European recommendations for the treatment of knee osteoarthritis (KOA), based on expert opinion and evidence based medicine, were published by a EULAR task force in 2000.¹ The task force took into account the 23 types of treatment that were available for knee osteoarthritis and recommended 10 guidelines for the management of the disease (table 1⇓).

Although the ultimate goal of guidelines is to improve the management and thus the outcome of a specific disease,² scientific data are not always translated into practice, resulting in non-standard patient treatment and eventually a poor overall outcome. Thus the compliance of physicians in putting guidelines into practice plays a key role in this process.³ Compliance is influenced by several factors, such as the physician’s background, the recommendations themselves, and the disease for which they are developed, and also the way they are implemented.⁴ Thus, despite the importance of clinical guidelines, their acceptability by physicians to whom they are addressed is not routinely quantified. Further, in the management of KOA, physicians with different backgrounds are involved—general practitioners, rheumatologists, rehabilitators, and orthopaedic surgeons.

We therefore carried out a survey to evaluate the opinion of physicians involved in the management KOA across Europe on the 10 recommendations and the 23 treatment modes.

METHODS

A questionnaire was sent to general practitioners, rheumatologists, rehabilitators, and orthopaedic surgeons in five European countries—Belgium, France, Italy, Spain, and Switzerland—between February and April 2001.

In all countries the questionnaire was sent by the scientific societies of each specialty involved, to all their members. However, in Switzerland, a general mailing was undertaken to all general practitioners, rheumatologists (including rehabilitators), and orthopaedic surgeons, using a national database. In Italy, only rheumatologists were involved.

The two questions addressed in this paper are:

• What is your overall opinion about the usefulness of each of the 23 treatment modes of the European recommendations on the treatment of knee osteoarthritis (using a 100 mm visual analogue scale (VAS), ranging from 0 “I do not recommend its use” to 100 “I recommend its use”)?

• What is your overall opinion on each of the 10 European recommendations on the treatment of knee osteoarthritis (using a 100 mm visual analogue scale ranging from 0 “I totally disagree with this recommendation” to 100 “I totally agree”)?

Completed questionnaires were collected and data were recorded and analysed, using SAS 8.2 software for PC (SAS Institute Inc, Cary, North Carolina, USA), as follows. Descriptive analyses (means, standard deviations, medians, and quartiles) and comparisons between practitioners’ and experts’ opinions were carried out on the whole survey population, and by country and specialty. Comparisons of the mean scores between countries (whatever the specialty), and between specialties (whatever the country), were made using an overall analysis of variance (Fisher’s test). Multivariate descriptive analysis using principal components analysis (PCA) was employed to summarise high dimensional data into fewer dimensions.⁵ PCA aimed to examine relations among all the answers provided by practitioners. This statistical analysis “reduces” data to the pertinent components with respect to variability and identifies correlations between therapeutic modes; these can determine “medical practices” or “general opinions” within the sample of the survey participants. In these newly calculated scales, the country effect and the specialty effect on medical practices and general opinion can be examined.

Missing answers were considered as missing data in the analyses. In the multivariate analysis we used only fully completed questionnaires.

RESULTS

More than 40 000 questionnaires were sent out, to nearly 30 000 general practitioners, 3500 rheumatologists, more than 2000 rehabilitators, and 4900 orthopaedic surgeons in five European countries. As the questionnaires were sent by the scientific society to their members only, not all physicians were represented. This explains the variability in the percentage of practitioners across countries and specialties (table 2⇓).

In all, 4204 completed questionnaires were returned, corresponding to 10.4% of the practitioners contacted. This percentage varied across countries (from 6.5% (Spain) to 18% (Italy), and across specialty (from 7.4% of general practitioners to 22.9% of rheumatologists) (table 2⇑).

Nine hundred and sixty four questionnaires (23% of responses) contained one or more incomplete items and each item was completed in at least 91% of the returned questionnaires. PCA were conducted on the 3240 fully completed questionnaires.

The age distribution of practitioners was almost identical in France, Belgium, and Switzerland (table 2⇑). For France, the age distribution of physicians was based on the national statistics, as an indicator of sample representativeness. In all countries, between one quarter and one fifth of practitioners were in the 30 to 40 year age group. Spain had the highest percentage of practitioners in the 41 to 50 year age group. Data from Italy were insufficient and the percentage of given answers was too small to be analysed. Fifty two per cent of practitioners answering this question (94.6% response rate) were in private practice, 29% in a hospital, and 19% had a mixed mode of practice.

DISCUSSION

We report the results of a European survey of the views of the various specialists involved in the management of knee osteoarthritis. Overall, they showed a high degree of acceptance of the recommendations by the practitioners and little difference between the opinions of the European practitioners and those of the EULAR task force¹ concerning the treatments used in KOA. The greatest difference was that the experts were more likely to recommend IA corticosteroid injections and joint lavage than European practitioners, who were more likely to recommend spa therapy.

