Table 3.
Multivariable models for GlycA (log) in persons with rheumatoid arthritis and controls
Parameter estimate | Partial R 2 | P value | |
---|---|---|---|
Rheumatoid arthritis: model R 2 = 0.74, n = 48 | |||
hsCRP (log mg/l) | 58.4 | 0.59 | <0.0001 |
ESR (mm/hour) | 1.8 | 0.11 | 0.002 |
Thigh muscle density (Hu) | −2.7 | 0.05 | 0.01 |
Controls without rheumatoid arthritis: model R 2 = 0.47, n = 33 | |||
Thigh subcutaneous adiposity (cm2) | 0.54 | 0.35 | 0.0003 |
IL-18 (log pg/ml) | 132.6 | 0.12 | 0.01 |
Multivariable modeling was performed using linear models with forward stepwise variable selection. Variables for forward selection were based on significant results from bivariate analyses shown in Table 2. For RA patients, variables were selected from disease activity (DAS-28), ESR, hsCRP (log), IL-6 (log), total abdominal adiposity, thigh muscle density, fasting glucose, age, and gender. For controls, variables were selected from hsCRP (log), IL-18 (log), total abdominal adiposity, thigh subcutaneous adiposity, acute insulin response, insulin sensitivity, age, and gender. Final models are shown
DAS-28 Disease Activity Score, ESR erythrocyte sedimentation rate, hsCRP high-sensitivity C-reactive protein, IL interleukin