Defining osteoarthritis: a moving target.

1063-4584/$ – see front matter 2011 Osteoarthritis doi:10.1016/j.joca.2011.10.008 The challenge of disease definition in osteoarthritis (OA) continues to grow. As described by Pereira et al., in a recent issue of Osteoarthritis and Cartilage (OAC), there are substantial differences in prevalence and incidence estimates when defining OA by radiographic changes alone, as symptomatic OA (requiring a combination of both symptoms and radiographic change) or as selfreported OA, using patient/participant self-report of a previous OA diagnosis1. The authors find, not surprisingly to those in the field, a higher prevalence of radiographic OA in comparison to the symptomatic definitions. What may be more surprising is the close relationship between symptomatic OA, which requires both radiographs and a report of symptoms, and self-reported OA, requiring perhaps only a single question, both having a prevalence w10– 40% lower than that estimated based on radiographic OA. The estimates are most dissimilar for hand OA, which has a remarkably high radiographic prevalence on the order of 50%, but is symptomatic in 15% and reported by only 5% in the pooled estimates. In contrast, the prevalence estimates for hip OA are more comparable, with a radiographic prevalence of 15%, symptomatic prevalence of 6%, and was reported by 7% of subjects (Fig. 1). As discussed by the authors, even these three seemingly straightforward definitions are highly variable across studies. Radiographic OA can be determined based on one of many available scoring systems, including theKellgren–Lawrence scale, assessment of individual radiographic features such as osteophytes or joint space narrowing and their combinations, or quantitative assessment of joint space width. These measures are most commonly applied to the knee, and may be less reliable at other joint sites. Hand OA, in particular, is a challenge to define. Should one distal interphalangeal joint with a Kellgren–Lawrence grade 2 define OA of the hand, or should multiple, bilateral joint involvement be required2? While Pereira et al. have necessarily combined different radiographic definitions to achieve their goal of pooled prevalence estimates, such differences in radiographic definition are likely responsible for the greater variation demonstrated in the forest plots (Pereira et al., Figures 2–5) among radiographic OA prevalence estimates compared to symptomatic or self-report estimates1. The concept of symptomatic OA is attractive to clinicians, as it seems to represent the best of both worlds, by combining information on structural damage with the patient’s symptoms. However,