Background/Purpose:  Knee osteoarthritis (OA), foot pain, and falls are common in older adults and limit mobility. We previously showed that knee OA increases the risk of indoor falls for men and outdoor falls for women. Because foot pain often accompanies knee OA and women report more foot pain than men, understanding whether the increased risk of falls in subjects with knee OA is mediated through foot pain can help us better target interventions to reduce falls. We examined the associations of knee OA with foot pain and falls, and assessed the extent that foot pain may mediate the association between knee OA and the risk of falls, and whether this varies by sex.  

Methods:   This study included 764 participants from the MOBILIZE Boston Study, a population-based cohort of older adults. Knee OA was assessed at baseline using the ACR clinical criteria. Falls data were prospectively collected using monthly calendars, with phone follow-up to assess location of falls. The presence of foot pain was assessed at baseline. Using negative binomial regression, we examined the sex-specific association of knee OA with the risk of indoor falls and outdoor falls adjusting for confounders. We applied a counterfactual approach of mediation analysis using logistic regression and marginal structural modeling to estimate the direct (through mechanisms excluding foot pain) and indirect (through the foot pain mechanism) effects to determine the extent that foot pain mediates the associations between knee OA and falls. 

Results:   Among study participants (486 women and 278 men, mean age: 78 years, mean BMI: 27.3), 25% had clinical knee OA. The proportions with foot pain were 35.8% in men with and 14.4% in men without knee OA; foot pain was reported by 33.6% and 24.5% of women with and without OA. Over an average of 2.2 years, 43% in men with and 35% in men without foot pain had ≥ 1 indoor falls while 39% and 36% of the women with and without foot pain had ≥ 1 outdoor falls. The adjusted rate ratio (RR) and 95% confidence interval (CI) for the total effect of knee OA (i.e., including both direct and indirect effects) on risk of indoor falls in men was 1.58 (95% CI: 0.99, 2.52), and on risk of outdoor falls in women was 1.70 (95% CI: 1.21, 2.40). The effect of knee OA on risk of indoor falls in men mediated by foot pain was 0.93 (95% CI: 0.73, 1.32) and on risk of outdoor falls in women was 1.00 (95%CI: 0.81, 1.25).   Conclusion:   Despite the increased risk of falls in people with OA and a strong association of knee OA and foot pain, there is no evidence that knee OA increases the risk of falls through foot pain in older men and women. The counterfactual method of mediation analysis reduces potential confounding and selection bias that occur with the conventional method of including a potential mediator in a model as a covariate. Future studies should explore other possible mechanisms through which knee OA affects the risk of indoor and outdoor falls differently for men and women.