Dr. Fortin is a Senior Research Scholar of The Arthritis Society.
Timing of total joint replacement affects clinical outcomes among patients with osteoarthritis of the hip or knee
Article first published online: 12 DEC 2002
Copyright © 2002 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 46, Issue 12, pages 3327–3330, December 2002
How to Cite
Fortin, P. R., Penrod, J. R., Clarke, A. E., St-Pierre, Y., Joseph, L., Bélisle, P., Liang, M. H., Ferland, D., Phillips, C. B., Mahomed, N., Tanzer, M., Sledge, C., Fossel, A. H. and Katz, J. N. (2002), Timing of total joint replacement affects clinical outcomes among patients with osteoarthritis of the hip or knee. Arthritis & Rheumatism, 46: 3327–3330. doi: 10.1002/art.10631
- Issue published online: 12 DEC 2002
- Article first published online: 12 DEC 2002
- Manuscript Accepted: 8 AUG 2002
- Manuscript Received: 25 JAN 2002
- Canadian Orthopaedic Foundation
- Canadian Arthritis Network
- NIH. Grant Number: AR-36308
- Arthritis Foundation
- NIH. Grant Numbers: K24-AR-02123, P60-AR-47782
To determine the predictors of outcome in patients with osteoarthritis 2 years after receiving total hip or knee replacement.
A prospective cohort study of 222 osteoarthritis patients undergoing total hip or knee replacement in Boston and Montreal was done. Their postoperative outcomes at 6 months were previously reported. This followup reports on the outcomes after 2 years among the 165 patients (74%) who remained. The subjects were divided into 2 groups according to the median value of their preoperative Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) function score. The Short Form 36-item physical function subscale and the WOMAC pain and function subscale scores were collected at baseline and at 3, 6, and 24 months postoperatively. Clinical outcomes were analyzed at 2 years, using descriptive and multiple regression analyses.
Improvements in pain and function at 2 years were similar to those observed at 6 months. Those subjects with the worst function and pain at the time of surgery (baseline) had comparatively worse function 2 years after surgery.
In this comparison, the poor outcomes observed at 6 months following total joint replacement in patients with worse baseline functional status persisted after 2 years. Although there are no validated indications for when a patient should optimally have total joint replacement, these data suggest that timing of surgery may be more important than previously realized and, specifically, that performing surgery earlier in the course of functional decline may be associated with better outcome.
Total hip replacement (THR) and total knee replacement (TKR) are both highly successful and cost-effective procedures for end-stage joint disease. The influence of the timing of these two procedures on patient outcomes has not been studied formally. In a natural experiment, 2 cohorts of patients treated in academic health centers in Boston and Montreal were studied. Although all patients showed improvement at 6 months postoperatively, those with the worst function and pain at the time of surgery (baseline) had comparatively poorer function at followup (1). In this study, we examined whether these results persisted up to 2 years after surgery, when the majority of patients have achieved optimal postoperative results.
PATIENTS AND METHODS
We performed a prospective cohort study of 222 consecutive persons undergoing elective THR and TKR for osteoarthritis at a Boston and a Montreal teaching hospital. At 2 years postoperatively, followup information was available on 165 patients.
All consecutive patients undergoing elective surgical knee or hip replacement for primary osteoarthritis at Brigham and Women's Hospital in Boston and at Montreal General Hospital in Montreal between August 5, 1994, and June 3, 1996, were asked to participate in this study (1). Patients were assessed preoperatively (baseline), and at 3, 6, and 24 months after surgery. A demographic questionnaire on age, sex, and education (number of years of education) and the Comorbidity Illness Rating Scale (2) were collected at baseline. The cohort was divided into two groups according to the median value of their baseline Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) physical function score (high and low function groups) (3). At each assessment, patients completed the Medical Outcomes Study Short Form 36 (SF-36) (a generic measure of quality of life) (4) and the WOMAC (a disease-specific measure of quality of life) (3).
