Update: Treatment of osteoarthritis


The treatment of osteoarthritis (OA) is largely symptomatic and unsatisfactory; it is made more complicated by the heterogeneity of the condition and the fact that OA of different joint sites is expressed in varying ways and may respond differently to the many therapeutic options (1). Many people in the community with symptomatic OA never seek help from traditional Western medicine, and the utilization of complementary and alternative therapies by people with OA is high (2). Within traditional medicine, available options range from education and advice, through physical therapy, aids and appliances, local and systemic drug therapy, to joint replacement surgery. This update summarizes 6 recent articles, each of which provides new and important data on a different treatment option for people with OA.

Long-term effects of glucosamine sulphate on osteoarthritis progression: a randomised, placebo-controlled clinical trial (Lancet, 2001) (3)

Glucosamine (sometimes mixed with chondroitin) is an extremely popular over-the-counter medicine, advertised as being of benefit to people with OA. The data available to support these claims has been considered weak (4).

This randomized controlled trial compared the effect of glucosamine sulfate (1,500 mg once per day) with placebo over a 3-year period on progression of OA of the knee. A total of 212 people were recruited: 106 to the glucosamine group and 106 to the placebo group. The inclusion criteria were age >50 years and OA of the knee as defined by the American College of Rheumatology (ACR) classification criteria (5). The exclusion criteria were other rheumatic disease; severe articular inflammation; traumatic knee lesions; body mass index >30; substantial hematologic, renal, hepatic, or metabolic problems; and intraarticular corticosteroid injections to the knee in last 3 months. The primary outcome measure was joint space narrowing as measured on a radiograph, and the secondary outcome measure was change in Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) score. The trial found (using intent-to-treat analysis) a significant difference between the treatment and placebo group for joint space narrowing at the end of 3 years: –0.06 mm (95% confidence interval [95% CI] –0.22, 0.09) in the glucosamine group versus –0.31 mm (95% CI –0.48, –0.13) in the placebo group; difference of 0.24 mm (95% CI 0.01, 0.48; P = 0.043). Change in WOMAC was also significantly different between the groups, with a 9.8% (95% CI –6.2, 25.8%) increase in the placebo group and an 11.7% (95% CI –20.3, –3.2%) decrease in the glucosamine group; difference 21.6% (95% CI 3.5, 39.6; P = 0.02). There was no significant difference in reported adverse events (P = 0.72). Dropout rates were similar in both groups: 36% in glucosamine and 33% in placebo (P = 0.77).

The results of this trial suggest that glucosamine is effective at slowing the progress of OA and reducing the symptoms of OA. The analysis, however, does not link symptoms and joint space narrowing, which is an important problem because the relationship between symptomatic and radiographic progression may not be high (6). The symptomatic benefit could theoretically have affected the radiographic evidence of joint space narrowing, by, for example, allowing patients to stand with less flexion or rotation of the knee for the followup views (7). Commercial bias may also be an issue, and this has been raised by others and us elsewhere (4, 8). Until these issues are clearly addressed, the wholesale endorsement of glucosamine as a “structure-modifying” agent in OA cannot be made. Nevertheless, this trial of glucosamine is a major advance on previous studies, most of which were of much poorer quality, over much shorter periods, and involving far fewer patients (7, 9). The effect on treatment patterns could be significant, as it could lead to wide-scale prophylactic use of glucosamine by patients, with or without the endorsement of health professionals.

Physical therapy is effective for patients with osteoarthritis of the knee: a randomized controlled clinical trial (J Rheumatol, 2001) (10)

Physical therapy is widely recommended for the treatment of OA, particularly knee disease (11). However, the evidence base for its effectiveness is weak (12).

