Table 3.
Discrimination ability of CALIBRA. Analysis was performed both in the FMC NephroCare cohort and in the GCKD cohort; endpoint horizon: 1 year.
| Cohort | Contrast | Effective Sample Size | AUC CALIBRA | AUC benchmark | Delta AUCs | P-value |
|---|---|---|---|---|---|---|
| FMC Nephrocare | ||||||
| CALIBRA | 14,880 | 0.79 | - | - | - | |
| ASCVD - CALIBRA | 3,960 | 0.76 | 0.59 | -0.17 | <0.001 | |
| FHS - CALIBRA | 3,960 | 0.76 | 0.54 | -0.22 | <0.001 | |
| INDANA - CALIBRA | 3,960 | 0.77 | 0.63 | -0.14 | <0.001 | |
| G-CKD | ||||||
| CALIBRA | 4,822 | 0.73 | - | - | - | |
| ASCVD - CALIBRA | 4,792 | 0.73 | 0.61 | -0.12 | <0.001 | |
| FHS - CALIBRA | 4,792 | 0.73 | 0.57 | -0.16 | <0.001 | |
| INDANA - CALIBRA | 4,792 | 0.73 | 0.69 | -0.04 | <0.001 | |
CALIBRA performance was assessed either including all cases or excluding patients with missing information. The benchmark model scores were computed considering only complete cases. The column “Effective sample size” reports the number of patients included in each analysis. Imputation method: Listwise. Non-inferiority was defined as ΔAUC<0.05, while superiority was set at ΔAUC >=0.05.