Osteoarthritis (OA) is the most common form of arthritis and the third-leading cause of life-years lost to disability [1–3]. It affects 10% of the population and is more common in women than in men. By age 65 years, 50% of the population have OA [1], and, as the population ages in demographic terms, the prevalence of OA is expected to rise. The knee is a commonly affected joint, with a prevalence of OA of 40% in people age 65 years or older [2].

Self-management (SM) is considered to be an effective strategy in the treatment of chronic illnesses, including OA [4]. Numerous SM programs have been developed for different illnesses, such as diabetes, hypertension, asthma and arthritis. Various models, both disease-specific and generic, have been employed, including individual [5], group-based [6], post and internet programs [7]. Most arthritis SM programs use lay leaders who deliver a scripted program. Face-to-face interaction with health care professionals is an important component of some programs, especially when medication compliance is considered relevant [4].

A number of studies have examined the effectiveness of SM for people with arthritis [4, 7, 8]. One systematic review of SM interventions for various chronic diseases found a trend towards a small benefit from arthritis programs, the majority being Stanford University's Arthritis Self-Management Program (ASMP) [9] or derivatives of the ASMP. The results were not significant, and publication bias was suspected [7]. A comparison of lay leaders versus health care professional leaders has also been published [7]; however, the study required all leaders, both lay leaders and health care professionals, to deliver the same scripted program, which limited the capacity for the knowledge and skills of the health care professionals to be optimally utilised [10]. Given the design of the study, it was not surprising that no differences were demonstrated. Despite the popularity of layperson-led SM programs for arthritis, and taking into account possible cost advantages, Taylor and Bury suggested that such programs have little or no advantage in terms of improved self-efficacy or the management of chronic illness [11]. Another study, by Hurley et al. [5], demonstrated positive findings. They compared group and individual rehabilitation programs run by physiotherapists in combination with SM and education (ESCAPE-knee pain (Enabling Self-management and Coping with Arthritic Knee Pain through Exercise)) with normal care in 418 participants with knee OA. The authors found that people in the ESCAPE-knee pain groups had better functioning than those in the normal care group.

Although lay leaders have the potential to be role models (as they often have musculoskeletal conditions themselves), health care professionals, with their knowledge and skills, can also have powerful influence as models. A modelling approach has the potential to transmit knowledge and skills to which people may aspire, particularly if the information is perceived to be important and relevant, resulting in behaviour changes that are more likely to be maintained over the long term [12]. This platform for behaviour change is constrained in layperson-led programs because of the limited knowledge of lay leaders.

Self-efficacy is an integral component of SM, and resilience goes hand in hand with self-efficacy. Resilient people tend to have well-developed self-efficacy and, when confronted with an obstacle, will see it as a hurdle to overcome rather than an insurmountable problem. Furthermore, when repeated problems are encountered, they are more likely to persist longer in attempting to overcome them [13]. Those who attempt and succeed will benefit in terms of improved self-efficacy.

Factors influencing self-efficacy, such as problem-solving, pain management, exercise, modelling, social persuasion, weekly goal-setting and cognitive therapy, are interconnected. Pain management is important because often people are hesitant to undertake new activities for fear of pain, regardless of whether pain has previously been experienced with that particular activity. Many people with OA rely on medication for pain relief, but are reluctant to take medication because of possible side effects. Such people prefer to be aware of the pharmacologic and treatment options available and then decide on a course of management. If knowledge about the available options is lacking, however, the treatment choices are more limited, and, importantly, this may have an impact on adherence to treatment [14].

The lack of demonstrated benefits of SM in patients with arthritis highlights the need to develop and test alternative models of SM. In this report, we describe the evaluation of an SM program for people with OA of the knee, the Osteoarthritis of the Knee Self-Management Program (OAK), which is designed to be delivered by health care professionals.

