Background/Purpose: Physical activity may help reduce symptoms in adults with knee osteoarthritis (KOA), yet few meet the recommended levels. This analysis identified participant characteristics associated with improvement in physical activity over 28 weeks in a clinical trial setting aimed at testing an incentive program utilizing elements of gamification and social incentivization. We hypothesized that greater self-efficacy (confidence in their ability to exercise) would be a significant predictor of improvements in average daily step counts.

Methods: This was a multi-center randomized controlled trial with a factorial design. The study included 221 participants aged 40-85 years from 4 Veterans Affairs centers with a Kellgren-Lawrence (KL) knee radiograph grade ≥1 and clinical indication for a corticosteroid joint injection. Participants were randomized into either the incentive program or control and then further randomized to receive corticosteroid or lidocaine-only injections. A web-based platform collected daily step counts (from a Fitbit activity monitor) and patient-reported outcomes. Daily step counts were averaged over 2-week periods throughout the 28-week trial. Self-efficacy was measured at enrolment using the Exercise Self-Efficacy Scale. Linear regression incorporating generalized estimating equations was used to evaluate the association of participant characteristics with the change in average daily step counts. Variables associated with the outcome (p< 0.1) in univariate analyses were tested in multivariable models and removed in a backwards selection process to arrive at a final model.

Results: A total of 221 participants were randomized. Mean participant age was 64 years-old and 87% male. Unadjusted associations between participant characteristics and change in average daily steps are shown in Table 1. Individuals with greater self-efficacy for exercise demonstrated significantly greater increases in the average daily step count (Figure 1). This effect was greatest for participants in the highest tertile, whose daily step counts increased about 1145 steps more than those in the lowest tertile (Figure 1). This pattern was observed in both incentive and control arms and was an independent predictor in the multivariable model (Table 2). Patient characteristics associated with less improvement in average daily step count included higher pain scores, worse sleep quality, greater fatigue, kidney disease, and higher BMI. Participants that lived in areas in the highest quartile of Area Deprivation Index (ADI) also failed to improve.

Conclusion: We identified predictors of improvement in physical activity level among patients with KOA in the context of a clinical trial. Higher self-efficacy for exercise was a strong predictor of greater improvement in physical activity levels throughout follow-up. Less improvement was observed among those with greater pain, comorbidity, BMI, fatigue, and among those living in unsafe areas or areas with high ADI. These observations highlight the need to address low self-efficacy and other barriers in order successfully promote exercise in this population.

Table 1: Factors associated with change in average step counts in univariate models. **Also tested but not significant: smoking status, heart disease, degenerative disc disease, heart failure, depression, employment status.

Table 2: Factors associated with changes in average step counts in a multivariable model.

Figure 1. Average change in daily steps (with 95% CI) overall and by incentive group among Veterans with knee osteoarthritis, predicted from univariate linear regression models.

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/predictors-of-change-in-physical-activity-level-among-veterans-with-knee-osteoarthritis-in-a-behaviorally-designed-incentive-program/