Osteoarthritis [OA] is the most common condition affecting synovial joints [1]. The number of persons affected by OA in the western world will increase because its prevalence increases with age [1]. Patients with knee OA are managed in primary care, and they represent a large group seen by physiotherapists. An overview of systematic reviews covering physiotherapy interventions for patients with osteoarthritis of the knee demonstrates that exercise can reduce pain and improve function in patients with knee OA [2]. It also indicates that low-level laser, transcutaneous electrical nerve stimulation and acupuncture can reduce pain, and that psychoeducation, including patient education and self-management programmes, can improve psychological outcomes. Thus, physiotherapy can improve pain and function and play an important role in the management of patients with knee OA.

Improving the quality of care is a major issue for all health care systems, and measuring performance is essential for the planning and evaluation of quality improvement strategies [3–5]. Measuring performance means comparing actual clinical practice to desired clinical practice. Patient perspectives of care and patient outcomes can also be included in performance measurements [3]. Most performance studies in physiotherapy have described management of low back pain [6–9]. Although OA is a highly prevalent disease, little is known about the performance, including physiotherapy for patients with OA [10].

The aim of this study was to measure physiotherapy performance in patients with knee OA by comparing actual clinical practice to evidence from systematic reviews.

The study was conducted among private physiotherapy practitioners in Norway, who are integrated into primary health care. The National Committees for Research Ethics in Norway approved the protocol for the study.

We received a response from 527 therapists. Among these, 297 had treated one patient with knee OA and had completed the data-collection form. The responders that did not complete the form (n = 230) reported various reasons for not completing, e.g. no patient referred during the study period (n = 109), not working in clinical practice (n = 41) or specialist in other areas such as neurology, child or mental health (n = 46).

When assessing the reliability of the data-collection form we found that the different items had a kappa score that varied from 0.8 to 1.0. For some types of exercise, e.g exercise aimed at increasing strength, coordination and stability, the score was lower.

The most important aim for the treatment, as reported by the therapists, was to reduce pain [92%], followed by increasing muscle strength [85%].

There is evidence of moderate quality that transcutaneous electrical nerve stimulation [TENS], low-level laser therapy and acupuncture reduce pain. Each of these modalities were used by less than 25% of the therapists [Table 2]. Moderate quality evidence suggests that short-wave or pulsed electromagnetic energy has no effect on outcomes for knee OA. This modality was provided by only 10% of physiotherapists.

The physiotherapists applied a median number of four [range 1–10] different treatment modalities for each patient throughout the sessions. Massage, traction/mobilisation and stretching were the next most common modalities after exercise, and were applied in approximately half of patients [Table 2]. There is no evidence from systematic reviews about the effect of these treatments.

There is evidence of moderate quality that psychoeducation, including patient education and self-management programmes improve psychological outcomes, e.g., scales of psychological disability, mental functioning, self-efficacy or depressive symptoms. Sixty eight percent used interventions that were classified as psychoeducation, such as education about OA and coping with the disease. Almost all physiotherapists [90%] provided information and guidance about physical activity, and 76% prescribed a home exercise programme.

The physiotherapists provided advice and information about weight reduction to 59 [58%] of the 102 patients that they considered overweight. On the other hand, almost all patients that the physiotherapist assessed to need more physical activity [n = 101] received advice and support for increasing activity level [n = 92].

To our knowledge, this is the first study of physiotherapy performance for patients with knee OA. The study describes clinical practice in terms of individual patients, as recorded prospectively by therapists during every treatment session. We compared the treatment to findings from an overview of systematic reviews. Quality of care includes many elements. We have studied one important factor that contributes to quality, – the factor of clinical effectiveness.

Almost all therapists in this study used exercise in all treatment sessions. This current practice is desirable, since it is supported by evidence of high quality. Less than 35% of physiotherapists used acupuncture, low-level laser therapy or TENS which have moderate-quality evidence for reducing pain. In addition, physiotherapists used many treatment modalities with low-quality evidence or no evidence from systematic reviews, e.g., traction, massage and stretching.

The physiotherapists provided different types of exercises. Because there is no evidence from systematic reviews to support one specific type or dose [11], we merged different types of exercise into one treatment modality. Clearly, we lost some information about practice by this procedure, but as long as no type of exercise is shown to be more beneficial than another we think this was reasonable. We also categorised different information modalities, but separated simple information about exercise and weight reduction from pscychoeducation and self-management programmes. There is clearly an overlap between these interventions that might introduce information bias or misclassification in this study. The effects of advice and information about exercise and weight reduction provided by physiotherapists to patients with knee OA is unclear, although systematic reviews have demonstrated that exercise and weight reduction improve outcomes in knee OA [2]. However, professional advice and guidance with continued support can encourage people from the general population to be more physically active [12]. Long-term adherence to exercise is required to maintain the benefits of exercise in knee OA, and because long-term adherence requires regular motivation, supervision and monitoring [12], physiotherapists should include such guidance in all treatment sessions. Although many gave advice about physical activity, few physiotherapists [15%] reported having provided this in more than 80% of the sessions.

