Introduction

Osteoarthritis (OA) is the most common cause of musculoskeletal disability and pain in the world. In Finland, 6% of men and 8% of women over the age of 30 years suffer from clinically diagnosed knee OA. The incidence of radiological and clinical knee arthritis increased with age both in men and women 1. During 2004, a total of 5905 total knee replacements (TKRs) were performed with the median waiting time (WT) to surgery being 209 days. According to statistics, the number of TKRs increased to 9033 and median WT decreased to 149 in 2006 2. In 2005, the Finnish Social Insurance Institution paid EUR 85 million in drug reimbursements to 990,637 patients with musculoskeletal diseases, with the mean cost being EUR 86 per year 3.

Patients with OA experience increasing pain and progressive loss of physical function, walking, and stair climbing 4-8. Earlier studies have established that arthritis causes difficulties in participating in daily activities, which in turn affect quality of life 9-12. Physiotherapy and pharmacological treatment (paracetamol and nonsteroidal anti-inflammatory drugs) are used to reduce the symptoms, namely joint pain, stiffness, and swelling. But, if OA pain is not otherwise manageable, patients' ability to manage everyday tasks is essentially compromised because of OA, or if there is significant restriction of motion, or a joint malposition, a TKR should be performed 1.

In many Western countries, WTs for TKR are long 13. In discussion on health policy in Finland, long WTs are regarded as a major problem for several reasons. They are claimed to violate the principle of equal access to treatment, which is a key performance indicator of health care in Finland. They are also claimed to result in extra suffering, extra cost to the patients and the society, and poorer treatment outcomes. Nevertheless, it is unclear to what extent these claims are true as there is little solid evidence to substantiate them.

To our knowledge, no prior studies have looked at the effect of WT on health-related quality of life (HRQoL) and the costs of disease-specific medication (DSM) among knee replacement patients in a randomized study design. The aim of this prospective randomized study was to identify the effects of WT on health and quality-of-life outcomes, and the use and costs of disease-specific medication among two different patient groups: a short WT group (≤3 months) and a nonfixed WT group (>3 months). The question of whether the length of WT for TKR affects the costs of medication and health, and quality-of-life outcomes is a contested issue.

Methods Data Collection

Between August 2002 and November 2003, 555 TKR patients in three Finnish hospitals were invited to participate in the study: two of the hospitals (the Surgical Hospital and Jorvi Hospital) are part of the Helsinki University Central Hospital, while the third is the Coxa Hospital for Joint Replacement. Patients were recruited into the study through contact with orthopedic and practice staff.

The key inclusion criteria were the need for a primary TKR due to OA of the knee joint as evaluated by the hospital surgeon, the patient was adult (aged 16 or older) and placed on the waiting list in a research hospital, and the patient was willing and mentally able to participate in the study. The key exclusion criteria were patients with rheumatoid arthritis, fractures, and congenital hemophilia or congenital deformities.

Randomization

After being placed on the hospital waiting list, the patients were randomly assigned to one of two groups: 1) a short waiting time (hereafter SWT) with a maximum 3 months wait; or 2) a nonfixed waiting time (hereafter NFWT) with surgery performed according to the hospital's routine procedure, with the waiting period measured from the date the patient was added to the waiting list to the date of admission for surgery. The number of patients placed on the waiting list varied from 1 month to another, being specific to each hospital. Therefore, no advance estimate could be made of the number of patients to be placed on the list.

The patients were recruited into the study in three (in one hospital) or four recruitment periods (in two hospitals), each period lasting 3 months to avoid the WT for the SWT group exceeding 3 months. Patients randomized into the SWT group were operated within 2 weeks after the end of each recruitment period, and only half of the hospital's 1 month surgical capacity could be allocated to the SWT group, so the number of SWTs was restricted and determined specifically for each hospital. The size of the NFWT group was not restricted to ensure that all eligible patients placed on the waiting list had an opportunity to be recruited into the study. The two patient groups therefore differed in size.

Computer-generated randomization sequences were produced by the National Research and Development Centre for Welfare and Health, and supplied to the hospitals using consecutively numbered and sealed opaque envelopes. The patient's named nurse assigned participants to their groups after the decision for surgery had been made. The randomization envelopes contained information on whether the patient belonged to the SWT or NFWT group. Surgeons were blinded to patient allocation. For ethical reasons, double-blinding was not possible.