This article reports a pragmatic, randomized controlled trial comparing physical therapy with placebo for the treatment of OA of the knee over an 8-week period. The study recruited 126 patients: 43 were assigned to individual treatment, 40 to group treatment, and 43 to a waiting list control group. The individual exercise regimen was delivered by a physiotherapist and was not standardized across patients. Group therapy involved standardized twice-weekly physiotherapist-supervised exercise classes, and a standardized home exercise program. The waiting list group received no treatment for 8 weeks, after which they were randomized to 1 of the 2 active treatment arms to increase the statistical power of comparison between treatment groups.

Inclusion criteria were age >50 years, knee pain, and radiographic signs of OA. Exclusion criteria were intraarticular cortisone injections, lower limb arthroplasty, unstable cardiac comorbidity affecting ability to exercise, or comorbidity affecting gait. The primary outcome was based on change in WOMAC score (using a 100-mm visual analog scale [VAS]). Analysis was undertaken using intent-to-treat.

The results showed that both treatment groups had significant improvements over the control group for pain, function, and quality of life. Change in WOMAC pain score for the treatment group was 10.6 (95% CI 6.3, 15.0) and for the control group –1.5 (95% CI –5.5, 2.4; P < 0.01). (WOMAC score was reversed in this study, so a higher score was a better outcome.) Change in WOMAC function was 7.7 (95% CI 4.2, 11.2) in the treatment groups versus –0.1 (95% CI –3.9, 3.7) for the control group (P < 0.01). Three people dropped out of the individual exercise group, 2 from the group exercise group, and 2 from the control group. At 8 weeks, those in the control group were randomized to 1 of the 2 active treatment groups, and another assessment was made after 8 weeks comparing the 2 treatment groups. There was no statistically significant difference between the group exercise and individual exercise arm, although the group exercise arm was consistently better than the individual group.

This trial shows that physical therapy can have a significant effect on patient pain and function. It also suggests that a more cost effective, standardized group program may be more clinically effective than personalized physical therapy. The quality of this trial is an improvement on much of the existing evidence (12). However, blinding of allocation was not complete, which has been shown to be a major potential source of bias (13). The trial was relatively short term, so the issue of long-term adherence cannot be addressed; a recent study from our group suggests that lack of concordance with physiotherapy over longer time periods may weaken its value (14). The intervention was also relatively intensive, with people in the group treatment arm receiving twice-weekly sessions, and expected to undertake home exercise. A separate trial of an exercise program over a 12-week period showed that the beneficial effects of an exercise regimen diminished over the 24-week followup period, and concluded that patients require booster or refresher sessions to ensure that they maintain exercise regimens (15). Nevertheless, the use of standardized group exercise programs offers an inexpensive, simple, and perhaps even enjoyable treatment option for patients.

Safety and efficacy of meloxicam in the treatment of osteoarthritis: a 12-week, double-blind, multi-dose, placebo-controlled trial (Arch Intern Med, 2000) (16)

The new cyclooxygenase 2 (COX-2) inhibitor nonsteroidal antiinflammatory drugs (NSAIDs) have been heavily promoted for use in OA, based on safety. These promotions are based on the fact that COX-2 inhibitors should be less likely to cause gastrointestinal (GI) problems; however, evidence has only recently become available, and it is not conclusive.

This randomized controlled trial compared the COX-2 inhibitor meloxicam (at dosages of 3.75, 7.5, or 15 mg/day) with diclofenac (at 100 mg/day) or placebo over a 16-week period in patients with OA of the knee or hip. Patient inclusion criteria were current NSAID use, age >40 years, and radiographic and clinical evidence of OA. Exclusion criteria were intolerance for NSAIDs, analgesics, or antipyretics; aspirin hypersensitivity; abnormal renal, heptic, or hematologic function; a history of bleeding disorder; anticoagulant use; corticosteroid use in the last 2 months; intraarticular injection of hyaluronic acid in the past 3 months; long-term use of GI medication; history of alcohol or narcotic abuse; and upper GI perforations, ulcers, or peptic ulcers in the last 6 months. A total of 774 patients were randomized into the study: 157 to the placebo group; 154 to the meloxicam group receiving 3.75 mg/day; 154 to the meloxicam group receiving 7.5 mg/day; 156 to the meloxicam group receiving 15 mg/day; and 153 to the diclofenac group receiving 50 mg twice daily.

