Background
The current consensus of expert opinion is that "it is important to separate conceptually the disease process of osteoarthritis and the syndrome of musculoskeletal pain and disability" [
1]. Given that classification is "arguably one of the most central and generic of all our conceptual exercises" [
2] and that "all our activities in public health, in epidemiology, and in clinical practice depend on the way we classify, recognize, and identify diseases" [
3] this issue deserves attention.
The disease of osteoarthritis (OA) is considered to be an active process involving the entire synovial joint with both degenerative and repair processes. It has multiple determinants that differentially affect incidence and progression [
4], and may not comprise a single uniform disease but several (e.g. tibiofemoral vs patellofemoral; isolated knee vs generalised OA).
It has been argued that "from a clinical perspective, the most compelling definition of (the) disease (of osteoarthritis) is one that combines the pathology of disease with pain that occurs with joint use" (termed "symptomatic knee OA") [
5]. Symptomatic knee osteoarthritis (OA) affects an estimated 10–12% of the adult population aged 55 years and over [
6] with an annual rate of radiographic progression of approximately 3–4% [
7,
8]. However, there are grounds to doubt whether this definition is, in fact, the basis for using the diagnostic label of "osteoarthritis" in primary care, where it is one of the most common diagnoses made in older adults [
9]. Current guidelines discourage the routine ordering of x rays to confirm a diagnosis of OA [
10] making it unlikely that the 'pathology' has been verified in many cases. Although plain x-rays may be ordered by primary care clinicians, the decision to do so appears to be determined less by clinical features at the point of presentation as by provider characteristics and other considerations such as patient expectations of a diagnosis or a predetermined management plan [
11,
12]. Knee osteoarthritis in practice may be better characterised as a "preference-sensitive" diagnosis (with respect to clinicians' and patients' preferences) that will include a combination of radiographically verified symptomatic knee OA and non-radiographic knee pain in older adults which is given the same diagnostic label [
13]. The effect of using such a 'mixed' approach to OA classification in practice on the accuracy of prognoses or the effectiveness of management is unclear. However, it cannot be assumed that adopting a more rigorous approach to diagnosis using only the disease-based definition of symptomatic knee OA would be an improvement although much of the current evidence for the effectiveness of interventions has been based on participants defined in this way.
One proposed alternative to disease-based classification is instead to view knee pain in older adults (with the possible exception of severe OA) as a regional pain syndrome [
14] in which psychosocial rather than pathoanatomical features contribute to variance in symptom severity and disability [
1,
15]. This observation alone, however, does not provide direct evidence about exactly how such an approach might be implemented in practice nor how useful this might be. The debate on the relative merits of disease-based and regional pain syndrome-based approaches to classifying and diagnosing knee pain in primary care is also found across a range of other musculoskeletal pains [e.g. [
16‐
18]].
The general aims of the cross-sectional component of this study are to investigate the relative usefulness of disease-based and regional pain syndrome-based approaches to classifying knee pain in older adults and to develop simple assessment tools that are clinically practicable for the primary care setting. We have taken our starting point as knee pain in older adults in the general population to reflect the diagnostic challenge in primary care where the presentation of undifferentiated symptoms is common [
19]. This starting point encompasses a larger proportion of older adults in the general population – approximately 25% of those aged 55 years or over experience knee pain that has lasted four weeks or longer at any given point in time [
6].
Specifically our study will consider the following questions:
• What is the association between symptomatic radiographic knee OA and chronic knee pain? Does this association differ between tibiofemoral joint (TFJ) and patellofemoral joint (PFJ) OA?
• Can simple clinical signs and symptoms accurately identify symptomatic radiographic knee OA in older adults with knee pain?
• How does the distribution of other clinical features (e.g. signs and symptoms indicative of periarticular pathology), coexisting hand OA and non-clinical characteristics (e.g. psychosocial factors) compare between symptomatic radiographic knee OA and other knee pain?
• In what respects do consulters and non-consulters differ in their characteristics at baseline?
• How does a GP diagnosis of "knee osteoarthritis" relate to disease-based and regional pain syndrome-based classification?
It has been argued that deciding what to do about a problem is often of more interest to clinicians and patients than what to call it [
20]. Accurate information on the likely future course will play an important role in the decision-making of both parties. Classifications of knee pain at a single point in time, whether on the basis of a disease or regional pain syndrome approach, need to be qualified by descriptions of their subsequent course over time. Such prospective studies offer the potential of describing intra- and inter-individual changes, investigating mechanisms to explain these changes, and forecasting change or outcome [
21,
22]. The current study has been designed with attention to previously published requirements for reporting longitudinal studies in rheumatology [
23,
24].
The general aims of the longitudinal component of this study are to describe the clinical course and patterns of health care utilisation of knee pain in older adults in the general population, and to develop prognostic indicators of clinical course and predictors of consultation. Specifically, it will address the following questions:
• What proportion of this sample consult their general practitioner for knee pain within the follow-up period? Can this be predicted by clinical and/or non-clinical variables collected at baseline?
• How common is clinical deterioration (in terms of increasing pain/disability severity) in this sample? Can it be predicted?
• What is the relative contribution of disease-based, clinical, and regional pain syndrome-based variables as prognostic markers?
Authors' contributions
All authors participated in the design of the study and drafting the manuscript.