Table 4.
Models performance and additional predictive value of smoking habit, when added to the fully adjusted Cox model, on the study endpoints.
| CV fatal and non-fatal events | All-cause death | ESKD | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Model 1 | Model 2 | p | Model 1 | Model 2 | p | Model 1 | Model 2 | p | |
| LR test | – | 0.018 | – | 0.095 | – | 0.507 | |||
| c-index | 0.705 (0.669–0.741) | 0.716 (0.678–0.750) | 0.001 | 0.684 (0.644–0.722) | 0.692 (0.654–0.731) | 0.002 | 0.672 (0.631–0.714) | 0.673(0.630–0.714) | 0.943 |
| R2 | 66.0 (59.8–78.0) | 69.9 (64.1–82.1) | – | 72.3 (62.8–89.3) | 74.2 (65.6–90.0) | – | 63.2 (50.1–84.6) | 67.1 (52.3–87.1) | – |
| AIC | 2373 | 2330 | – | 2227 | 2204 | – | 2375 | 2334 | – |
| BIC | 2425 | 2393 | – | 2279 | 2266 | – | 2421 | 2386 | – |
LR test (Likelihood Ratio test). Comparisons are made between Model 1 (without smoking habit) and Model 2 (including smoking habit). Model 1 and 2 were adjusted for all the covariates included in Table 3.
AIC akaike information criterion, BIC Bayesian information criterion, R2 Royston’s modification of Nagelkerke’s R2. Estimates with p value <0.05 are shown in bold.