The questionnaire consisted of a case scenario, task instructions, the final list of statements and a consent form. The patient presented in the scenario had a problem with their knee:

● A 57 year old attends the GP for the first time with a knee problem. The problem has worsened over the past few months and the patient has come to ask for help coping with it.

The GPs and patients undertaking the consensus exercise were given a framework for the consultation which was based on the set of principles listed in stage 1. The instructions stated that: i) the treatment algorithm in the 2008 NICE OA Guideline should be followed, and a figure of the NICE target algorithm was included; ii) the consultation was to support the patient’s self-management of OA; iii) the diagnosis would be made clinically; iv) if a diagnosis of OA was made the GP should offer the OA Guidebook and a follow-up appointment with the specially trained healthcare professional. To simplify the consensus task, the scenario focused on one joint, the knee, and the tasks to be considered were those for the assessment and treatment of a problem in the knee rather than any peripheral joint. Participants were asked to consider the statements regarding explanation and planning on the basis that a diagnosis of OA had been made and that the OA Guidebook would be given to the patient.

The level of agreement used to define consensus is often arbitrary 14 . Some studies have “set the bar” for agreement at the level of a simple majority, while others have set the bar higher 14 . We wanted to identify a set number of consultation tasks which could realistically be undertaken in a 10-minute consultation, and not to pre-define an arbitrary level of agreement for a task to be included in the model OA consultation. For this reason an analysis of the number of statements at different levels of agreement for inclusion was undertaken to consider where to “set the bar”.

Setting the bar for consensus at 90% resulted in the identification of 25 consultation tasks to be undertaken during the initial consultation between a GP and a patient presenting with peripheral joint pain. The 25 tasks provide detailed advice on how the following elements of the consultation should be addressed: assessment of chronic joint pain, patient’s ideas and concerns, exclusion of red flags, examination, provision of the diagnosis and written information, promotion of exercise and weight loss, initial pain management, and arranging a follow-up appointment. There was high level of agreement in the GP group to include many of the tasks proposed for the model consultation; the patient group had high levels of agreement for fewer tasks.

The items included in the consensus study for the model OA consultation cover both the assessment of the problem and its treatment if a diagnosis of OA has been made and is to the authors' knowledge the first study using consensus methodology to characterise such a consultation. Two trials 19 20 have previously evaluated the effect of a standardised approach to consulting for OA. One of these 19 included both assessment and treatment, but in both studies the content of the consultation was developed by a group of experts through discussion and reference to published guidelines, and the methodologies for these have not been published. Standard textbooks on clinical methods 21 22 are focused primarily on the examination rather than history taking and do not cover in detail the assessment of peripheral joints in older people. A textbook on the 10-minute clinical assessment 23 includes, in the section on the assessment of knee pain, many of the tasks with a high level of agreement for inclusion in our model OA consultation such as eliciting ideas and concerns, taking a “pain history” and understanding the effect of the problem on mobility and work.

The two tasks given the highest priority, those which all the participants from both groups included, concerned the pharmacological management of pain. However, they did not prioritise psycho-social tasks such as assessing mood and asking about social circumstances, suggesting that both groups favoured a bio-medical approach to the initial consultation rather than a biopsychosocial one. This suggests a discordance between “current thinking” of practising GPs, and patients, and “current best thinking” from research evidence, which suggests that an integrated biopsychosocial approach should be adopted for OA 6 . Possible reasons for this discordance might be; the dominance of the practicalities of achieving something in the 10 minutes of a consultation, the GPs’ lack of awareness of this research, the influence of the prevalent bio-medical approach to osteoarthritis 24 25 , that the relevance of psychosocial management to clinical management of OA has yet to be established or GPs’ perceptions of clinical priorities in a first consultation for such a problem. Concerning the last point, similar patient views supporting a biomedical approach for initial consultations for a problem have been identified previously in other clinical areas; for example Calnan et al 26 . found that patients’ explanations for upper limb disorders were initially biomechanical, with psychosocial explanations only being invoked when these were no longer appropriate. Neither of the two groups in our study prioritised tasks eliciting patient expectations, which is counter to a patient-centred approach propounded in the biopsychosocial approach, or in current notions of the “patient-as-person”, sharing power and responsibility and therapeutic alliance 27 .

The inclusion of patients in the consensus exercise represents a particular strength of this study. The levels of agreement for the statements were lower and more varied in the patient group than the GP group and, by “setting the bar” at the same level for both groups, the GP group contributed more tasks to the model than the patient group. However, the majority of the patient group was in favour of including all the 25 consultation tasks in the model and lowering the bar in the patient group to 80% would only have included two additional tasks. The response in the GP group was low, but this was in line with responses in other studies with GPs as participants 28 and still resulted in 15 GPs completing the consensus exercise, a number which has been shown to be sufficient for such exercises 14 . The participating GPs may not have the same views as GPs as a whole, as they all declared a special interest in musculoskeletal disorders, but it does seem reasonable to use the views of “specialist” GPs when evidence suggests that GPs in general have not fully engaged with the management of OA.

The tasks which the consensus groups prioritised produced a model that had a bio-medical focus and was not fully patient-centred – “eliciting patient expectations” for example was not included - and obtaining this result could be seen as a weakness of a methodology to develop a patient-centred consultation. However, the patient group could have, but did not, prioritise “patient expectations”, suggesting that for this group such an aspect of the consultation was not an essential feature of patient centredness, and our aim was to elicit consensus around current views of patients and professionals on consulting for OA as an important starting point when planning how to implement change.

The consensus exercise was undertaken in the context of the development of an approach to the management of OA to be used in an intervention study investigating how to implement best primary care for OA. The consensus we have obtained will form part of the framework, together with other clinical, scientific, guideline and policy evidence, in shaping the final content of a model for the initial consultation between a GP and a patient presenting with peripheral joint pain for use in the implementation study. These insights from the consensus exercise, into current GP and patient opinion on priorities for such a consultation will be used to inform the development of the training programme in the study.

More generally, the results of this consensus study can inform primary care training for OA management. Although the context of day-to-day practice is different from that used in the consensus exercise, for example the provision of a specially trained healthcare professional to support the self-management of OA is not generally provided in clinical practice currently, many of the tasks which were identified for inclusion in the model OA consultation do not rely on such a service being available and would be relevant to current clinical practice.