Table 2. Prediction of DGF using KeGFR versus the clinical model.
| Variables | AUC (95% CI) | p | IDI-DGF (95% CI) | IDI-Non-DGF (95% CI) | |
|---|---|---|---|---|---|
| Base Model 4h (n = 56) | 0.72 (0.58 to 0.86) | a | a | ||
| Base Model + | sCr | 0.73 (0.59 to 0.87) | 0.79 | 0.01 (-7 x 10–4 to 0.07) | 0.01 (-4 x 10–4 to 0.05) |
| Base Model + | KeGFR sCr | 0.77 (0.63 to 0.89) | 0.24 | 0.03 (1 x 10–6 to 0.11) b | 0.02 (3 x 10–5 to 0.08) b |
| Base Model + | pCysC | 0.77 (0.64 to 0.90) | 0.13 | 0.02 (-5 x 10–5 to 0.11) | 0.02 (-1 x 10–5 to 0.08) |
| Base Model 8h & 12h (n = 52) | 0.68 (0.52 to 0.83) | ||||
| 8h (n = 52) | |||||
| Base Model + | sCr | 0.67 (0.52 to 0.83) | 0.39 | 0.00 (-0.004 to 0.004) | 0.00 (-0.001 to 0.002) |
| Base Model + | KeGFR sCr | 0.81 (0.67 to 0.94) | 0.16 | 0.07 (9 x 10–4 to 0.21) b | 0.04 (6 x 10–4 to 0.12) b |
| Base Model + | pCysC | 0.78 (0.65 to 0.92) | 0.10 | 0.03 (-5 x 10–4 to 0.14) | 0.02 (-2 x 10–4 to 0.09) |
| Base Model + | KeGFR pCysC | 0.78 (0.64 to 0.92) | 0.19 | 0.06 (2 x 10–4 to 0.20) b | 0.04 (0.15 to 0.12) b |
| 12h (n = 52) | |||||
| Base Model + | sCr | 0.71 (0.56 to 0.86) | 0.32 | 0.01 (-0.004 to 0.06) | 0.00 (-0.001 to 0.03) |
| Base Model + | KeGFR sCr | 0.88 (0.78 to 0.99) | 0.01 b | 0.18 (0.04 to 0.35) b | 0.10 (0.03 to 0.21) b |
| Base Model + | pCysC | 0.80 (0.67 to 0.93) | 0.11 | 0.06 (-8 x 10–4 to 0.19) | 0.03 (-2 x 10–4 to 0.11) |
| Base Model + | KeGFR pCysC | 0.82 (0.69 to 0.95) | 0.09 | 0.11 (0.02 to 0.25) b | 0.06 (0.01 to 0.15) b |
Model enhancement was analysed by calculation of the IDI. The clinical base model was derived from recipient-, donor- and transplant related factors (reference [8]). There is no KeGFRpCysC at 0–4h since no 0h pCysC data were available.
Key:
a: metrics are not calculable for baseline model alone
b: p < 0.05 vs. base model; IDI: integrated discrimination improvement.