Background/Purpose: Studies examining the relation of osteoarthritis (OA) to mortality or cardiovascular disease (CVD) have reported conflicting results, raising the question of whether OA itself increases such risks, or rather OA may simply be a marker of these outcomes due to its relation to other factors that heighten such risks, such as obesity. We assessed the relation of knee OA to mortality and incident CVD in a population-based cohort of older adults in which knee OA status, death and CVD were comprehensively assessed in a standardized, validated manner.

Methods: We included participants aged 50-75 from the Framingham OA Study comprising the Original and Offspring cohorts with knee x-rays obtained at exam 18 (1983-85) for the Original cohort, and exam 7 (2002-2005) for the Offspring cohort. Radiographic knee OA (ROA) was defined as Kellgren and Lawrence grade ≥2, and symptomatic knee OA (SxOA) as ROA plus pain in the same knee. All-cause mortality was defined as death due to any cause. Incident CVD was defined as a new diagnosis of coronary heart disease, intermittent claudication, CHF, stroke or TIA in the absence of any of these diseases prior to knee OA assessment.  All deaths and CVD events were adjudicated by a panel of cardiologists and neurologists using published criteria to review clinical data and hospitalization records. Follow-up started from the date of knee OA ascertainment and continued until the outcome (death/incident CVD) or last assessment in 2009. We truncated follow-up at 10 years to avoid universal occurrence of events in these older subjects. In sensitivity analyses, we evaluated outcomes beyond 10 years. For the incident CVD analysis, only those free of CVD at the time of knee OA ascertainment were eligible for inclusion.  The relation of knee OA to overall mortality and incident CVD were examined in both cohorts combined as well as separately, using Cox proportional hazards models, adjusting for age and sex in the first model, and then additionally adjusted in a second model for body mass index (BMI), diabetes mellitus, hypertension, renal disease, aspirin use, lipid-lowering medications and smoking status.

Results: Among 2037 participants (mean age 66 yrs, 58% women, mean BMI 27 kg/m², 499 ROA, 163 SxOA), there were 873 deaths and 624 incident CVD cases. No association was found between knee OA (ROA or SxOA) and overall mortality or incident CVD (Table), with effect estimates in the adjusted models ranging from 0.93-1.27. Results were similar when the cohorts were analysed separately and in the sensitivity analyses. While OA was associated with obesity and other risk factors for CVD, additionally adjusting for these factors did not affect the risk of the outcomes.

Conclusion: In this large cohort of community-dwelling older adults, there was no significant relation of knee OA to mortality or incident CVD, suggesting knee OA itself likely does not have systemic effects.