The patients completed a self-administered questionnaire when placed on the waiting list, at admission, and at 3 and 12 months postoperatively. The questionnaires were either distributed to the patients at the hospital or in some cases mailed to patients, as happened with one hospital for the third and fourth questionnaires. All questionnaires were returned by prepaid post. Common guidelines for administering the questionnaires were provided in each hospital. The patients completed a sociodemographic form, reported their medication and comorbidities as diagnosed by a medical doctor, and completed the disease-specific modified Knee Society Clinical Rating System (KS) and a separate questionnaire for HRQoL. Each patient provided informed consent. The study was approved by the Helsinki University Central Hospital Surgery Ethics Committee.

Measurement Instruments

HRQoL was measured by the generic 15D instrument. The 15D is composed of 15 dimensions: moving, vision, hearing, breathing, sleeping, eating, speech, eliminating, vitality, usual activities, mental function, discomfort and symptoms, depression, distress, and sexual activity. Each dimension has five ordinal levels to choose from. The 15D can be used as a profile measure or to give a single index score by means of population-based preference weights. The index score (15D score) ranges from 0 (dead) to 1 (completely healthy) 14. Completing the 15D questionnaire takes 5 to 10 min, and it describes the HRQoL of the respondent at present. A difference of >|0.03| in the 15D score is clinically important in the sense that on average people can feel the difference 15. The 15D was chosen for three main reasons: 1) it has been used successfully in earlier studies dealing with knee replacement and facilitates thus a comparison to the presurgery scores in these studies; 2) earlier research has shown that in most of the important properties (reliability, content validity, sensitivity in terms of discriminatory power, and responsiveness to change), the 15D compares at least equally with other similar instruments that produce a valuation-based single index number 16, 17; and 3) recent research has since confirmed that especially in the rehabilitation of musculoskeletal disorders, the 15D was at least as responsive as the SF-6D, if not better. These two instruments were anyway the most responsive ones 18.

By using the mean 15D scores from each measurement point and assuming a linear change in the scores between the measurement points, we also estimated the possible gain in quality-adjusted life-years (QALY gain) for both groups within the observation period.

The disease-specific, self-reported modified KS 19 was used to measure knee pain and physical function, because it has been found to be responsive to change 20. Pain score (0–50 points) and function score (0–100 points) are presented separately; clinical dimensions (range of motion, stability, flexion contracture, extension lag, and alignment) could not be measured in a patient's self-administered questionnaire, and thus the total score could not be derived.

The use of DSM during the week preceding every measurement point was measured based on self-report. The unit costs of medication per tablet were obtained from CD-Pharmacy 21. The costs of medication during the waiting period were calculated as a product of the weeks spent on the waiting list and the medication costs per week. The calculations for medication costs were made at each of the four measurement points.

Statistical Analysis

The sample size estimate was based on the primary outcome variable 15D. A total subgroup of 177 patients would provide the 80% power (two-tailed α error 5%) to detect clinically important differences >|0.03| in the 15D score between the randomized groups.

Primary analyses were conducted with an intention-to-treat (ITT) principle 22, so that patients were followed in the groups to which they were randomly allocated. When comparing the mean scores of 15D, knee pain, function, and costs of DSM at the various points of follow-up between SWT and NFWT, a general linear model (GLM) for repeated measures as tests of between-subject effects was used. P < 0.05 was considered statistically significant. Also, the mean differences, SDs, and confidence intervals (CIs) of 95% of the variables 15D, knee pain and function, WT, and costs of DSM were presented for the estimated effect.

Baseline characteristics of the randomized groups and the patients who were lost to follow-up were compared using either the independent sample t test or chi-square test, depending on whether the variable was on a continuous or nominal scale.

Missing values on the 15D were replaced if a minimum of 80% of dimensions had been completed, using a regression model with the patient's responses for other dimensions, age, and sex as explanatory variables 14. Data analyses were performed using SPSS for Windows v.14 and v.16 (Chicago, IL).

