Particle-in-Cell Direct Simulation Monte Carlo (PIC-DSMC) is a widely used method for simulation of non-equilibrium plasmas, especially when the plasma flow is rarefied, and the applicability of fluid models is questionable. However, the PIC-DSMC method is subject to stochastic noise, and depending on the process being simulated, might require extremely large computational efforts. Therefore, the improvement of accuracy of PIC-DSMC methods is a topic of active research. In the present work, a recently developed collision and boundary condition treatment scheme, dubbed "event splitting", which aims at improving simulation of low-probability processes, is applied to ionization processes in xenon and helium plasmas, where it shown to reduce the level of stochastic noise compared to standard DSMC collision schemes.