The primary outcome measures were change in WOMAC score and incidence of adverse events. Changes in total WOMAC score were –10.2 for placebo, –14.2 for meloxicam at 3.75 mg/day, –15.3 for meloxicam at 7.5 mg/day, –18.9 for meloxicam at 15 mg/day, and –21.1 for diclofenac. Meloxicam at 7.5 mg/day, meloxicam at 15 mg/day, and diclofenac were found to be significantly better (P < 0.5) at reducing pain than placebo. Premature discontinuation of therapy for adverse events was 3.8% for placebo, 8.4% for meloxicam at 3.75 mg/day, 7.1% for meloxicam at 7.5 mg/day; 8.3% for meloxicam at 15 mg/day, and 8.5% for diclofenac.

This trial was selected as an example of work currently been undertaken on COX-2 inhibitors. These drugs have been replacing older NSAIDs because of the perceived lower GI risks associated. However, this trial does not show any significant difference between diclofenac and meloxicam in terms of GI problems. Furthermore, a recent study has suggested that COX-2 inhibitors may be associated with a significant increased risk of cardiovascular events (17). The study reports the standardized rate for myocardial infarction (MI) as being 0.52% (based on a meta-analysis of placebo arms from existing community-based cardiovascular studies), and this compared with rates for MI of 0.74% with rofecoxib (which was significantly higher than the standardized rate of 0.52%, P = 0.04) and 0.8% with celecoxib (which was significantly higher than the standardized rate, P = 0.02) (17). The methodology of this study has been heavily criticized (18), but nevertheless the findings raise a cautionary flag and need to be examined in subsequent trials. Studies have also suggested that the use of aspirin at the same time as COX-2 inhibitors may increase GI problems, and given that a significant percentage of OA patients will also have cardiovascular risk factors for which they will be using aspirin, this is an important issue (19).

This trial suggests that meloxicam is as effective as diclofenac (the most widely used COX-1 inhibitor for OA patients), but it does not answer the question as to whether NSAIDs are preferable to simple analgesics such as acetaminophen (preferences for pharmaceutical treatment are likely to differ between patients). Furthermore, as in most trials, many of those at highest risk of side effects were excluded. As with any commercially sponsored trial, the issue of reporting bias has to be remembered (20). COX-2 inhibitors are likely to represent a significant part of pharmaceutical treatment for OA over the next few years, so it is important that evidence is available to show in which patients these drugs are most useful.

Intraarticular injection of hyaluronan as treatment for knee osteoarthritis: what is the evidence? (Arthritis Rheum, 2000) (21)

Hyaluronan injections have become a popular symptomatic therapy for people with knee OA over the last few years. Many different products, with different molecular weights and properties, are available, and the published evidence for effectiveness is contradictory.

This randomized controlled trial compared injected hyaluronan sodium (Orthovisc; 2 ml injected once a week for 3 weeks) with injected saline on OA of the knee over a 28-week followup period. A total of 226 people with OA of the knee entered the study: 114 were randomized to the hyaluronan arm and 112 to the saline arm. Inclusion criteria were OA as defined in the ACR classification criteria (5), radiographic OA of Kellgren/Lawrence scale grade II or III, WOMAC pain score >13 in the index knee and <13 in the lateral knee (possible range 5–25), and willingness to stop all analgesics. Exclusion criteria were steroid use within 2 months, hyaluronan injection within 12 months, Kellgren/Lawrence scale grade IV, anticoagulant use, immunosuppressive use, muscle relaxant use, inability to tolerate acetaminophen, clinically significant renal or hepatic disease, allergy to lidocaine, or previous arthroplasty. The primary outcome measure was change in WOMAC score.