Results

Of the 555 eligible patients invited to participate in the study, 117 patients (90 women) with a mean age of 71 refused to participate and were excluded. Thus, 438 patients after providing a signed informed consent were randomly allocated to either the SWT (n = 172) or NFWT (n = 266) group. Of these patients, 13 did not return the questionnaire at baseline, although they had signed informed consent. During the WT, 95 patients were lost to follow-up, 3 exited the queue, 8 had severe comorbidities, 4 were operated on in a private hospital, 29 had canceled operations, 4 died while waiting, and 45 did not return the questionnaire at admission. Primary ITT analyses are based on 330 (77%) patients (237 women) with a mean (±SD) age of 68 (±9.9) years, of which 132 were in the SWT and 198 in the NFWT group, and the final GLM repeated analyses are based on 289 patients, who completed all four questionnaires. Of them, 119 were in the SWT and 170 in the NFWT (Fig. 1).

Patients flow through the trial.

Baseline Characteristics

The baseline characteristics of the groups were similar (Table 1). The mean (±SD) 15D score in the SWT group was 0.772 (±0.108) and 0.779 (±0.119) in the NFWT group; the difference was not statistically significant or clinically important (P = 0.539). The mean (±SD) pain score was 19.9 (±11.8) and 20.9 (±12.4) in the SWT and NFWT groups, respectively (P = 0.355). The percentage of patients receiving DSM was 87% (n = 118) in the SWT and 88% (n = 181) in the NFWT group. The mean weekly medication costs were €5.7 and €5.9 in the SWT and NFWT groups, respectively. The difference was not statistically significant (Table 1).

Table 1. Characteristics at baseline in patients waiting for total knee replacement and those lost to follow-up Characteristics SWT NFWT Lost to follow-up P value† P value‡ n = 132 n = 198 n = 95 Age, mean ± SD 67 ± 9.5 68 ± 8.9 69 ± 9.3 0.208 0.619 Sex (female), (n [%]) 98 (70) 139 (69) 64 (66) 0.308 0.294 Living alone (n [%]) 50 (32) 65 (37) 33 (38) 0.180 0.900 Professional education (n [%]) 31 (23) 38 (19) 11 (13) 0.218 0.168 Employment status Retired (n [%]) 107 (81) 169 (85) 71 (83) 0.050 0.834 Comorbidity (n [%]) 102 (77) 154 (78) 70 (81) 0.509 0.566 Medication to arthritis (n [%]) 115 (87) 175 (88) 71 (82) 0.428 0.208 Cost of medication €/week 5.72 ± 7.2 5.91 ± 6.1 4.89 ± 6.2 0.803 0.456 Body mass index (kg/m2) 29.4 ± 7.0 29.4 ± 4.4 28.55 ± 7.1 0.121 0.43 Pain score, mean ± SD§ 19.9 ± 11.8 20.92 ± 12.4 19.96 ± 11.0 0.355 0.030* Function score, mean ± SD|| 48.42 ± 22.1 46.95 ± 23.3 52.38 ± 22.6 0.567 0.171 15D score, mean ± SD¶ 0.772 ± 0.108 0.779 ± 0.119 0.779 ± 0.11 0.539 0.835 * P < 0.05. † Differences between the randomized groups. ‡ Differences between the patients who completed the questionnaire and those lost to follow-up. § The scale 0–50, worst to best. || The scale 0–100, worst to best. ¶ The scale 0–1, worst to best.

A comparison between patients who completed the questionnaire and those who were lost to follow-up showed a statistically significant difference in pain score (P = 0.030) (Table 1).

Outcomes and WT

At admission, there was a statistically significant difference in mean WT between the groups: 95 (±81) days in the SWT and 239 (±135) days in the NFWT group (F value 100.6; 95% CI 116, 99–172, 22; P < 0.001). There was a statistically significant difference between the groups in the weekly costs of DMS, at €5.33 and €3.57, respectively (F value 4.85; 95%CI 3.54–5.10; P = 0.029). There were no differences between the randomized groups in the mean 15D, pain, and function score, and in the costs of DSM during the WT (Table 2).

Table 2. Outcomes at admission in total knee replacement patients, intention-to-treat analyses Outcome SWT NFWT 95% Confidence interval for mean F value P value Mean, (±SD) (n = 133–136)* Mean, (± SD) (n = 194–203)* 15D score 0.768 (0.15) 0.779 (0.10) −0.020 to 0.029 0.031 0.744 Pain score 18.93 (11.8) 22.79 (12.40) −1.478 to 3.969 0.308 0.369 Function score 48.42 (22.1) 46.95 (23.31) −6.659 to 3.493 0.737 0.540 Cost of medication €/week 5.33 (7.55) 3.57 (5.38) 3.54 to 5.10 4.85 0.029† Cost of medication €/waiting time (WT) 89.19 (199.19) 120.54 (194.82) 80.75 to 131.09 1.50 0.222 WT (days) 94.60 (81.3) 239.2 (135.1) 154.5 to 188.09 100.60 0.000‡ * Number of observations varies because of missing values. † P < 0.05, ‡ P < 0.001. Outcomes after TKR