Only 58% of the patients that the physiotherapists categorized as overweight were given information and advice about weight reduction. The therapists rated subjectively if the patient was overweight. This method might be prone to bias because we do not know how this measure compares with body mass index, which is commonly used to identify overweight. However, clinical judgement and experience might be as important as body mass index for offering patients advice about weight reduction. There are many plausible explanations why many physiotherapists did not focus on weight reduction, e.g., they do not have enough knowledge and/or skills on how to address the problem, the topic is too intimate or they provide advice on physical activity instead. Still we think that physiotherapists might contribute to the positive outcomes of weight reduction by supervision and guidance, perhaps in cooperation with a dietician.

Our findings are comparable to studies of physiotherapy performance for low back pain which demonstrate that adherence to guidelines varies across different treatment modalities [6, 9, 13]. Treatments for which evidence is limited or absent are also frequently used [6, 9]. Interestingly, our study shows that electrotherapy modalities that can reduce pain supported by moderate quality evidence were used by less than 35% of the physiotherapists. In studies of low back pain [6, 14], electrotherapy was more frequently used even though there was no evidence of effect. However, interventions should always be specified to meet the need from individual patients, and the physiotherapists might choose not to use these modalities if the patient had mild pain. If providing electrotherapy, the physiotherapist should choose modalities supported by moderate quality evidence instead of modalities with no evidence, or with evidence of no effect. Still, almost all therapists used exercise, and exercise can also reduce pain. Though, we can not argue that the therapists were providing inadequate care by not using low level laser, TENS or acupuncture.

There are some limitations to this study. The response rate was low, and this might be a threat to the validity of the data of physiotherapy performance because the therapists that responded might have different practice pattern than the study population. We feared that a low response rate might be a problem, and we tried to develop a strategy to get a large and unbiased sample of responses from Norwegian physiotherapists. We invited all private practitioners in Norway to the study. We used finding from a systematic review on how to increase response rate [15]. We contacted the physiotherapists before the study started, the data-collection form was user-friendly with pre-paid return and we had several follow-up contacts. In addition, we enclosed a bar of dark chocolate with a sticker saying "Thank you for contributing to physiotherapy documentation" randomly to half of the physiotherapists.

The physiotherapists who participated were comparable to private practitioners in Norway regarding age [mean age reported by the Norwegian Physiotherapists Association is 48], although a higher proportion of men responded to our study. We have no additional information about the non-responding physiotherapists. Surprisingly many physiotherapists reported that they did not treat a patient with knee OA during the study period. This might also be the case for many of the non-responders.

Other studies of physiotherapy performance in primary care that have used a prospective design have experienced the same lack of participation [6, 16]. When Swinkels et al established a network to collect practice data on a continuous basis in The Netherlands they only collected data from 90 physiotherapists [9].

Another potential source of bias is the self-selection of patients, because the therapists might choose patients that are not representative to patients normally treated in private practice. We asked the physiotherapists to report the management of the first patient with knee OA. The characteristics of the patients in the study are comparable to patients included in 36 trials in a systematic review on physical interventions for patients with OA [17]. The mean age was 65.1 years and the mean baseline pain score was 62.9 on a 100-mm VAS.

We collected data by self-report from the therapist. Self-report of practice might represent a threat to validity because some therapists might report treatments that they do not perform. Some might also adopt new practice pattern because they think it is expected. This might mean that self-reported adherence rates to guidelines could exceed the rate measured by medical records or observation [18]. There might be variation in how the therapists interpret and respond to the data collection from.

We measured performance by comparing practice to findings from systematic reviews. For some interventions we lack evidence because we did not identify any systematic reviews. Evidence of high quality from primary studies not included in systematic reviews might be available for such interventions. This is clearly a limitation to our approach. Secondly, some reviews needed updating. Inclusion of new primary studies might change the estimates of effect and the quality of evidence. Finally, it is crucial to remember that "no evidence from systematic reviews" does not imply "evidence of no effect".

It is difficult to measure physiotherapy performance because physiotherapy practice is complex. Treatment can differ both within and across sessions. Type, dose and frequency vary and the interaction and communication between patient and therapist are important parts of the therapy. In the present study we assessed performance for one measurable part of physiotherapy practice, but we excluded interpersonal communication, structural aspects of care, organisational culture, teamwork and access. These are other important parts of high quality physiotherapy care. Multiple data collection methods might be used to get a more comprehensive picture of actual physiotherapy practice.

Despite clear limitations in our methods, this study contributes to the knowledge about physiotherapy performance in patients with knee OA. We need research to develop valid and reliable methods to measure physiotherapy performance in primary care, as well as research on how to bridge research and clinical practice. Specifically, we should identify effective ways to promote interventions supported by high quality evidence. Finally, in order to be able to measure performance in physiotherapy, we need more research and more systematic reviews on the effects of physiotherapy interventions for patients with knee OA. Because physiotherapists use exercise regularly for patients with knee OA, and there are different opinions about optimal exercise regimen, studies should compare different types, settings, intensities and volumes of exercise. Interventions that are frequently used by physiotherapists without evidence from systematic reviews, e.g., traction, massage and stretching for patients with knee OA, should be tested in rigorous trials and summarised in reviews.

This study provides information about physiotherapy performance in patients with knee OA. Exercise is the most common treatment and this is supported by high quality evidence. Physiotherapists also provide several treatment modalities based on moderate and low quality evidence of benefit, or without evidence from systematic reviews. We need more research to develop and identify the best methods to measure physiotherapy performance in primary care.