**2. Patients and methods**

### **2.1 Study population**

Study patients enrolled in the ongoing Veterans Affairs Rheumatoid Arthritis (VARA) Registry prior to January 2009 were included in the study. The characteristics of this population have been previously described (Mikuls et al., 2007). Briefly, VARA is a multicenter chronic disease registry initiated in 2003 now including collection sites at VA Medical Centers across the U.S. The current study included participants from VA sites in Dallas, Denver, Jackson, Omaha, Salt Lake City, and Washington, DC. Rheumatoid arthritis (RA) patients with disease onset after age 18 years who fulfill American College of Rheumatology (ACR) classification criteria for RA (Arnett et al., 1988) are invited to enroll. Participating sites prospectively collect and archive clinical and laboratory observations associated with RA during routine visits and standard of care. VARA has been approved by the Institutional Review Board (IRB) and VA Research and Development Committee at each participating site. All subjects provide written consent for a single blood draw and ongoing review of their electronic medical records.

#### **2.2 Metabolic syndrome definition**

The NCEP/ATPIII criteria (National Cholesterol Education Program, 2001) for MetS were modified so that a body mass index (BMI) of ≥ 30 kg/m2 replaced waist/hip circumference. Because waist and hip measurements are not routinely available on study subjects, BMI was used as a surrogate for waist-to-hip circumference with the following rationale: a) weight and annual height measurements are routinely obtained on all our patients, b) data support BMI to be as accurate as waist circumference in identifying individuals for risk of CVD (Farin et al, 2006), c) BMI is included in the WHO definition for MetS (World Health Organization [WHO], 1999) and there appears to be at least modest levels of agreement between BMI and waist-tohip ratio in both men and women (Balkau et al., 2006), and d) to obviate the varied cut off points for waist circumference for different ethnic groups (Alberti et al., 2005). Additionally, documented prior use of disease specific medications were used as surrogates for the other three MetS components. Hence MetS was defined as presence of ≥ 3 of 4 criteria (BMI ≥ 30 kg/m2, anti-hypertensive, lipid lowering, and/or diabetes agents) (Grundy et al., 2005).

#### **2.3 Clinical measures**

RA disease severity was assessed by rheumatoid factor (RF) positivity, duration of disease, presence of radiographic changes suggestive of RA (based on ACR criteria) and nodules at enrollment. Disease activity was assessed at the time of the most recent clinic visit by erythrocyte sedimentation rate (ESR), C-reactive protein (CRP), total swollen and tender joint count (0-28 joints), patient global well-being 100 mm visual analog scale, modified health assessment questionnaire (MD-HAQ, 0-3) and three-variable Disease Activity Score (DAS28-3v; a composite measure based on swollen/tender joint counts and ESR) (Bawa et al., 2005; Prevoo et al., 1995). The three-variable DAS was used in place of a four variable DAS to minimize the impact of missing data from patient global scores.

Disease-modifying anti-rheumatic therapy (DMARD), biologic treatments, and prednisone use were also recorded at the most recent visit. DMARD agents included methotrexate, sulfasalazine, hydroxychloroquine, and leflunomide; biologic treatments included antitumor necrosis factor (adalimumab, etanercept, infliximab), anti-signal 2 inhibitor (abatacept) and anti-CD20 (rituximab) agents. Patient report of ever or never smoking was tabulated, and subjects were categorized based on a history of CVD (documentation of prior coronary artery disease, myocardial infarction, angina, cerebrovascular event, transient ischemic attack, or peripheral vascular disease).
