Cohort-relative knee pain, walking performance, and risk of total knee arthroplasty: An Osteoarthritis Initiative analysis

Objectives: The objective of the study is to examine whether cohort-relative knee pain and 400-m walking performance are associated with subsequent total knee arthroplasty and whether preserved walking identifies a lower-hazard subgroup among higher-pain knees. Methods: This was a secondary analysis of the Osteoarthritis Initiative, comprising 4,247 participants contributing 8,494 knees. Knees were grouped by cohort-median splits of Western Ontario and McMaster Universities Arthritis Index pain (>2/20) and 400-m walk time (>297.8 s) into four phenotypes: Lower Pain/Faster Walk (reference), Lower Pain/Slower Walk, Higher Pain/Faster Walk, and Higher Pain/Slower Walk. A knee-specific marginal Cox model with cluster-robust variance estimated the cause-specific hazard of first arthroplasty, with death as a competing event, adjusted for age, sex, body mass index, Kellgren–Lawrence grade ≥2, depressive symptoms, and subcohort. Results: Over a median of 14.0 years, 658 first arthroplasties and 540 competing deaths occurred. Both higher-pain phenotypes showed elevated hazard versus reference (Higher Pain/Faster Walk 2.58 [2.03–3.28]; Higher Pain/Slower Walk 2.23 [1.76–2.84]; both p < 0.001); Lower Pain/Slower Walk did not (1.16 [0.91–1.48]; p = 0.24). Neither the direct higher-pain contrast (1.16 [0.93–1.44]; p = 0.19) nor the pain × walk interaction (0.75 [0.54–1.03]; p = 0.07) was significant. Nine-year cumulative incidence, with competing mortality, was 3.8%, 6.0%, 15.4%, and 17.6%. Adjustment for medial joint-space width attenuated but did not remove the associations. Conclusion: Knee pain above the cohort median was associated with a higher hazard of subsequent arthroplasty, whereas dichotomized 400-m walk performance did not clearly discriminate hazard among higher-pain knees. These cohort-relative findings, based on a single non-knee-specific walking measure, are hypothesis-generating and carry no implication for arthroplasty appropriateness.