologic treatment, including current number of analgesics (acetaminophen) and nonsteroidal antiinflammatory drugs (NSAIDs) taken per week, and other medications related to knee OA in tablets per week; 2) patient perceptions of the result of TKR, including satisfaction in terms of improvement in pain and function on a 5-point Likert scale, with responses ranging from “very satisfied” to “unsatisfied” (17) and the question, “Would you undergo the operation again in similar circumstances?” (yes/no), “If not, why not?”; and 3) asking about regular postoperative physical activity in the previous 6 months using the question, “How many times a week do you exercise (low, moderate and high) for at least thirty minutes?” (17). Followup interviews were completed in the outpatient clinic by an independent researcher.
At 7 years, clinical data included (medical records): number and type of complications after discharge (lower extremity deformity, dislocation of the TKR [femorotibial, femoropatellar], infection, pain, septic or aseptic loosening [patellar, tibial, femoral], extensor muscle failure, deep venous thrombosis, and others [yes/no]).
Accepting an alpha risk of 0.05 and a beta risk of 0.20 in a bilateral contrast, 71 patients were needed to detect a difference of ≥10 points between mean pre- and postoperative scores for the WOMAC pain and physical function dimensions (12), which was judged to be a clinically important difference (12, 20). A common SD of 30 was assumed. The sample was overestimated by 20% to allow for possible losses.
A descriptive analysis was performed using univariate frequency tabulation for categorical variables or mean ± SD values for continuous variables. The difference between groups for categorical variables (patient perceptions of their TKR and regular physical activity) and WOMAC dimension scores were analyzed using the t-test.
The Student's t-test (paired samples) was used to evaluate differences between mean scores at baseline and 7 years in the WOMAC pain, stiffness, and function dimensions. Effect sizes (ES) were calculated for the different outcome measures using the formula ES = mean change/SD of preintervention (baseline) results.
Explanatory multiple linear regression models were constructed to determine the influence of independent variables (sociodemographic and clinical characteristics; intraoperative surgical, inpatient, and postoperative clinical variables; and preoperative WOMAC scores) on the dependent variables (WOMAC scores at 7 years). Variables were selected by combining the stepwise and all subset regression methods. Variables with more than 2 categories were coded as dummies: BMI was coded as BMI <35 kg/m2 and BMI ≥35 kg/m2. Separate regression models were constructed for the 3 WOMAC dimensions. All models included age, sex, and number of comorbidities as potentially confounding variables. Residual plots and standard diagnostics were used to check that model assumptions were verified. The results of the models are presented using the adjusted R2.
Differences between the study sample values and the age- and sex-matched Spanish NPVs were analyzed using the t-test. Spearman's correlation coefficient was used to evaluate the relationship between the WOMAC pain and function dimensions and the SF-36 bodily pain and physical function dimensions.
The 95% confidence intervals (95% CIs) were calculated. Statistical analyses were performed using SPSS, version 12.0 for Windows (SPSS, Chicago, IL).
One hundred forty-six patients were eligible for inclusion, 4 of whom refused to participate. Thirty patients (21%) were lost to followup (4 patients died of unknown causes, 7 refused to attend the interview, and 19 could not be contacted). The remaining 112 patients completed followup data and were included in the analysis. No significant differences between participants and nonparticipants were detected (Mann-Whitney U test for continuous variables and Fisher's exact test for categorical variables). Table 1 shows the main patient characteristics. All of the patients had ≥1 comorbidities.
Table 1. Sociodemographic and clinical characteristics*
Respondents (n = 112)
Nonrespondents (n = 34)
Values are the number (percentage) unless otherwise indicated. BMI = body mass index.
Age, mean ± SD years
67.3 ± 16.6
69.2 ± 11.4
Disease duration, mean ± SD years
13.9 ± 10.1
14.7 ± 12.3
No. comorbidities, mean ± SD
5.9 ± 3.0
5.4 ± 3.2
Type of comorbidity
Low back pain
BMI, mean ± SD kg/m2
30.6 ± 4.4
30.7 ± 5.2
With respect to current pharmacologic treatment, 23 patients were taking either low doses of NSAIDs plus acetaminophen (n = 6) or only acetaminophen (n = 14) and opioids (n = 3). Thirty-one patients were taking either NSAIDs plus acetaminophen (n = 8) or only acetaminophen (n = 23) when required.
