In the article by Felson et al, which appears elsewhere in this issue of Arthritis Care & Research (1), the authors support current public health recommendations that all adults accumulate at least 30 minutes of moderate physical activity on most days of the week (2). Results from their study of 1,279 individuals in the Framingham Offspring cohort indicate that prior physical activity did not increase the risk of developing knee osteoarthritis (OA) over a 9-year study period. The authors conclude that this finding was not affected by the amount of activity (more miles walked per week), intensity of activity (working up a sweat), or being overweight or obese. An additional purpose of this study was to investigate whether physical activity was chondroprotective. Comparing radiographs taken at baseline and at study completion, the authors reported that joint space width was not related to physical activity or weight and that there was no evidence that physical activity benefits cartilage health.

In addition to the encouraging outcome that moderate recreational physical activity does not increase the risk for developing knee OA in older adults, the study by Felson and colleagues contributes credible data by overcoming a number of methodologic problems that have clouded understanding of the relationship between knee OA and physical activity. This longitudinal study included over 1,000 subjects for an approximately 9-year study period, which was a reasonable sample and time period to answer the study question. Felson and colleagues' definitions of knee OA were precise and comprehensive and included radiographic and symptomatic criteria, joint space width measurement, and determination of disease in tibiofemoral and patellofemoral compartments. The assessment by Felson et al of prior physical activity appeared to be valid for collecting recalled information. They approached the aim of evaluating the true incidence of knee OA by eliminating subjects with evidence of radiographic disease at baseline. Taken together, these methods result in a useful and valid study that supports recommending regular moderate physical activity without undue fear that such activity may increase the risk for knee OA.

As usual, asking an interesting question with a database as rich as the Framingham cohort piques a reader's curiosity and leads to more questions. A most intriguing question arising from this study concerns the difference between the subjects who developed knee OA during the study period and those who did not. If we take from this study the findings that prior physical activity, OA in the other knee, knee injury, and greater weight do not increase the risk for developing knee OA, then we wonder what does. What other variables might be making important contributions to manifestation and progression of knee OA?

Our inability to find the answer to this question within the study by Felson et al, as well as in most other studies of knee OA, is not due to poor design or implementation, but results from missing data. That is, data that we do not collect–variables we do not consider, and questions we do not ask. It is generally agreed that people with knee OA are a heterogeneous population. In fact, there is a great deal of unexplained variation in all aspects of knee OA, such as age of onset, occupational and medical history, rate and site of progression, remission and repair, and responses to a variety of pharmacologic and physical interventions. However, few studies consider differences in individual subjects. There is growing evidence that a number of individual characteristics are important variables in understanding knee OA progression and response to treatment (3–7). These variables should be identified and included in any study that proposes to answer questions related to knee OA development, progression, or response to treatment.

This community-based study by Felson et al is a step toward evidence-based recommendations regarding exercise and the risk for knee OA (1). Prior research performed with subjects with existing joint disease also indicated that moderate exercise did not exacerbate disease, and may have a number of beneficial effects (8). Therefore, regular moderate physical activity can be recommended for people with and without knee OA with a fair degree of certainty in its safety and effectiveness for improving function and general health.

The question remains whether physical activity can improve the joint environment and reduce OA progression. Future information to answer this question will come from both epidemiologic studies and clinical trials such as that by Roos and Dahlberg, who reported improved cartilage glycosaminoglycan content in people who participated in a randomized trial with an intervention of moderate-intensity neuromuscular and aerobic training (9). In their study, all subjects had undergone partial resection of the medial meniscus 3 to 5 years earlier. The authors characterized subjects as individuals with early joint disease. Of a total of 30 subjects, 22 reported monthly knee pain, 16 reported at least mild functional difficulty, 21 reported mild knee joint stiffness, 26 reported awareness of knee problems monthly, and 11 (37%) satisfied ACR clinical criteria for knee OA (10). In addition to significant between group differences, there was a strong relationship between increased physical activity and improved cartilage glycosaminoglycan content (r = 0.74, 95% confidence interval 0.52–0.87).

The cohort study of disease incidence, joint space width, and prior physical activity contains little information about individual subjects (1). The clinical trial of physical activity and cartilage content reported more individual information, including subjects' current physical activity level and knee strength (9). However, it would have been more informative if both studies had considered the contribution of potentially mediating factors and reported values for variables such as knee alignment, laxity, and neuromuscular fitness.

At this time, there is little agreement on how the research subject should be characterized, i.e., which individual and/or environmental factors should be measured or reported. Currently, there are published criteria to classify disease, grade severity, measure outcomes, and quantify response to treatment in knee OA. However, there has been little discussion about what individual factors might affect incidence or outcomes. Research recommendations developed at the 2002 International Conference in St. Louis, Missouri on Health Promotion and Disability Prevention for Individuals and Populations with Rheumatic Disease: The Evidence for Exercise and Physical Activity, point to the need for more precise characterization of subjects in terms of biomechanics, neuromuscular fitness, occupation, sports participation, and injury (11–13). Meaningful interpretation of research results and translation of findings to clinical settings for evidence-based practice require more complete characterization of study subjects. Research has begun to identify a number of relevant clinical variables that may aid in the interpretation of study results.

Future research, whether designed to evaluate the effectiveness of interventions or to identify risk factors for development or progression, should characterize subjects in terms of variables relevant to knee OA. These variables will likely include individual characteristics as well as social and environmental factors. The relevant variables need to be identified and a set of standardized measures consistently used to explore the variability in knee OA. Inquiry must now identify factors that make a difference to individual response. We must identify and agree upon meaningful characterization of research subjects and move beyond general statements of risk and efficacy in the aggregate. In addition to improving the usefulness of knee OA research, our ultimate aim must be to produce evidence that assists clinical decision-making and individualized recommendations regarding safety and effectiveness of interventions, including physical activity.


  1. Top of page