Abstract / Summary
Abstract Physical activity is important for health, but the association between the temporal distribution of weekly moderate-to-vigorous physical activity (MVPA) and incident osteoarthritis (OA) remains uncertain. This observational cohort analysis included 70,608 UK Biobank participants. Physical activity exposure was based on UK Biobank processed accelerometer variables derived from wrist-worn Axivity AX3 devices. Participants were classified as inactive (< 150 min/week MVPA), regularly active ( > = 150 min/week with < 50% accumulated during the two most active days), or weekend warriors (WW; >=150 min/week with > = 50% accumulated during the two most active days). The two most active days were not required to be Saturday and Sunday. Cox proportional hazards models estimated associations with incident overall, knee, hip, and hand OA after covariate adjustment; associations were also examined across OA polygenic risk strata. Compared with inactivity, both the WW and regularly active patterns were associated with lower overall OA risk. A direct comparison of the WW and regularly active groups showed no statistically significant difference in overall OA hazard. Site-specific associations differed: the two patterns showed similar associations with knee OA, while the WW group had a 21% higher risk of hand OA than the regularly active group. For hip OA, an inverse association was observed for the WW group, whereas the estimate for the regularly active group was not statistically significant. Associations between activity patterns and OA were observed across polygenic risk strata. The WW pattern defined by the two most active days was associated with lower overall OA risk compared with inactivity, and the direct comparison with regularly distributed activity was not statistically significant. Site-specific findings should be interpreted cautiously. Because this was an observational analysis of processed wrist-accelerometer variables, the results do not establish causality or identify joint-level mechanisms.