Table 2.
Results of multivariable logistic regression models.
| Variables | Basic Model | Best AIC Model | ||
|---|---|---|---|---|
| Coefficients, β | p Value | Coefficients, β | p Value | |
| (Intercept) | −182.013 | <0.001 | −161.37 | <0.001 |
| Ln (Age) | 45.716 | <0.001 | 40.465 | <0.001 |
| Male | 8.516 | 0.011 | 8.252 | 0.013 |
| HTN | 6.302 | 0.011 | 6.622 | 0.007 |
| DM | 3.917 | 0.263 | 0.635 | <0.001 |
| Ln (eGFR) | 3.259 | 0.708 | −0.81 | 0.035 |
| Ln (BMI) | 23.992 | 0.008 | 24.307 | 0.007 |
| Ln (non-HDL) | 12.269 | 0.003 | 11.723 | 0.004 |
| Interaction | ||||
| Sex * Ln (age) | −1.898 | 0.021 | −1.833 | 0.025 |
| HTN * Ln (age) | −1.419 | 0.023 | −1.498 | 0.015 |
| DM * Ln (age) | −0.81 | 0.347 | ||
| Ln (eGFR) * Ln (age) | −1.035 | 0.638 | ||
| Ln (BMI) * Ln(age) | −5.94 | 0.009 | −6.018 | 0.008 |
| Ln (non-HDL) * Ln (age) | −3.014 | 0.003 | −2.876 | 0.005 |
Note: Akaiki information criterion (AIC): basic model, 2638.4; Best AIC model, 2635.4. Multivariable logistic regression models with full variables and best AIC. Continuous variables were log-transformed. To prevent overfitting, 10-folds cross-validation with 10 times of iteration was done. Among these models, the model with the best AIC was selected. Abbreviations: HTN, hypertension; DM, diabetes mellitus; eGFR, estimated glomerular filtration rate; non-HDL, non-high density lipoprotein cholesterol.