<p>Life-threatening arrhythmias like ventricular fibrillation are the leading cause of sudden cardiac death worldwide. The standard clinical treatment involves delivering a high-energy defibrillation shock that interacts with the heterogeneous cardiac tissue. The significant induced electrical currents lead to severe side-effects like tissue damage or post-traumatic stress. The introduction of biphasic waveforms, instead of monophasic ones enabled a significant reduction in energy leading to mitigated side-effects. While many hypotheses for the increased efficiency of biphasic waveforms exist, the underlying mechanisms are not entirely understood. In a statistically driven multi-scale study, we use GPU-based simulations to investigate the influence of different cardiovascular structures on the success rate of defibrillation. The main result of our study is that biphasic waveforms do not excite a significantly larger fraction of the tissue than monophasic waveforms. As the leading mechanism for the increased efficiency, we identify the fact that biphasic shocks lead to significantly more uniform microscopic excitation patterns of the tissue and therefore a decreased reinitiation rate of fibrillation. Revealing the detailed interaction between electric fields and cardiac tissue enables the development and optimization of low-energy defibrillation strategies and waveforms, opening the way towards mitigating side-effects and improving life quality of patients.</p>