The results showed that hyaluronan was significantly (P < 0.05) better than saline at reducing pain as measured on the WOMAC pain score at 7, 11, 15, and 27 weeks (but not at weeks 1, 2, 3, or 21); no overall figures for WOMAC pain scores were presented, only a graph. However for the WOMAC function and WOMAC stiffness scores, hyaluronan was only significantly better (P < 0.05) than placebo at improving symptoms at week 11. Improvements in scores compared with baseline were maintained throughout the trial period in both groups.

Although this trial shows that hyaluronan is more effective than saline, saline alone did result in significant improvements in patient outcome compared with baseline. For example, 40% of those treated with saline compared with 58% of those treated with hyaluronan improved >5 units on the WOMAC pain score. This was a statistically significant difference in favor of hyaluronan (P = 0.04), but demonstrates that saline does have a large effect size.

This trial is one of several large trials that have been published on hyaluronan in the past few years. Although this trial and the one by Altman and Moskowitz (22) show hyaluronan to be effective, the trial by Lohmander et al (23) showed no significant difference between hyaluronan and saline. Overall, the weight of evidence points to hyaluronan injections being a modestly effective treatment for pain associated with OA of the knee in people with moderate OA, and that the effect lasts for several months, but saline injections also produce large effect sizes over the same period of time and have fewer associated side effects. Whether this effect is due to the placebo effect or due to other factors, including debris removal and aspiration, is unknown. In addition, Evanich et al (24) show that hyaluronan is inappropriate for patients with severe OA, and raise the question about what is the optimal severity in which injections should be used.

A randomized controlled study of the Arthritis Self-Management Programme in the UK (Health Educ Res, 2000) (25)

The provision of education has always been highlighted as a problem area within OA (26). It is popular with patients and user organizations, however the amount of work in this area has been limited.

This randomized controlled trial compared a standardized self-management education program with no education over a 4-month period on people with OA. The inclusion criteria were age >18 years, ability to complete questionnaires, and a diagnosis of arthritis. A total of 544 people were randomized either to the education program (311 people) or to a waiting list for the education program (233 people). The primary outcome measures were arthritis self-efficacy and health behaviors. The results showed, at 4 months followup, statistically significant differences in favor of the intervention group for assessment of self efficacy (P < 0.0005), depression (P < 0.0005), anxiety (P = 0.014), communication with health professional (P = 0.001), diet (P = 0.001), fatigue (P = 0.02), and positive feelings (P = 0.005). However, there was no significant effect on VAS pain (P = 0.707), negative feelings (P = 0.429), Health Assessment Questionnaire (HAQ) score (P = 0.351), EQ-5D EuroQol (P = 0.485), or EuroQol VAS (P = 0.736), though figures did not deteriorate. A total of 44 people from the control group and 77 from the intervention group failed to complete the 4-month followup. The blinding of patients and potential for “contamination” of treatment from intervention to control group is an issue with this study, which was undertaken on people of a wide age range with self-reported diagnoses of arthritis, rather than proven OA.

Education is constantly highlighted by patient surveys as being a major unmet need, and the results of this study and others (26, 27) show that education interventions can be inexpensive and effective methods of helping patients deal with living with arthritis. However, one of the major issues with education interventions has been the heterogeneity of interventions provided and the poor quality of the trials. This perhaps reflects the variety of outcomes the interventions are trying to target: self help, psychological coping, etc. This study does not score high on quality criteria. However, the use of standardized methods, such as the one used here, offers the opportunity to overcome these problems, and further investigations need to be undertaken.

Differences between men and women in the rate of use of hip and knee arthroplasty (N Engl J Med, 2000) (28)

Joint replacement is the ultimate, irreversible treatment for large joint OA, and the only intervention likely to produce major benefits for those with advanced disease. However, huge variations in the provision of joint replacements within and between countries have been highlighted as an important issue for health services research. The provision of the best possible treatment for OA patients, who form the majority of those coming for joint replacements, has also been highlighted.