People with OA may initially resist physical activity due to discomfort, fear of pain, or previous advice to avoid exercise [15, 16]. Many believe that exercise will result in bone and cartilage loss and are therefore resistant to exercise in general [16], yet avoidance of activity is known to contribute to disability long term [17]. The OAK program includes general information about the benefits of exercise and specific advice on joint protection during exercise for those with OA of the knee. The program aims to maximize the benefits of physical activity and promote long-term adherence to an exercise regimen by using structured exercise participation that is linked to weekly 'SMART' goals (Specific, Measurable, Attainable, Realistic, Time-bound) [18]. The program offers the added reassurance that health care professionals are on hand to give advice, and the group dynamics offer incentives for participants to comply with and maintain the exercise regimen. Participants' success in meeting goals each week increases self-efficacy [19], which is the strongest and most consistent predictor of physical activity behaviour and its maintenance over the long term [20]. Although exercise features strongly in the OAK program, OAK is not an exercise school. Exercise is only one component of the program, and it is up to each individual to decide how much emphasis is given to exercise from week to week during the program.

Cognitive symptom management strategies are encouraged to help eliminate 'negative' symptoms associated with OA. Such negativity not only affects individual symptom control but also contributes to and exacerbates the symptoms of OA [21]. Guided imagery, relaxation techniques, positive self-talk and problem-solving are taught to participants. These techniques enable participants to understand how such influences contribute to their symptoms and provide them with the skills necessary to prevent the symptoms' becoming an overwhelming negative influence. Health care professionals also use their knowledge to assist participants with problem-solving to overcome hurdles and promote resilience [22].

In this study, we sought to determine whether a disease-specific SM program for people with osteoarthritis of the knee (the OAK program), implemented by health care professionals, would achieve and maintain clinically meaningful improvements in health-related outcomes compared with a control group. This study was approved by the Human Research Ethics Committee at Curtin University of Technology (HR141). Data access and storage were in keeping with National Health and Medical Research Council guidelines [23]. License agreements were obtained for the Short Form 36 version 1 questionnaire (SF-36) and the Western Ontario and McMaster Universities Arthritis Index questionnaire (WOMAC).

The study design adhered to Consolidated Standards of Reporting Trials Statement guidelines and intention to treat principles. This trial was registered with the Australia and New Zealand Clinical Trials Registry no: 12607000080426. The protocol has previously been described in greater detail [24]. Amendments to the trial protocol included analysis to determine participant's response to treatment (responders). This was necessary to meet the requirements suggested in the Osteoarthritis Research International Outcome Measures in Rheumatology (OMERACT-OARSI) guidelines [25]. In addition, the proportion of people attaining minimal clinically important improvements (MCIIs) in WOMAC physical function SF-36 domains, Visual Analogue Scale (VAS) pain (at 8 weeks) and the Timed Up & Go Test (TUG) was determined.

Study design

We conducted a two-group, randomised (1:1 ratio), controlled, repeated-measures study to examine group differences regarding changes over time. Convenience sampling was employed. The research sample was selected from among people who were referred to the program. Suitable candidates were invited to enrol in the OAK program. Those who agreed to participate and provided their written informed consent were randomised either to an OAK group (immediate start) or to a control group (delayed start). For ethical reasons, those participants who were randomised to the control group were offered the opportunity to enrol in the OAK program at the conclusion of the 6-month study. Independently of the study, all participants were allowed to continue standard medical management for knee OA. Figure 1 shows the design of the study and the time points at which the outcome measures were recorded. All assessments were performed at Arthritis Western Australia.

Table 3 shows the number, characteristics and distribution of all subjects. The male-to-female ratio was not significantly different between groups (Pearson's χ² = 2.311_(1,146), P = 0.182). Sixty-eight participants from each group were assessed at 6-month follow-up examinations. All participants in the intervention group (OAK group) included in the analyses attended at least four of the six SM sessions. The mean (median) attendance in the OAK group was 5.77 (6) sessions. The reasons cited for withdrawal were overseas relocation; work, family and time commitments; and not having been randomised to the OAK group. Participants from the highest socioeconomic group were overrepresented, and approximately 90% had coexisting disease (Table 3).