A total of 86 (76.8%) patients were satisfied or very satisfied with the result of their TKR. Patients who were satisfied had significantly lower (better) postoperative WOMAC dimension scores (P ≤ 0.01 for all dimensions) (Figure 1). Eighty-nine (79.5%) patients said they would undergo the operation again in similar circumstances, and 23 (20.5%) said that they would not undergo the operation again due to difficult recovery (n = 9), increased pain (n = 8), and insufficient improvement (n = 6). Patients who responded positively had significantly lower (better) WOMAC dimension scores than those who did not (P < 0.05) (Figure 1).
Fifty-six percent of patients did regular physical activity at 7 years and had lower (better) WOMAC dimension scores than patients who did not (P < 0.05 except for the stiffness dimension [not significant]) (Figure 1).
With respect to the variables of surgical difficulty, all patients were operated on by 1 of the 3 senior surgeons, the mean ± SD surgical time was 130 ± 41 minutes, the mean ± SD ischemia time was 80 ± 12 minutes, and 2 patients required osteotomy of the tibial tuberosity. Forty-seven percent of the patients had no intraoperative difficulty (grade 0), and the degree of intraoperative difficulty was low in the remainder of the patients (grade 1).
The main inpatient complication was problem healing of the surgical wound (n = 10). All of the patients had correct alignment of the lower extremity. The mean ± SD length of hospital stay was 6.8 ± 1.3 days.
Seventy-seven percent of the patients (n = 86) had no complications during the 7 years of followup and 23% had some complications (mean ± SD 1.7 ± 0.9; infection [n = 1], pain [n = 16], septic [n = 3] or aseptic [n = 4] loosening, extensor mechanism failure [n = 1], deep venous thrombosis [n = 1], and 5 additional complications [2 femoropatellar problems, with patellectomy required in 1 case; 1 stiff knee due to arthrofibrosis; 1 periprosthetic femoral fracture; and 1 skin lesion over the anterior tibial tuberosity after osteotomy of the anterior tibial tuberosity during TKR surgery]).
Table 2 shows WOMAC scores at baseline and at 7 years. There were significant improvements in the 3 WOMAC dimensions (mean difference between baseline and 7 years of followup), as shown in Table 2 (P values of improvement). The ES is also shown in standardized units in Table 2, and shows an improvement of 51% for pain, 40% for stiffness, and 35% for function.
Table 2. WOMAC dimension scores at baseline and 7 years after total knee replacement*
WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; 95% CI = 95% confidence interval; ES = effect size.
By t-test (paired samples).
52.9 ± 16.3
25.9 ± 21.5
43.3 ± 26.6
25.9 ± 25.5
55.8 ± 17.6
36.4 ± 22.7
54.2 ± 16.0
33.3 ± 21.3
Table 3 shows sociodemographic and clinical characteristics, which were independently significant in the multiple linear regression analysis. The variables retained in each of the models explained between 14.1% and 32.5% (adjusted R2) of the variability of each WOMAC dimension.
Table 3. Variables independently associated with the WOMAC pain, stiffness, and function dimensions at 7 years in the multiple linear regression models*
Pain (adjusted R2 = 0.325)
Stiffness (adjusted R2 = 0.141)
Function (adjusted R2 = 0.207)
Coefficient (95% CI)
Coefficient (95% CI)
Coefficient (95% CI)
The coefficients of the regression models indicate if an increase in the independent variables is associated with an increase (worse; positive coefficient) or decrease (better; negative coefficient) in WOMAC dimensions. The adjusted R2 is the proportion of variance in the dependent variable explained by the relevant independent variables shown. WOMAC = Western Ontario and McMaster Universities Osteoarthritis Index; 95% CI = 95% confidence interval; BMI = body mass index.
BMI >35 kg/m2 (classes II and III)
Complications after discharge
Of the potentially confounding variables included in the analysis (age, sex, and number of comorbidities), only sex was independently associated with the WOMAC function dimension score. The remaining variables did not improve the models and produced only minimal modifications in the other coefficients. Therefore, they were excluded from the final models.
Female sex was associated with higher (worse) scores in the WOMAC function dimension (P = 0.015). Severe and morbid obesity, BMI ≥35 kg/m2 (classes II and III), and the number of complications after discharge were all associated with higher (worse) scores in all WOMAC dimensions (P < 0.05 for all).
