Background/Purpose: Body mass index (BMI), healthy diet or components, physical activity, and tobacco smoking have been examined independently with risk of knee osteoarthritis (OA). Sleep quality has been shown to be related to chronic pain but its relation to knee OA is unclear. In this study, we evaluated the combined effects of lifestyle factors with or without BMI on risk of knee OA.

Methods: We used data from the Osteoarthritis Initiative (OAI), a prospective, multicenter study of 4,796 U.S. men (41.5%) and women [mean (SD) age: 61.2 (9.2) y and BMI: 28.6 (4.8) kg/m²] with or at risk of knee OA. Lifestyle factors assessed at baseline of the study including overall diet (assessed by Dietary Approaches to Stop Hypertension, DASH), physical activity (Physical Activity Scale for the Elderly, PASE), tobacco smoking (never, former and current smokers), and sleep quality (frequency of restless sleep in the Center for Epidemiologic Studies Depression Scale, CESD), and BMI (kg/m²) were examined separately for their relationship with risk of incident symptomatic (SxOA) and radiographic (ROA) knee OA. Because physical activity did not show a consistent relationship with knee OA, each participant was scored on all factors except PASE from least healthy (0 point) to healthiest (1 point) for each factor with a total score of 0 (least healthy) to 3 (healthiest) excluding BMI in the composite or a total score of 0 (least healthy) to 4 (healthiest) including BMI (see Table 1). Incident ROA was defined as a new onset of Kellgren and Lawrence grade ≥2 and incident SxOA was defined as new cases of both ROA and a painful knee on most days in past month. We applied Generalized Estimating Equations to account for correlations of both knees in each person in multivariable regression models to assess the association between the lifestyle composite and risk of knee OA using 0 (least healthy) as the referent group.

Results: Up to 96 months in OAI, we identified 1340 knees with incident SxOA among 7662 knees and 625 knees with incident ROA among 5861 knees after excluding knees with missing values of OA status or knees with prevalent OA. When BMI was excluded in the composite, a higher score was associated with a lower risk of SxOA (p-trend<0.0001), but no association was found with ROA. When BMI was included in the composite, there was a significant inverse relationship with risks of SxOA and ROA (p-trend<0.0001) (see Table 2).

Conclusion: Data from the OAI suggested that persons who lived a healthy lifestyle independent of BMI and physical activity had a lower risk of symptomatic knee OA, but the combined impacts of diet, sleep and tobacco smoking was not related to risk of radiographic knee OA.