Table 2. Simulation Results.

Estimated power to detect an association using a Cox-model based approach vs. a two-sample t-test comparing progressors and non-progressors. We assume rank-transformed GMD values, a sample size of n=123, and a two-sided alpha level of 0.05.

	Effect Size Expressed in terms of Hazard Ratioa
Analysis Performed	1.5	2.0	2.5	3.0
Cox model approach	0.39	0.80	0.97	1.0
Two-sample approach
1 y follow-up cutoff	0.15	0.30	0.53	0.62
2 y follow-up cutoff	0.24	0.54	0.80	0.88
3 y follow-up cutoff	0.29	0.63	0.88	0.94
4 y follow-up cutoff	0.27	0.58	0.84	0.91
5 y follow-up cutoff	0.18	0.39	0.62	0.70

^aHazard ratio based on comparing the 25th percentile to the 75th percentile