The SF-36 has 8 subscales with scores ranging from 0 (worst) to 100 (best). The WOMAC has 3 subscales that measure pain, stiffness, and function. Each WOMAC subscale score results from the summation of selected items, and each item is scored from 0 (absence of pain, stiffness, or functional limitation) to 4 (extreme pain, stiffness, or functional limitation). WOMAC scales range from 0 to 68 for function, from 0 to 8 for stiffness, and from 0 to 20 for pain. In contrast to the SF-36, a higher WOMAC score represents greater limitation. We elected a priori to examine the SF-36 physical function subscale and the WOMAC pain and function subscales as the primary outcomes, and to report the 24-month results.
The ethics committees of both participating institutions reviewed and approved this study.
Between-center differences preoperatively and at 24 months postoperatively were calculated for the 3 selected subscales. We performed separate multiple linear regression models for THR and TKR. We hypothesized that worse preoperative function is associated with worse postoperative function. Since we have previously demonstrated a strong correlation between center and baseline function (1), we chose to include baseline subscale scores, and not the center, as one of the independent variables in our models. These models used the physical function subscale of the SF-36 or the pain and function subscales of the WOMAC at 24 months as dependent variables, and the baseline subscale scores, age, sex, education, and comorbidity as independent variables.
One hundred sixty-five (74%) of the original 222 patients completed the study at 24 months. The baseline characteristics of the 57 patients lost to followup were similar to those of the 165 participants in terms of age, sex, level of education, presence of comorbidity, baseline SF-36 physical function and WOMAC pain and function subscale scores, type of surgery (THR or TKR), and center at which surgery took place. Between the two centers, the mean age and the extent of comorbidity among the remaining 165 patients were not different, but other baseline characteristics did differ between the centers (Table 1). Boston patients had a higher mean number of years of education than did Montreal patients (difference 3.5 years, 95% confidence interval [95% CI] 2.3–4.6) and also had better scores on their SF-36 physical function subscale (difference 7.7, 95% CI 0.8–14.6) and on their WOMAC pain (difference −2.0, 95% CI −3.3, −2.7) and function (difference −7.5, 95% CI −11.7, −3.5) subscales. As previously noted (1), center and baseline function scores were highly correlated, with Boston patients having better function at the time of surgery than that of Montreal patients. There were no differences in these same variables between those subjects undergoing THR and those receiving TKR.
|MGH (n = 60)||BWH (n = 105)||THR (n = 84)||TKR (n = 81)|
|Sex, % male||33||49||41||44|
|Age, mean ± SD years||65.8 ± 8.4||67.5 ± 8.8||65.7 ± 8.2||68.1 ± 9.1|
|Education, mean ± SD years||11.3 ± 3.7||14.8 ± 2.9||14.0 ± 3.7||13.1 ± 3.5|
|CIRS score, mean ± SD||2.5 ± 1.6||1.95 ± 1.28||1.9 ± 1.3||2.4 ± 1.5|
|Baseline function and pain|
|SF-36 physical function score, mean ± SD||22.7 ± 20.2||30.4 ± 23.6||27.9 ± 24.0||27.3 ± 21.4|
|WOMAC pain score, mean ± SD||11.2 ± 4.3||9.2 ± 3.2||10.0 ± 3.8||9.9 ± 3.8|
|WOMAC function score, mean ± SD||39.8 ± 13.3||32.3 ± 11.8||36.7 ± 12.7||33.4 ± 13.0|
Postoperative status at 24 months.
Compared with the outcomes observed at 6 months, the results of the analysis show that there was no further improvement in function or pain at 24 months. Table 2 shows the comparisons of the between-group (high versus low function) and within-group differences in function and pain on the selected SF-36 and WOMAC subscales at 24 months for THR and TKR. Figure 1 illustrates this for TKR outcomes on the basis of the WOMAC physical function subscale scores, in which the gap between the high and low physical function groups that was observed at 6 months remains the same at 24 months. A similar pattern could be observed on the SF-36 physical function and the WOMAC pain subscales between 6 and 24 months.
