Study: Skinny-Leg Runners Face Higher Rates of Knee Arthritis

Researchers have long wondered whether activities that put weight stress on the knees, such as walking, running, or ball games, raise risks for osteoarthritis (OA) of the knee. Now a lengthy prospective study provides an answer.

Among some 5,000 participants in the Rotterdam Study, individuals who reported regular participation in weight-bearing activities were 22% more likely to develop knee OA per each 1-unit increase in metabolic impact (OR 1.22, 95% CI 1.10-1.35), according to Joyce B.J. van Meurs, PhD, of Erasmus Medical Center in Rotterdam, the Netherlands, and colleagues.

But this increased risk was actually concentrated in just one subgroup: those in the lowest third of lower-limb muscle mass, the researchers reported in JAMA Network Open. In this group, people who engaged in regular weight-bearing activities had more than 50% increased risk for knee OA per unit of metabolic impact (OR 1.53, 95% CI 1.15-2.04), while people in the middle and upper tertiles showed no relationship between their level of exercise and knee OA risk (OR 0.93 and 1.15, respectively, P>0.05).

And in all cases, increasing levels of non-weight-bearing exercise such as swimming or cycling had no association with knee OA risk.

"Although physical activity is known to have numerous health benefits, our study suggests that caution is needed when engaging in weight-bearing activity, especially for individuals with low levels of lower-limb muscle mass," van Meurs and colleagues concluded.

It stands to reason that people whose daily life includes regular activities that impact the knees would, over time, incur more damage to joint structures. But past studies looking to demonstrate it failed to turn up a clear association. The current authors cited a recent, large meta-analysis that "found no association of total physical activity with increased risk for knee osteoarthritis." It still remained uncertain, though, whether increasing degrees of more directly impactful activities could raise the risk.

Thus, van Meurs's group turned to the Rotterdam Study, a long-running prospective, observational project that enrolled people in the Dutch seaport city in three sequential cohorts beginning in 1990. Cohort 2 began enrolling in 2000, and cohort 3 in 2006. To be included, participants needed to have provided extensive baseline data on their physical activities (recreational and otherwise) and other lifestyle factors. DXA scans were used to measure lower-limb muscle mass, with participants divided into tertiles according to the resulting scores. Follow-up time averaged 6.33 years, during which 8.4% of the 5,003 included participants developed new-onset knee OA.

For the statistical analysis, the researchers quantified participants' average output as metabolic equivalent of task (MET)-hours/week, separately for weight- and non-weight-bearing activities, and separately for each study cohort, and then standardized as z-scores to allow comparisons between cohorts. Averages were as follows:

• Weight-bearing: cohort 1, 35.41 (SD 31.19); cohort 2, 29.55 (SD 22.01); cohort 3, 15.69 (SD 31.13)
• Non-weight-bearing: cohort 1, 11.38 (SD 14.92); cohort 2, 8.05 (SD 13.57); cohort 3, 6.83 (SD 15.64)

Those averages served as the incremental units of metabolic impact for calculating the odds ratios. Thus, lead author Yahong Wu, MD, also of Erasmus Medical Center, explained in an email, "for weight-bearing physical activity, 1 unit represents an average of 15.69-35.41 MET-hour/week."

A wide range of activities qualified as weight-bearing: in addition to walking and running, they included gardening, golf, sailing, ice skating and other winter sports, tennis and other ball sports, bowling, dancing, and "moving to music."

The primary analysis included adjustments for sex, body mass index, follow-up time, and Kellgren-Lawrence grade, a measure of radiographic knee damage. Results were similar when further adjustments were taken for education level, alcohol use, smoking, blood pressure, lipid levels, and diabetes.

These statistics excluded 1,511 participants reporting knee pain at baseline; their data were examined separately. Numerical trends in this group were similar to those seen for participants without knee pain, but the associations fell short of statistical significance. The authors also pointed out that, as one might expect, individuals with knee pain were less active on average than those without.

"Given the intricate nature of these relationships and the inherent limitations of our observational study," van Meurs and colleagues wrote, "future studies with larger sample sizes and different methods, such as randomized clinical trials, are needed to provide more definitive insights."

Comment