Table 2.
Baseline model for risk of renal artery stenosis based on 426 patients and DLR model based on a nested case–control substudy of serum creatinine. Shown are log odds ratios for P(D = 1|X) and coefficients for covariates in the corresponding model for logDLRX(Y). Continuous variables are standardized to mean 0 and variance 1. p-values in parentheses are calculated by comparing estimates with standard errors. Standard errors for coefficients in the logDLR model are calculated by adopting the techniques described in the appendix
| Factor | Baseline risk | log DLRX(Y) | log DLRX(Y)∗ |
| Intercept | – 2.54 (< .001) | 0.06 (0.56) | 0.07 (0.44) |
| Gender (female) | 0.38 (0.18) | 0.44 (0.01) | 0.47 (0.01) |
| Age (per 10 years) | 0.61 (< 0.001) | – 0.18 (0.002) | – 0.18 (0.003) |
| Hypertension | 0.66 (0.03) | 0.03 (0.81) | 0.00 |
| BMI (kg/m3) | – 0.20 (< 0.001) | 0.03 (0.06) | 0.03 (0.09) |
| Abdominal bruit | 1.41 (< 0.001) | 0.33 (0.15) | 0.00 |
| Atherosclerosis disease | 0.91 (0.002) | – 0.45 (0.01) | – 0.42 (0.01) |
| log serum creatine | — | 0.92 (< 0.001) | 0.91 (< 0.001) |
*
After setting to 0 coefficients for abdominal bruit and hypertension.