This epidemiologic study examined sex differences in total joint arthroplasty. All 48,218 people older than 55 years in 2 counties in Ontario, Canada were given a postal or telephone survey. The response rate was 72%. The questionnaire asked about existing joint problems. People who highlighted problems were assessed using WOMAC, HAQ, and Medical Outcomes Study Short-Form 36. A sample had their joint assessed by a health professional. The study formed part of the Patient Outcome Research Team study on joint replacement. The results showed that women had a higher prevalence of moderate to severe self-reported arthritis of the hip or knee than men (age-adjusted odds ratio 1.76; P < 0.001), but women were less likely to have undergone arthroplasty (adjusted odds ratio 0.78; P < 0.001). The study showed potential need for arthroplasty was 44.9 per 1,000 among women and 20.8 per 1,000 among men. This was further adjusted for people's willingness to undergo immediate surgery, which gave numbers of 5.3 per 1,000 for women and 1.6 per 1,000 for men. This shows that in the community there may be a pool of both men and women with severe arthritis requiring surgery and willing to undergo surgery, but the female group is 3 times greater than the male group.

Although this study does not represent an advance in treatment, it does present an extremely important issue. If these results are confirmed in other countries, then it will have major implications for the pattern of joint replacement. Research has shown that people with higher levels of disease severity who underwent surgery, although they exhibited great improvement in terms of pain and function, never obtained the outcome levels of those with lower levels of disease severity who underwent surgery (29).


The articles reviewed here highlight advances that are being made in the treatment of OA. However, OA remains a frustrating and difficult condition to treat because no single treatment can cure, or even help, all people with OA, and management strategies must change over time as the condition and the circumstances of the patient change. This means that successful management of OA is as much about patient and physician working together, and health professionals cooperating with one another, as the impact of any individual intervention. This statement is perhaps borne out by the results presented in Table 1, which compares the effect sizes of the 4 treatments that used the WOMAC outcome measure to report results, and shows how similar the effect size is for each treatment. At the same time, the large differences in the outcomes of the placebo groups is striking and perhaps reflects different followup times and control group procedures.

Table 1. Change in Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores*
 Fransen et al (10)Brandt et al (21)Reginster et al (3)Yocum et al (16)
VAS WOMAC at 16 weeksLikert WOMAC at 27 weeksVAS WOMAC at 3 yearsLikert WOMAC at 12 weeks
Exercise (% change)Control (% change)HyaluronanControlGlucosamine (% change)Control (% change)Meloxicam 3.75 mgMeloxicam 7.5 mgMeloxicam 15 mgDiclofenacPlacebo
  • *

    Percentage change from baseline shown in parentheses for Likert WOMAC scores. VAS = visual analog scale.

  • Score was reversed in this study so a positive result signified a better outcome, here the figures have been reversed back to make them comparable to the other studies.

WOMAC pain−10.61.5∼−5∼−4  −2.7−3.4−4.5−4.5−2.2
   (−30%)(−24.5%)  (−24.1%)(−31.2%)(−39.5%)(−38.9%)(−19.3%)
WOMAC function−7.70.1−14.7−9.8  −10.2−10.4−12.6−14.9−7.2
   (−26.8%)(−18%)  (−25.1%)(−27.4%)(−32%)(−36.7%)(−18.1%)
WOMAC stiffness  −1.7−1.1  −1.3−1.5−1.8−1.8−0.7
   (−24.3%)(−16.2%)  (−24.5%)(−30%)(−35.3%)(−35.3%)(−14.3%)
WOMAC total∼>−18.3∼>1.6∼−21.4∼−14.9−11.79.8−14.2−15.3−18.9−21.2−10.2
   (−27.4%)(−19.2%)  (−24.8%)(−28.7%)(−33.9%)(−37%)(−1%)


Bristol is the lead center of the MRC Health Services Research Collaboration.