In this randomised controlled trial, we have demonstrated that participants in a SM program designed specifically for people with OA of the knee and delivered by health care professionals experienced improvements in a number of health domains that people with OAK have identified as important problems associated with their condition [47].

SM aims to motivate people to undertake the changes in behaviour necessary to improve their condition. The priorities of people with OA knee have been identified as problems with pain and activities of daily living, and their preference is to actively manage their condition [14, 47]. The OAK program was designed as a community-based SM education program that aims to improve pain, function and quality of life and to empower people to address these preferences with the support of health care professionals who have expertise in this area. The OAK program incorporates education with an emphasis on OA-related information and the benefits of exercise within SM constructs to promote improved self-efficacy and changes in behaviour. Utilising the knowledge and skills of health care professionals is a chief component of the OAK program because knowledge is an important part of self-efficacy in that no amount of confidence will produce success unless the required knowledge and skills are present [48].

The mechanisms involved in successful SM are not well-understood. The highly structured nature of the intervention may be important, and other nonspecific mechanisms such as group dynamics may be contributory. Nevertheless, there appears to be consensus that the efficacy is likely to be due at least in part to increased adherence to medications [4]. This is a positive effect, especially in people with chronic diseases, for whom compliance (adherence and persistence) with all measures, including pharmacologic ones, are important in optimal management. It should be noted that in the OAK program, educational material concerning pharmacologic therapy and pain relief are included in the syllabus.

The WOMAC and SF-36 questionnaires are both tools that can demonstrate improvements in pain and in overall health status [49]; in people with OA, however, the WOMAC questionnaire is more sensitive than the SF-36 to changes in pain and physical function [40]. Improvements in pain scores demonstrated on the VAS (weeks 1 to 8) were also reflected on the WOMAC and SF-36 questionnaires in the OAK group compared with the control group, and, moreover, they were maintained to 6 months.

Similarly, in the OAK group, a significantly greater proportion of responders were observed at 8 weeks compared with the control group, though by 6 months this proportion did not differ significantly between groups. Reviews of SM show that there is a decline in the extent of improvements by 6 months. This decline is commonly observed in such studies and highlights the limited duration of the benefit that accrues due to SM in the management of OA of the knee, which, of course, requires long-term care [8, 11, 50]. These and other observations attest to the need for more research in this area, especially studies which investigate the differences in cost between SM programs that utilize health care professionals and layperson-led SM programs.

Determining the value that patients place on improvements in pain can be difficult. In studies in which patients were treated with nonsteroidal anti-inflammatory drugs (rofecoxib or ibuprofen), patients have perceived improvements of 9% to 10% in WOMAC scores as beneficial for OA of the knee [51]. The OAK group demonstrated better improvements in WOMAC pain of 23% between pre- and postintervention and of 13.7% from preintervention to 6 months. By contrast, at the same time points, the control group had improvements of 2.3% and 7% in WOMAC pain.

One limitation of the OAK/control study is that it compared a treatment program with a no-treatment control group. Therefore, the only blinding that could be maintained was assessor blinding, an important consequence of which is the risk of reporting, attrition and other types of bias. In addition, self-reported pain may be affected by bias, as patients are keen to 'do well' and to please health care providers by reporting an improvement when there may not have been one. Moreover, the perception of the efficacy of the treatment by the health care providers may influence how the patients perceive their pain and thus may result in an improved pain rating [52], suggesting that the bias related to no-treatment control groups is generally underestimated [53].

As with pain, there were significant improvements in quality of life and function in the OAK group compared to the control group, with improvements seen in WOMAC and SF-36 scores maintained to 6 months. Physical improvements were also maintained at 6 months compared with the control group. Self-reported functional outcome measures tend to be influenced by pain, so it is important to have functional as well as self-reported outcome measures because the combination allows a more realistic appraisal of functional ability than self-reported outcome alone [54].

