Activity Increase Despite Arthritis (AÏDA): design of a Phase II randomised controlled trial evaluating an active management booklet for hip and knee osteoarthritis [ISRCTN24554946]

We have developed a similar booklet encouraging increased activity in patients with hip or knee osteoarthritis. The theoretical framework underpinning this new booklet is Leventhal's theory of self regulation, which states that our coping response to illness is governed by our beliefs about the nature of the illness: how well we understand the symptoms (its identity), its chronicity, its controllability, its cause, and the seriousness of its consequences [19]. Educational interventions should emphasise that control is possible and within individuals' capabilities. This model has been extended to include treatment beliefs, so that when considering an intervention patients weigh up the perceived benefit in health gain with the perceived cost in terms of pain, fear and expectation of exacerbating the condition [20]. In addition, social learning theory states that an individual's ability to perform an activity (self-efficacy) is crucial to behaviour change [21]. The evidence based messages for this booklet were obtained from a review of systematic reviews and evidence based management guidelines. These were then converted into patient centred messages and combined in the narrative of a draft booklet. This draft booklet was examined in four focus groups of patients with hip or knee osteoarthritis, which improved the phrasing and emphasis of the patient centred messages in the final booklet [22]. The development of this booklet has been described in more detail elsewhere [23]. This paper reports upon the design of a RCT to test the effectiveness of this new booklet for patients with hip and knee osteoarthritis in terms of influencing illness and treatment beliefs, and to assess the feasibility of conducting a larger definitive RCT to measure change in health status and exercise behaviour [24].

This study has several overlapping aims. In order to test the effectiveness of the new 'Hip and Knee Book' for changing illness and treatment related beliefs a RCT design is employed using outcome measures designed to measure such beliefs. We also want to assess the feasibility of conducting a future definitive RCT and concurrent economic evaluation of the effectiveness and cost-effectiveness of this new booklet on health status and exercise behaviour, so we will use this RCT to pilot recruitment and retention of participants recruited from primary care with hip or knee osteoarthritis, measure changes in health status, levels of physical activity and public agency costs. The results will inform the sample size calculation for a future Phase III RCT. We have developed instruments measuring illness and treatment beliefs from outcome measures designed for different purposes, so we will use this RCT to concurrently test the validity and responsiveness of these adapted instruments.

Participants randomised to the intervention arm will be sent the new educational booklet by mail. This booklet 'The Hip and Knee Book: helping you cope with osteoarthritis' [22] describes the development of osteoarthritis as a dynamic process with the potential for some repair and discourages the more passive concept of 'wear and tear'. Physical activity is emphasised as being beneficial for arthritic joints, as well as for general physical health, and a variety of different types of exercise are promoted. Patients are also encouraged to lose weight if overweight, use simple analgesia and anti-inflammatory gels. Prescribed medication and additional methods to control pain are described. The role of different therapists is discussed as are specialist treatments including arthroplasty. The booklet emphasises that control over their condition is possible and within individuals' capabilities.

Participants randomised to the control arm will be sent a patient information booklet about osteoarthritis produced by the Arthritis Research Campaign (ARC) [25], which can be obtained from the ARC website, and is widely available in rheumatology out-patient clinics. This booklet discusses the mechanism of osteoarthritis, treatment options and self-management. Physical activity and exercise is mentioned, but not emphasised. In particular, it does not address the same exercise-related beliefs and behaviours as the intervention booklet. Clinical care of patients will not be affected or influenced in any other manner by participating in the study.

Randomisation will balance the two groups by site of pain, age, gender and duration of symptoms. As all patients are consented prior to randomisation, this will use an optimal allocation approach proposed by Raab [38] and algorithms developed by Carter [39]. Patients will be recruited from each practice and then randomised as a block. Within the block all potential allocations to two groups will be calculated and their balance statistic calculated. The 1% with the nearest optimal balance will be passed to an independent statistician who will randomly select the allocation for use and randomly allocate groups. The balance calculations for subsequent blocks will incorporate the degree of balance for the chosen allocation to ensure close to optimal balance across groups.

In order to detect a moderate to large effect size of 0.55 in change in illness and treatment beliefs, with 80% power and a significance level of 5%, a sample size of 100 will be needed. We anticipate that 30% of those invited will agree to participate, and 80% of those recruited will return their final questionnaires; so that 400 patients will need to be sent letters inviting them to participate, 120 will be recruited, and 100 will return final questionnaires.

