Email zadetek: On the choice of time scales in competing risks predictions