The control group also demonstrated improvements in many outcomes. It is difficult to explain these improvements other than patient-provider interactions at assessments. Patients in untreated control groups may interact with health care providers. Hróbjartsson and Gøtzsche [53] suggested that the possibility of patient-provider interactions could have clinically useful effects. For example, there may have been a 'Hawthorne' effect, which is described as the awareness of being involved in a trial with resulting altered (that is, improved) behaviour or performance [55]. There was also an unexpected improvement in the number of responders in the control group from 8 weeks to 6 months. Using last value carried forward (LVCF) to replace missing data assumes that the participant's responses would have been constant from the beginning to the end of the study, and this assumption can result in the false conclusion that a difference exists when in fact there is none [56]. Using the LVCF creates the impression that the participant is in a state of equilibrium, neither better nor worse, an outcome that could be perceived as beneficial in terms of disease progression in OA because OA is a condition that generally deteriorates over time. This has the potential to bias the estimates of treatment effect [57].

Strategies to retain the study population included telephone and written notification of follow-up assessments, offers to reschedule missed assessment appointments and telephone contact to encourage rescheduling of assessments. Attendance at assessments at week 8 was 96% in the OAK group and 85% in the control group. The difference may be explained by some participants in the control group assuming that because the control period was 6 months, their participation was not required until that point. At 6 months, participation was relatively unchanged, with the OAK group's attendance being 93% and the control group's being 85%. All nonattending participants received posted self-report questionnaires with postage-paid, self-addressed envelopes enclosed. Not all participants returned the questionnaires. Although the posted questionnaires captured self-reported outcomes, they did not capture physical assessments, and other missing values were recorded with the LVCF, which is consistent with an intention-to-treat analysis.

Within-group improvements were evident on the WOMAC (Stiffness and Total scores) and the SF-36 (Physical Function, Role Physical, General Health and Vitality). The significant improvements seen in hamstring strength, but not quadriceps strength, are difficult to explain. The OAK program is not an exercise program, and, although participants are encouraged to exercise, the exercises are delivered within a SM format that requires individuals to adopt an exercise regimen that best meets their needs.

The highest socioeconomic group was overrepresented in this study. It is possible that the study results might overstate their likely impact on the wider community, because there is the potential for people with higher education levels to have better outcomes. Arthritis Western Australia is located in a middle socioeconomic area. Previous attempts to recruit from lower socioeconomic areas had limited success. Strategies for outer metropolitan and rural clinics were discussed and may be pertinent for future studies. Another possible limitation of this study is the self-initiated enrolment, which might produce potential bias because those people who volunteer may already be positively predisposed to SM [11, 50].

These results reflect the improvements seen in a previously reported quality assurance study in which the OAK program was tested [26]. The use of a more rigorous study design further strengthens the earlier findings. The combined information should prove useful for planning future models of SM in arthritis care. Although the use of health care professionals as facilitators will add to the cost of such care, there is only weak evidence to support SM programs that use lay leaders. Cost analysis was not within the scope of this study. Future research comparing the OAK program with a lay leader SM program should be undertaken to determine the most effective model.

In participants with OA of the knee, statistically significant improvements in pain, quality of life and function were observed in the group randomised to an OAK intervention program delivered by health care professionals compared with those randomised to a control group. The number of participants achieving MCIIs and responder criteria at 8 weeks and 6 months in the OAK group compared with the control group adds strength to these findings.

Future research is recommended to compare and estimate the cost of the OAK program in comparison to other SM models, in particular the ASMP. A comparison with ASMP would help to determine whether a policy shift between lay leaders and health care professionals is justified. Comparison of the OAK program and an exercise-alone program may provide useful information regarding the improvements noted in the OAK group. Such studies should incorporate longer follow-up assessments.