Skip to main content

Table 1 Time-dependent AUC estimated via nearest-neighbor smoothing [15, 16] at one year post-waitlist registration and one year post-transplant for the modified and existing LAS models. 95% confidence intervals were computed via the bootstrap percentile method; p-values were estimated using the normal approximation and standard deviation obtained from the bootstrap replicates. 1000 bootstrap replicates were used

From: Mitigating selection bias in organ allocation models

Cohorta

Modified LAS

Existing LAS

Difference

p-value of difference

Development

Pre-tx

0.732 (0.690, 0.760)

0.660 (0.619, 0.697)

0.071 (0.030, 0.106)

< 0.001

Post-tx

0.605 (0.580, 0.629)

0.560 (0.531, 0.585)

0.045 (0.026, 0.065)

< 0.001

Testing

Pre-tx

0.750 (0.686, 0.792)

0.693 (0.631, 0.738)

0.057 (−0.004, 0.122)

0.083

Post-tx

0.570 (0.536, 0.606)

0.540 (0.507, 0.576)

0.030 (0.003, 0.058)

0.030

  1. atx: transplant