marginal significanceP = 0.05
Because results using the summary score model with the validation cohort showed evidence of lack of fit at marginal significance ( P = 0.05), and because the AUC (0.65) was worse than it was using the standard model, we did not pursue the summary score with the merged cohort.