Table 3.
Multiple linear regression analyses of PWV (see “Methods”—“Statistical analysis” for details)
| Dependent variable: PWV | Main effects model | Model with effect modification |
|---|---|---|
| Beta (p value) | Beta (p value) | |
| Age, years | 0.06 (0.13) | 0.07(0.10) |
| MATSUDA | − 0.26 (< 0.001) | − 0.27 (< 0.001) |
| HDL cholesterol, mg/dl | 0.08 (0.04) | 0.07 (0.06)) |
| Hemoglobin, g/dl | 0.007 (0.85) | 0.02 (0.60) |
| AST, U/l | 0.04 (0.35) | 0.03 (0.41) |
| γGT, U/l | − 0.02 (0.62) | − 0.03 (0.47) |
| e-GFR, ml/min/1.73 m2 | − 0.11 (0.003) | − 0.10 (0.009) |
| Ferritin, ng/ml | 0.37 (< 0.001) | 0.19 (0.01) |
| hs-CRP, mg/l | 0.20 (< 0.001) | 0.05 (0.42) |
| Ferritin * hs-CRP | … | 0.28 (0.004) (see Fig. 2) |
Data are standardised regression coefficients (beta) and p values
By forcing gender into the main effect model does not modify the ferritin-PWV relationship (beta = 0.39, p < 0.001) and this was also true when forcing transferrin saturation (beta = 0.37, p < 0.001). In the same model, neither gender (beta = − 0.06, p = 0.16) nor transferrin saturation (beta = 0.07, p = 0.08) were related to PWV
PWV pulse wave velocity, HDL high density lipoproteins, hs-CRP high sensitivity C reactive protein, AST aspartate aminotransferase, γGT γ-glutamyltransferase, e-GFR estimated glomerular filtration rate