The use and costs of medication had decreased in both groups at 3 months and 1 year after the TKR. In the SWT group, the weekly costs at 3 months were €3.14 and in the NFWT group €3.10, and after 1 year they were €1.74 and €2.96, respectively (Table 3). In repeated measures analyses, the differences were statistically significant between measurement points (F = 13.17; P < 0.001), but not between the randomized groups (F = 1.17; P = 0.317) (Table 4).

Table 3. Outcomes after total knee replacement, intention-to-treat analyses Outcome Three months postoperatively 95% Confidence interval (CI) for mean P value One year postoperatively 95% CI for mean P value SWT NFWT SWT NFWT 15D score, mean 0.813 0.837 −0.002 to 0.05 0.06 0.813 0.852 0.01 to 0.07 0.012* ±SD 0.12 0.11 0.14 0.1 Pain score, mean 32.7 34.07 −1.55 to 4.85 0.311 36.27 36.95 −3.11 to 4.47 0.724 ±SD 13.03 13.49 13.15 12.83 Function score, mean 62.78 63.86 −7.12 to 4.96 0.725 73.5 74.63 −4.71 to 6.98 0.703 ±SD 25.58 25.22 23.32 22.28 Costs of medication, mean (€/week) 3.14 3.1 −1.48 to 1.20 0.838 1.74 2.96 −0.36 to 2.25 0.142 ±SD 5.88 7.29 3.96 4.07 Table 4. Waiting time effect between the randomized groups Outcome Source F value P value 15D Time 38.746 0.000† Time* 15D 1.66 0.177* Pain score Time 159.686 0.000† Time* pain score 0.645 0.563* Function score Time 118.47 0.000† Time* function score 1.102 0.346* Costs of medication Time 13.172 0.000† Time* costs of medication 1.17 0.317* * Difference between randomized groups. † Difference between the measurement points. General linear model, repeated measures, test of within-subjects effect.

At 3 months, the mean 15D score was 0.813 in the SWT and 0.837 in the NFWT group, and at 1 year they were 0.813 and 0.852, respectively. The difference between the randomized groups was statistically significant (P = 0.012) after 1 year postoperatively. The mean pain scores were at 3 months 32.7 and 34.1, respectively, and at 1 year they were 36.3 and 36.9, respectively. The mean function scores were at 3 months 62.78 and 63.8 in the SWT and NFWT respectively, and at 1 year 73.5 and 74.6, respectively. The differences between the randomized groups were not statistically significant (Table 3). There were statistically significant differences between the measurement points (Table 4).

There were improvements from admission to 12 months postoperatively in the mean 15D score, and pain and function scores, and a decrease in the cost of medication in both groups. Assuming that the final mean HRQoL score in the SWT group would carry forward until the final measurement point in the NFWT group, the latter group would gain 0.033 QALYs more than the SWT group during the whole observation period of almost 700 days (the sum of two areas between the curves in Fig. 2), even if the SWT group would gain 0.012 QALYs more in the short run by reaching a better level of HRQoL earlier (the first area between the curves in Fig. 2), although those in the NFWT had better quality of life postoperatively.

The quality-adjusted life-year gain (areas between the curves).

Interpretation

Scientific evidence on the relationship between WT and TKR outcomes is inconsistent, while the absence of randomized trials has prevented an assessment of whether longer waiting is somehow related to health and quality-of-life outcomes. To our knowledge, the present study is the first to assess the use and costs of DSM in TKR patients randomly allocated to SWT and NFWT groups. The study also analyzed whether the length of the WT was related to HRQoL, knee pain, and function as measured by the 15D and modified KS, respectively.

The main finding was that overall, there was no difference in HRQoL between the WT groups in different measurement points. Nevertheless, those in the NFWT group reached a higher level of HRQoL postoperatively and consequently gained 0.033 QALYs more than the SWT group during the whole observation period of almost 700 days. Those in the SWT group had a worse pain score at baseline, which reflected in an increased use of DSM during the waiting period. The weekly costs were almost identical in both groups at each of the three measurement points. The same applies to the mean pain and function scores, which reached their highest values 12 months postoperatively.

