Survival outcome with competing risks arises naturally in many clinical settings. For example, kidney transplant patients may die before experiencing adverse kidney and/or transplant-related outcomes (e.g. graft failure) or may experience adverse kidney and/or transplant-related outcomes. To identify and evaluate a candidate biomarker for its ability to accurately predict which patient would experience disease-specific adverse outcome and when, is an active research area. In particular, time-dependent prediction accuracy measures such as time-dependent ROC and AUC have been proposed for such competing risks settings. In our research, we adapt an existing non-parametric time-dependent AUC estimator for competing risks outcome. We show that the resulting estimator is intuitively meaningful and easy to evaluate even for a large dataset. We assess the properties of the estimator using extensive simulation studies and a real life kidney transplant data.