Methods

Systematic search to identify recommended interventions

Clinical guidelines [11] and systematic reviews of interventions for knee OA or knee pain in older adults, rather than original studies, were the target of the search for recommended interventions. Guidelines and reviews, rather than primary studies, were targeted as this is an extensively reviewed area, and it was felt these would be the sources of guidance sought by primary care practitioners. Guidelines and systematic reviews relating to knee OA were included, as they are an appropriate source of guidance for the treatment of knee pain in the older adult: knee OA is the commonest underlying pathology, and there are only a limited number of other pathologies causing knee pain in this age group.

Search strategies

Publications were retrieved by a computerized search of MEDLINE, EMBASE, Cumulative Index to Nursing and Allied Health Literature (CINAHL), Allied and Complementary Medicine Database (AMED) and British Nursing Index (BNI). Potentially relevant reviews or guidelines were identified using combinations of the search terms; knee, pain, OA, management, guidance, recommendation, guideline, review, treatment. Internet guideline or review sites [The Cochrane Library, Centre for Reviews and Dissemination (York), Clinical Evidence, Bandolier, National Institute for Health and Clinical Excellence, Scottish Intercollegiate Guidelines Network, National Electronic Library for Health, National Guideline Clearing House (USA), New Zealand Guidelines Group] as well as three other internet sites [Agency for Healthcare Research and Quality (USA), American Academy of Orthopaedic Surgeons, Canadian Medical Association] were searched for potentially relevant publications. Citations of full-text publications obtained for evaluation were searched for other potentially relevant papers.

Selection of relevant guidelines and reviews

Inclusion criteria

(i) English language (ii) published in the last 10 yrs (search date May 2004) (iii) was a clinical guideline, or included guidance or recommendations (based on a systematic identification of the evidence), for the management of knee OA or knee pain in older adults, or was a systematic review of an intervention(s) for the management of knee OA or knee pain in older adults.

Exclusion criteria

(i) was an original study; or that it was related to: (ii) anterior knee pain or patellofemoral syndrome, (iii) knee pain only in the young (aged < 50 yrs), (iv) the assessment or investigation of knee problems, (v) only in vitro research, (vi) operative or perioperative details of surgical procedures on the knee, (vii) rare conditions that would not normally present in primary care and (viii) complementary medicine other than acupuncture or symptomatic slow-acting drugs for OA. [This last exclusion was applied (a) in order to keep the number of potential interventions to a reasonable size, (b) because the size of the evidence base for complementary therapies other than those included is relatively small for OA, and (c) the availability, accessibility and uptake of many of these therapies in primary care is unclear and very variable]. Interventions could be self-care or professional care.

The inclusion and exclusion criteria were applied by one author (MP), with the two other authors also applying the criteria on the first 200 titles and first 100 abstracts to assess reliability. Full-text publications were obtained where necessary for further evaluation to ensure that the inclusion and exclusion criteria were satisfied.

Data extraction

A standardized methodology and extraction sheet was used to list all interventions discussed in each included publication, and to record whether the intervention was recommended for use. Recommendations were graded from +2 (strongly recommended) through 0 (no definite recommendation, equivocal recommendation) to −2 (recommendation strongly against use). In grading the strength of the recommendation the level of evidence behind the recommendation was not assessed, simply the final recommendation made in the paper.

Compilation of the list of recommended interventions

An intervention was included in the list of recommended interventions if a score of +1 or +2 was recorded from at least one guideline or review, provided there were no conflicting recommendations from other guidelines or reviews. A consensus group (MP, PC and KJ) met to review and clarify all the recommendations and agree inclusion, if there were conflicting recommendations, based on the weight of recommendations provided in the relevant reviews.

Development of the model of care

The model of care assumes that any ‘red-flag’ clinical features and diagnoses for serious causes of knee pain in this age group have been identified, and that the condition being treated is regarded by the clinician as being ‘clinical OA’ of the knee, regardless of whether there is radiographic evidence or not.

Members of the Primary Care Rheumatology (PCR) Society—a group consisting principally of GPs plus some physiotherapists and nurses with an interest in musculoskeletal problems—were identified as appropriate experts to undertake a nominal group consensus exercise [12, 13]. The aim of this exercise was to develop a model of how the interventions identified from the search could be used in practice in primary care for people aged ≥50 yrs with knee pain. The basis of the model of care was that it should be applicable to all degrees of knee pain seen in primary care in older adults, and would cover both self-care and professional care. A stepped-care model would allow the order of use of the interventions to be recommended: interventions in step 1 would be offered initially to all older adults with knee pain before trying interventions in higher steps for adults whose pain or disability persist.

The nominal group consensus exercise was undertaken by delegates to the annual PCR Society conference. During the initial plenary session, delegates were presented with the results of the systematic search and with the methodology for developing a stepped-care model for the treatment of knee pain in primary care (Box 1). Delegates were first asked individually to allocate each intervention to one of the steps in the model of care (round 1). The combined results were then presented in four consecutive workshops (the nominal groups). Discussion focused on interventions for which there had been less than 66% consensus on step allocation. Delegates individually re-allocated interventions to the model of care after the workshops (round 2).

Following round 2, an intervention was assigned to a step in the model of care if there was a 50% or greater consensus for it to be in that step. If this was not achieved then allocations for lower steps were added to those for higher steps until a 50% cumulative consensus was reached. The basis of this was that if, for example, a delegate would use an intervention in step 1, then he or she would logically allow its use in a higher step, if not ‘permitted’ to use it in a lower step.