WOMAC dimension scores analyzed according to sex showed no significant differences in any dimension at baseline, although women had higher (worse) scores than men. However, at 7 years, women had significantly higher (worse) scores than men in the function dimension (P = 0.005). The effect sizes for the function dimension were 1.1 in women and 1.4 in men (Figure 2).
Mean ± SD SF-36 scores for men and women at 7 years were 55.1 ± 27.1 and 39.5 ± 22.9 for physical function, 71.2 ± 36.5 and 51. 5 ± 42.7 for physical role, 66.2 ± 26 and 55.6 ± 28.9 for bodily pain, and 60.7 ± 17.1 and 50.7 ± 21.2 for general health, respectively.
There was a significant inverse correlation of 81% between the WOMAC function dimension and the SF-36 physical function dimension, and of 61% between the WOMAC pain dimension and SF-36 bodily pain dimension (P < 0.0001).
We studied health outcomes 7 years after TKR for severe OA and attempted to identify factors influencing outcomes, and found that HRQOL significantly improved after 7 years. Female sex, severe and morbid obesity, and complications after discharge negatively influenced WOMAC dimension scores.
At 7 years, there was a significant reduction in WOMAC dimension scores with the pain dimension showing the greatest improvement, which is in agreement with other studies with 5–11 years of followup (21, 22). This may be because in elderly people, pain is more likely to improve than overall function, which decreases with age (20). In addition, our findings show an ES lower than that described in the review by Kane et al (for WOMAC studies with 0–2 years of followup) and in a study with 3 years of followup. This may indicate a time effect on the results of TKR (8, 20). Chana et al also found that improvements in WOMAC scores 1 and 2 years after TKR fell at 5 years (21).
MacDonald et al found similar improvements (change scores) to ours in pain in a study with a mean of 7 years of followup, but reported greater improvements from preoperative to postoperative scores in the WOMAC stiffness and function dimensions (23). Although the preoperative pain dimension scores were similar to ours, the preoperative stiffness and function dimensions were higher (worse) and, as reported by other studies, patients with higher preoperative WOMAC scores had greater postsurgical improvement (20, 24). This suggests variability in the preoperative health status of patients undergoing TKR, which makes comparison between studies difficult (8, 9). This could be minimized by reporting not only change scores but also quantifying the magnitude of change (ES) following TKR. The ES is a standardized measure and provides information on the amount of change in the measure relative to the variation within the measure. This is not only more informative for physicians but may also suggest clearer indications for surgery (10, 25).
Women had significantly worse WOMAC scores at 7 years. This may be due in part to age, because the female participants were a mean of 5 years older than the male participants (69 versus 74 years), although age was not a significant factor in the regression analysis. In addition, women entered surgery with worse WOMAC scores than men, although again, this was not a significant factor. Petterson et al concluded that women are either more adversely affected by arthritis or further along in the disease process before considering arthroplasty (26). Reports vary on this aspect (5, 8), although this may be due to presentation of the results from different perspectives. A study with a mean followup of 10 years found that women improved more (change scores from preoperative to postoperative) after TKR than men in all of the WOMAC dimensions (23). In contrast, another study with 5–11 years of followup found that sex did not affect change scores after TKR (22). However, these studies reported that women had higher (worse) preoperative and postoperative WOMAC scores than men. The 2 studies coincide when outcomes are considered, but differ when improvements are considered (8, 21–23).
Severe and morbid obesity were significantly associated with worse scores in all WOMAC dimensions, which is in agreement with other studies (9, 20, 24, 27). Reports of obesity in these patients have mostly studied high mechanical prosthesis failure rates and long-term complications, whereas the consequence of weight on health outcomes after surgery is unclear (9). Some reports with a followup of 5–10.6 years have found that severely obese patients (mean BMI ≥30 kg/m2) have worse outcomes (measured by the Knee Society Scores and radiography) and a higher level of revisions (mean BMI ≥39 kg/m2) (28–30), whereas others found no differences between severely obese and nonobese patients (31–34). However, there is no definitive cutoff in BMI and studies use different classifications and categories of obesity (27).
