Article Text


Home based exercise for osteoarthritis
  1. E Roddy1,
  2. W Zhang1,
  3. M Doherty1
  1. 1Academic Rheumatology, Clinical Sciences Building, Nottingham City Hospital, Hucknall Road, Nottingham NG5 1PB, UK
  1. Correspondence to:
    Dr E Roddy
  1. Maxime Dougados2
  1. 2Rhumatologie B, Hôpital Cochin, 27 rue du Fbg Saint-Jacques, 75014 Paris, France

Statistics from

We read with interest the report by Ravaud et al.1 This largest ever study of exercise in osteoarthritis (OA) (2957 people followed up for 24 weeks) examined the effect of patient self assessment and the adjunct of a booklet and videotape to encourage unsupervised exercise. They found significant reductions in pain and disability in all groups including controls, but no between-group differences, thus concluding a “true negative” result.

There is convincing evidence that exercise reduces pain and disability from knee and hip OA.2 The authors suggest that small numbers and lack of attention control in previous studies and the unsupervised nature of their own programme may explain this discrepancy. Yet one UK study not cited by the authors recruited 786 subjects with knee pain (47% with radiographic OA) from three general practices and employed a minimally supervised exercise programme with attention control, demonstrating a clear benefit over 2 years.3

Several caveats of the Ravaud study warrant emphasis:

  • All patients received rofecoxib, with paracetamol as escape analgesia, decreasing baseline pain and reducing the maximum possible effect of exercise. This regimen is contrary to EULAR recommendations,4–6 which advocate paracetamol4–6 followed by topical agents for knee OA,5 then non-steroidal anti-inflammatory drugs/coxibs,4–6 and opioids for refractory pain.5,6

  • Adherence is a major predictor of outcome to exercise3,7 and adherence to the exercise regimen was very low (less than one third of the patients).

  • The rationale for “self assessment tools” is unclear. That patient self assessment itself might influence outcome is interesting. However, self assessment was used to guide adjustment of pharmacotherapy by the physician. Whatever the rationale, an alternative design incorporating assessment and adjustment for “response shift” (alteration over time in patient perception of symptom severity8) would have been preferable.

  • Controls received “usual care” from their rheumatologist. What guidance the rheumatologists received is unclear. However, although only 70% of controls recalled receiving muscle strengthening advice, one would expect rheumatologists to deliver optimum care, including exercise advice. Therefore controls received an expert management package not “attention control”. “Usual care” might equate to attention control in general practice where care is suboptimal9 but is an inappropriate strategy for specialist delivery. Although unstated, “usual care” presumably was given to all groups. Differences in exercise levels between groups are only presumed because exercise undertaken by controls was not assessed.

  • Over 800 rheumatologists recruited only four patients each, making treatment standardisation almost impossible. Previous studies have recruited large numbers from very few centres.3 Presumably the number of recruiting rheumatologists and rofecoxib use reflects industry sponsorship rather than study requirement.

Reassuringly, their study has not diminished the authors’ enthusiasm for exercise for OA. They conclude that for specialist care the addition of a booklet and videotape to encourage exercise is of no benefit. However, we believe the study was inadequately designed to examine effectiveness of unsupervised exercise over 6 months and is a “false negative” trial for exercise.


Author’s reply

We thank Dr Roddy and coauthors for their interest in our report. Obviously, we had some communication problems in the published manuscript because several points have been misunderstood:

  • Clinical efficacy versus clinical relevance

There is no doubt that exercises have shown clinical efficacy (improvement in symptoms) in sophisticated clinical trials, but the observed treatment effect was small and deserved to be evaluated in conditions of daily practice.

  • Potential caveats of our study

    • The protocol clearly mentioned that all the patients justified non-steroidal anti-inflammatory drug (NSAID) treatment. Moreover, it has to be noticed that after the EULAR recommendations were published, two studies have recently suggested that (a) patients do prefer NSAIDs rather than analgesics in crossover design trials1; (b) the effect of paracetamol treatment is questionable in knee osteoarthritis (OA).2

    • We do agree that adherence (compliance) is a major problem in OA. The question remaining is how to improve it. Our main objective was to check whether a simple route of administration of exercises (booklets, video) resulted in the same treatment effect as the sophisticated procedures previously used in clinical trials.

    • The design used in the study was a 2×2 factorial design, which is appropriate when two different, but potentially related, treatment modalities are evaluated. Such a design allows evaluation of each treatment separately and checking as to whether there is an additive or synergistic effect between the two treatments.

    • The fact that our “control” group was usual care is considered a potential caveat by Dr Roddy and as a sign of quality by us. Again, we do recognise the demonstrated symptomatic effect of exercises in OA. However, we considered that this effect has been demonstrated in sophisticated and quite artificial situations. To check whether such treatment is indicated for improvement in pain we conducted this study considering the control group as patients receiving the current usual care.

    • This rationale explains also why we did not conduct a conventional (but artificial) clinical trial in a few centres recruiting patients who are referred to this research centre. Instead we asked physicians who are in charge of patients with painful knee OA to participate.

To conclude, as for a drug development, we do believe that for non-pharmacological treatments, after a treatment effect has been shown in sophisticated clinical trials, a demonstration of its benefit has to be obtained through clinical trials conducted in conditions of daily practice. This was the rationale and the intrinsic quality of the study we reported. We do hope that such methodology will be further adopted for any non-pharmacological treatment.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.