All data will be anonymised and coded so that data collection and statistical analysis is blinded to treatment allocation. The code will only be broken after the main analysis has been completed.

Mean health care costs of service use by patients in the control and intervention groups will be calculated and compared using non-parametric bootstrapping. Health economic analysis will initially involve a cost-consequences analysis of all the costs and outcomes collected [35]. We will calculate a bootstrapped incremental cost utility ratio (ICUR) point estimate of the cost per Quality Adjusted Life Year (QALY) to be gained from this intervention as compared with the control, using EQ-5D as a source of utility weights [33]. We will use bootstrapping to generate a cost-utility acceptability curve to inform policy makers of the probability that the intervention is cost-effective [35, 44]. We will compare our estimate of the cost per QALY with the ceiling of £30,000 used by NICE. Given the relatively small sample size for this Phase II RCT, the findings will give a useful indication of the approximate size of the cost per QALY ratio for this intervention. It will inform the sample size calculation for a sufficiently powered future definitive RCT and concurrent economic evaluation. This trial platform offers an opportunity to compare EQ-5D values with SF-12 values and the other outcome measures of the trial in order to make some assessment of its sensitivity and appropriateness for use in a full trial [45].

Ethical approval was obtained from the North Wales East Research Ethics Committee [REC reference 09/WNo03/5].

From our theoretical framework for developing 'The Hip and Knee Book' we have presumed that people with OA knee or hip have unhelpful beliefs and fears which are responsible for increasing their level of disability. We hypothesise that reading the new booklet, and not the control booklet, will alter their illness and exercise-related beliefs (measured by HKBQ and EAQ-18 respectively), reduce their fear of movement (measured by TSK), change their behaviour to increase their physical activity (measured by IPAQ), which will in turn reduce pain, stiffness and disability (measured by the condition-specific outcome measure WOMAC) and lead to improved general health status (measured by SF-12) and health utility (measured by EQ-5D). Outcomes measuring links higher up this theoretical chain are likely to be more responsive than links further down. Definitive or phase III trials need primary outcomes measuring health status or health utility, which are 'downstream', less responsive and require larger sample sizes. For this smaller phase II trial we have presumed that their will be a larger change in more 'upstream' intermediate outcomes such as illness and treatment-related beliefs. The position of these outcome measures in this theoretical chain can be debated. It may be that before illness and treatment beliefs change participants' fear of movement must be addressed. Alternatively change in participants' views about the prognosis of their arthritic condition, and the benefit of exercise, may lead to a reduction in the fear of movement. Self-efficacy is also important in changing behaviour such as exercise. Although we did not include a separate measure of self-efficacy so as not to over-burden our participants, some of the individual items in the EAQ-18 outcome measure do address this.

According to the Medical Research Council's guidance for developing and evaluating complex interventions [24] the piloting stage includes: testing procedures for delivering the intervention, estimating the likely rates of recruitment and retention of participants, and estimating the effect size of the primary outcome measure in preparation for a sample size calculation. We have also taken the opportunity of testing outcomes measuring beliefs which we hypothesise will have a larger effect size than outcomes measuring behaviours or health status.

The HKBQ has been adapted from the Back Beliefs Questionnaire [BBQ], which is a measure of patients' beliefs about the inevitable consequences of back pain, and was shown to be responsive to change in a previous RCT of 'The Back Book' [46]. It has also been adapted to become the Whiplash Beliefs Questionnaire, which has undergone psychometric testing [47]. A similar instrument for hip and knee OA is not available, so we decided to adapt the BBQ for this purpose. We believe that the risk that such an adapted outcome measure will not be reliable, valid or responsive is less than that if it were to be constructed de novo. Whilst the ideal situation would be to test its psychometric properties for this different population before using it in a trial; we took the pragmatic decision to validate it concurrently with the phase II trial. We have used this approach successfully in other previous studies [47, 48]. The questionnaire is being administered with an intervention, which we hope will alter the outcome measures being tested, so the conditions are not ideal for assessing test-retest reliability. However, we will attempt to obtain some evidence about reliability by collecting data on those participants who claim that there is no change according to an accompanying transition question. Of course patients' responses to these outcome measures may shift independently of any trial intervention, in response to behavioural, cognitive and affective processes necessary to accommodate any chronic illness [49]. So long as these processes influence the outcome measures and transition question in a consistent manner, they should not interfere with the concurrent psychometric testing.