The number of postdischarge complications during followup also had an independent impact on TKR outcomes measured by the WOMAC. A review by Jones et al found that the lack of success reported by some patients can be attributed to complications or surgical-related factors (9). A review by Kane et al found that complications, as in our study, are defined by each investigator, and reported low rates (8).
The significant independent variables in the regression analysis explained between 14% and 33% of the variability (Table 3), suggesting the multifactorial nature of the health status in TKR (20) and also suggesting that factors not included in our study influence the health status.
Although our results showed that pain and disability improved 7 years after surgery, 14% of patients stated dissatisfaction with their TKR (9, 35, 36). Although satisfaction may be affected by factors unrelated to the surgical intervention, such as the patient–surgeon relationship, health care, etc. (35), which we did not study, we believe that when patients state that they are dissatisfied with their TKR, they are mainly referring to the fact that they have not gained as much as expected in pain relief and functional capacity. Furthermore, we found that patient satisfaction was associated with disease-specific outcome measures, which is in agreement with the findings of Robertsson and Dunbar for the WOMAC (36), with the greatest improvements being in satisfied patients. These authors also reported that patient satisfaction measured by a Likert scale significantly correlated with the WOMAC questionnaire (36).
We found a good correlation between the WOMAC and SF-36 questionnaires (pain, 61% and function, 81%), as did another study (36). SF-36 scores at 7 years approached the NPVs in all dimensions except for the physical function dimension, which remained worse. Similar findings have been described in other studies (at 5 years of followup) (9, 25). Weiss et al suggests that function after TKR may not be as satisfactory for patients as previous studies have suggested (37).
Female participants with a mean age of 74 years had a mean SF-36 function score of 39.5, which was lower than the Spanish NPVs for women in the 65–74 years age group (61.3) and the >75 years age group (45.2). Men with a mean age of 69 years had a mean score of 55.1, which was lower than the Spanish NPV for the 65–74 years age group (68.9) (13).
Exercise has been associated with significant improvements in function and pain in older adults with knee OA (4). For this reason, we emphasized the central role of exercise to our patients and evaluated it postoperatively (4). Regular postoperative physical activity positively influenced the health status, which is in agreement with other studies that found that exercise could benefit TKR outcomes (38). However, because we only measured physical activity in the 6 months before the 7-year control, we cannot assume that exercise favored better function or that this measurement was indicative of the entire 7-year period; it may be that in some patients, worse physical function was a causal factor in the lack of exercise.
The study was carried out in a single tertiary reference center. Therefore, the results are only representative of this group of patients and might vary in other geographic areas, although controlled data from a single center may provide valuable information. We could not contrast our results with those of Spanish studies because to our knowledge there are none, and we could only compare them with the results from countries with different health systems and cultures. We did not analyze possible confounding variables, including social support and the appropriateness of joint replacement. Therefore, our findings should be interpreted with care, although other studies have reported similar patient characteristics with respect to the preoperative state and the improvements obtained after TKR (21–23). In addition, making only 2 assessments during a 7-year period means that important factors such as speed of recovery from surgery or potential declines 3–5 years after surgery could not be studied.
To our knowledge, this is one of a few prospective long-term followup studies focusing on health outcomes in patients who had TKR for severe OA, including wide-ranging information on the amount of pain relief and functional improvement (magnitude of change) after surgery, patient perceptions on outcomes after TKR, and the role of regular physical activity on postoperative health outcomes. In addition, the degree of intraoperative difficulty has not been studied in this type of patient. We identified factors that negatively influence long-term outcomes after TKR and that could be minimized with better management.
In conclusion, our results show that specific HRQOL significantly improved 7 years after TKR in patients with severe OA. Female sex, severe obesity, and complications after surgery negatively influenced health outcomes measured by the WOMAC. However, 14% of patients stated that they were dissatisfied with the results of their TKR.
Improvements in pain and function suggest that other factors require deeper study, given the multifactorial nature of TKR. Standardized criteria are necessary to measure short- and long-term outcomes in TKR. HRQOL measures may help identify patients at an increased risk of negative outcomes after TKR.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Montserrat Núñez had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Montserrat Núñez, Lozano, Esther Núñez, Maculé.
Acquisition of data. Lozano, Sastre, Ortega.
Analysis and interpretation of data. Esther Núñez, Segur, Suso.
We thank F. Segura, G. Navarro, and D. Buss for their help and advice.