|High function||Low function||Difference|
|Mean ± SD||95% CI||Mean ± SD||95% CI||Mean||95% CI|
|Total hip replacement (n = 84)|
|SF-36 physical function|
|Preoperative||39.4 ± 26.1||–||17.9 ± 16.6||–||21.5||11.8, 31.2|
|24 months||74.2 ± 24.7||–||59.5 ± 30.4||–||14.7||2.6, 26.8|
|Difference (24 months − preoperative)||35.4||26.2, 44.6||41.6||33.4, 50.0||−6.2||−18.4, 5.9|
|Preoperative||8.2 ± 3.3||–||11.5 ± 3.4||–||−3.3||−4.8, −1.9|
|24 months||2.2 ± 3.0||–||3.7 ± 4.3||–||−1.5||−3.1, 0.1|
|Difference (24 months − preoperative)||−6.1||−7.3, −4.8||−7.9||−9.3, −6.4||1.8||−0.03, 3.7|
|Preoperative||26.0 ± 9.2||–||45.9 ± 6.4||–||−19.9||−23.4, −16.3|
|24 months||9.4 ± 10.5||–||16.0 ± 13.3||–||−6.6||−11.8, −1.4|
|Difference (24 months − preoperative)||−16.6||−20.5, −12.7||−29.9||−33.5, −26.3||13.3||8.1, 18.5|
|Total knee replacement (n = 81)|
|SF-36 physical function|
|Preoperative||37.4 ± 22.6||–||14.3 ± 9.8||–||23.2||15.5, 30.8|
|24 months||65.7 ± 23.9||–||42.9 ± 25.9||–||22.8||11.5, 34.2|
|Difference (24 months − preoperative)||28.0||18.9, 37.0||28.6||20.7, 36.5||−0.7||−12.5, 11.2|
|Preoperative||7.7 ± 3.1||–||12.6 ± 2.7||–||−4.9||−6.2, −3.6|
|24 months||2.8 ± 3.5||–||6.2 ± 4.4||–||−3.4||−5.2, −1.6|
|Difference (24 months − preoperative)||−4.9||−6.1, −3.8||−6.4||−8.1, −4.7||1.5||−0.5, 3.5|
|Preoperative||24.4 ± 8.9||–||44.9 ± 6.7||–||−20.6||−24.1, −17.1|
|24 months||10.3 ± 11.4||–||21.4 ± 13.0||–||−11.1||−16.6, −5.6|
|Difference (24 months − preoperative)||−14.0||−17.7, −10.4||−23.5||−28.1, −19.0||9.5||3.8, 15.2|
Furthermore, the results of the multiple regression models were very similar to those at 6 months (data not shown). The baseline subscale score was always the best predictor of outcomes at 24 months, with independent contributions from the level of education and the rating of comorbidity. The 24-month data also replicated our 6-month findings in that function and pain in patients with lower preoperative physical function were not improved postoperatively to the level achieved by those with higher preoperative function. In addition, at 24 months, there were 5 times as many persons requiring assistance from another person for their activities of daily living (ADL) in the low function group compared with those in the high function group (relative risk 5.2, 95% CI 1.9–14.6; unadjusted bivariate analysis) (results not shown).
The average physical function score on the SF-36 for a Canadian population between the ages of 65 and 74 years is 75.7 (95% CI 74.9–76.5) (5). In the present study at 2 years after surgery, the SF-36 physical function score for the low function group was 59.5, whereas for the high function group it was 74.2, showing a trend toward the population average. Moreover, there were 5 times as many persons in the low function group requiring assistance from another person for their ADL.
This raises the question as to what might be gained by performing surgery later in the disease course. One possible benefit is the one-time cost savings that results from delaying total joint replacement, which yields a lower present discounted cost of the procedure. A second cost savings would be the attrition of certain patients before the need arises for total joint replacement or for revision surgery. A formal cost-effectiveness study is needed to examine the economic impact of a strategy to intervene with total joint replacement earlier in the course of osteoarthritis. Our data, we believe for the first time, strongly suggest that the timing of total joint replacement is critical to the 2-year results. Since we had no information on the importance of contralateral joint involvement, we cannot comment on this potentially important parameter of long-term function in our study population; it is possible that the lower health status in the low function group might have been due to problems in other joints, and not related to the joint that had the surgery. While THR and TKR can arrest or reverse the decline in function at 2 years, we cannot decisively comment on the impact on overall function that the timing of surgery in one joint will have beyond 2 years.
The results of our study suggest that patients operated upon earlier in the course of functional decline had better outcomes. The differences in functional status were documented at 6 months (1) and persisted through 24 months. This observation supports a strategy to intervene earlier with total joint replacement in patients with lower extremity osteoarthritis.