The overall percentage of physicians who returned a completed questionnaire may seem quite low (10.4%); however, the absolute number of questionnaires analysed was high (4204), allowing analysis in subgroups, defined by countries or specialties.

We checked the representativeness of the survey by considering France and Belgium as an example. We compared the age of the physicians who replied to the questionnaire with that of the whole population of physicians and found very similar distribution curves, suggesting that the survey sample is representative of the whole population of physicians. In addition, all relevant specialties were contacted (that is, all members of scientific societies involved in the management of osteoarthritis). We also checked the representativeness of the general practitioners involved in the survey in France, as the number might seem very low compared with the total number of GPs. However, we found that our sample was indeed representative of the GPs involved in the management of osteoarthritis.

This survey had some limitations. For example, certain therapeutic modes (lavage,⁶ arthroscopic debridement,⁷ patellar strapping⁸) were evidently not clearly understood by some physicians and should have been better defined in the questionnaire.

Also, the overall responses should be interpreted with caution owing to the large number of GPs participating in the survey. To clarify the results, we carried out a principal components analysis. This revealed associations between treatment modes and identified differences between medical practices within the sample of responses. We found the following. First, orthopaedic surgeons, as opposed to general practitioners, were more likely to recommend “hard line” treatments (so called because they are invasive or potentially toxic). The positive association between such treatment modes suggests that physicians who prefer certain types of treatment are likely to prescribe others of a similar type. Second, we also showed differences between two groups of treatment methods: on the one hand, pharmacological treatment (paracetamol) or non-pharmacological measures (exercise, sticks, education) that are reasonably well validated and have low iatrogenicity were most often prescribed by orthopaedic surgeons and less often by general practitioners; on the other hand, treatments that are not so well validated (such as hormonal therapy) or “alternative medicines” (that is, vitamins, mineral salts) were most often prescribed by GPs and less often by orthopaedic surgeons. Finally, the age of the prescribers may play a role in the choice of treatment⁹ as younger physicians were more likely than their older peers to prescribe opioids, IA corticosteroids, and NSAIDs.

Principal components analysis of opinion about the 10 recommendations showed the same trends: orthopaedic surgeons were more in favour of the recommendations on NSAIDs and paracetamol than GPs and less in favour of those concerning symptomatic slow acting drugs for osteoarthritis (SYSADOA) and non-pharmacological modes including exercise, and vice versa. Not surprisingly, orthopaedic surgeons were more likely than GPs to recommend knee replacement than the use of oral NSAIDs, and vice versa. The age of the practitioners also played a role, as orthopaedic surgeons (older) were more in favour of recommendations on paracetamol and NSAIDs than other specialists (younger) and less in favour of those on non-pharmacological modes.

Overall, such analyses suggest that the management of the patient with knee osteoarthritis depends in part on the age, specialty, and country of origin of the practitioner consulted. It was interesting that rheumatologists were more likely to prescribe local injections (of both corticosteroids and hyaluronic acid) and NSAIDs, that orthopaedic surgeons most commonly recommended surgical procedures, that rehabilitators recommended non-pharmacological measures, and that general practitioners preferred spa therapy and the use of opioids. Different specialists are seeing patients at different stages of the disease, which may have a direct impact on their prescriptions.

In the management of a chronic disease with variable and progressive clinical manifestations, it is normal that treatment measures should be graduated. At any given stage of the disease, an optimised strategy should be determined using an evidence based approach, and this should favour a multidisciplinary form of management of this chronic disease. Further studies are needed to establish predictive factors of response to treatment and decision trees allowing a “tailored” treatment.

Although we found high scores of agreement with the recommendations, we do not know the extent to which practitioners are ready to apply all these recommendations in clinical practice. A recent survey conducted in general practice in France allowed such an evaluation.¹⁰ The major findings were that there was a very low prescription rate of non-pharmacological treatments (ranging from 9% for exercise to 42% for joint sparing advice), a high rate of prescription of paracetamol as first line treatment (90%), a high rate of co-prescription of NSAIDs as first line treatment (43%), and a low degree of prescription of corticosteroid injections in case of flare up (22%). These results reflect different degrees of compliance with the guidelines.

The existing gap between research and practice has been highlighted on several previous occasions, and studies—including those involving osteoarthritis guidelines—have been carried out to define effective interventions in osteoarthritis.^4,11 As far as physicians are concerned, lectures, distribution of recommendations, and audiovisual materials are of little use, while interactive educational meetings are consistently effective at promoting behavioural changes among health professionals.¹² To conclude, our findings suggest that a multidisciplinary approach is optimal in the management of this chronic disease with such a variable course.