Knee confidence as it relates to physical function outcome in persons with or at high risk of knee osteoarthritis in the osteoarthritis initiative

Objective: To evaluate whether low knee confidence at baseline is associated with poor baseline-to-3-year physical function outcome in the Osteoarthritis Initiative.

Methods: Knee confidence was assessed using an item from the Knee Injury and Osteoarthritis Outcome Score instrument. Physical function was assessed using self-report measures (Western Ontario and McMaster Universities Osteoarthritis Index [WOMAC] function score and Short Form 12 physical component scale) and performance-based measures (20-meter walk and chair stand test). Poor function outcome was defined as moving into a worse function group or remaining in the 2 worst function groups between baseline and 3 years. Logistic regression was used to evaluate the relationship between baseline knee confidence level and poor baseline-to-3-year function outcome, adjusting for potential confounders.

Results: The sample included 3,975 men and women with or at high risk of developing osteoarthritis of the knee, of whom 37-53% had poor baseline-to-3-year function outcome. For both self-report measures, increasingly worse knee confidence was associated with a greater risk of poor function outcome, and trend tests supported a graded response (e.g., the adjusted odds ratios [95% confidence intervals] for the WOMAC function score for worsening confidence categories were 1.26 [1.07-1.49], 1.43 [1.16-1.77], and 2.05 [1.49-2.82], P for trend <0.0001). Similar associations between confidence and performance-based function outcome were observed, but statistical significance did not persist in adjusted analyses. Factors independently associated with poor function outcome for all 4 outcome measures were depressive symptoms, comorbidity, body mass index, and joint space narrowing.

Conclusion: These findings indicate that worse knee confidence at baseline is independently associated with greater risk of poor function outcome by self-report measures, with evidence of a graded response; the relationship with performance measures is not significant in fully adjusted models.