In this study, the patients' HRQoL at baseline was deteriorated. In fact, it seems to be worse than in two earlier studies, where the mean scores before operation measured by the 15D were 0.83 23 and 0.81 24 compared to our 0.77 to 0.78. The patients had pain and difficulties in functioning. Nevertheless, in an earlier study among these patients, it was found that they used very little health and social services during the WT with no statistically significant difference between the randomized groups 25. The length of WT alone did not affect HRQoL or pain and function scores, which is in line with the findings of some earlier studies 7, 8, 19. TKR has been found to be effective 24, 26, 27, and this study also showed a significant improvement in HRQoL, and in pain and function 3 and 12 months postoperatively in both randomized groups.

To our knowledge, no studies have so far been published on the effect of WT on the use and costs of DSM after the TKR. Earlier prospective studies on the costs and outcomes of the WT for TKR have found that waiting more than 6 months was associated with higher total costs and deterioration in physical function while waiting 28, 29. These cost results are not comparable to ours, as they estimated all costs during the waiting period. Nunez et al. (2007) findings on HRQoL and weekly medication costs in patients with OA on a waiting list for TKR were in line with ours.

Strengths and Limitations

The strengths of this study were that the patients awaiting THR were prospectively followed from the time of being placed on the waiting list to admission—with WTs recorded precisely—and further for a year of follow-up postoperatively, providing evidence of the effect of WT on pre- and postoperative health status. Further, the patients were randomly assigned to either the SWT or NFWT group, the randomization was successfully completed, and the groups did not differ from each other at baseline. The findings were based on the simultaneous use of patient-reported generic and disease-specific outcome instruments. The results, based on ITT analyses, indicated that there were no statistically significant differences in health outcomes between the randomized groups during the WT. This was also tested with repeated measures analyses. The within-subject test indicated that there was a significant time effect (i.e., the outcomes did change over time), but an insignificant interaction effect (i.e., there was no difference between the groups).

Some limitations pertain to this study. First, the patients who refused to participate in the study were older, and second, a total of 29 patients in the SWT group waited more than 3 months. The reasons were hospitals' limited capacity to carry out TKR within the 3 months WT period or the patient's unwillingness to have TKR within 3 months, or they were too tired to complete the questionnaires. Because of these factors, the differences between the randomized groups may have been underestimated. Nevertheless, the primary analysis was based on the ITT principle to address the question of clinical effectiveness and to avoid the bias associated with a nonrandom loss of participants.

Third, although SF-36 has been widely used in OA area, we chose to use the 15D. SF-36 is basically a profile instrument and does not thus belong to the category of instruments producing a valuation-based single index number. Nevertheless, in 1998, the first algorithm was published to generate a valuation-based single index score, SF-6D score, from the SF-36. Hawthorne et al. 16 showed that the 15D performed at least as well as the SF-6D. Since a new algorithm has been devised for the SF-6D in 2002 30. Recent research with that algorithm indicates that especially in the rehabilitation of musculoskeletal disorder, the 15D was at least as responsive as the SF-6D 18.

Further, we looked at medication cost only, and medication information was obtained only from patient self-reports, while there was no distinction between self-care and prescription medication. For temporary medication, we used mean dosages. A study is ongoing, where the effect of WTs on the total cost of hip and knee replacements is being investigated.

Conclusion

In Finland, the so-called treatment guarantee was introduced in 2005 with a maximum WT of less than 6 months 31. Since then, there has been discussion in health-care policy about the optimal and effective WT in elective surgery. When considering the optimal timing, information is needed on the effect of WT on key parameters. This study showed that the length of the WT, at least as realized in practice in this study, did not result in different health outcomes in the randomized groups in three cross-sectional follow-up measurements from baseline to 1 year postoperatively. Those in the SWT group reached better HRQoL 3 months earlier than those in the NFWT group, but the latter had better HRQoL after the operation and gained more QALYs overall during the study period. To be useful for future policy work, further research is needed to determine the optimal timing.

Source of financial support: This study was financially supported by the Academy of Finland (no. 51871), HUS HUCH Jorvi Hospital, Coxa Hospital for Joint Replacement, Medical Research of Tampere University Hospital, HUS HUCH Surgical Hospital, Orton Orthopaedic Hospital, and Yrjö Jahnsson Foundation.